2014-12-12 Marc Glisse <marc.glisse@inria.fr>

[official-gcc.git] / gcc / ada / gnat_ugn.texi

\input texinfo   @c -*-texinfo-*-
@c %**start of header

@c oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
@c                                                                            o
@c                            GNAT DOCUMENTATION                              o
@c                                                                            o
@c                             G N A T _ U G N                                o
@c                                                                            o
@c           Copyright (C) 1992-2014, Free Software Foundation, Inc.          o
@c                                                                            o
@c oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo

@setfilename gnat_ugn.info

@copying
Copyright @copyright{} 1995-2014 Free Software Foundation,
Inc.

Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with no
Invariant Sections, with no Front-Cover Texts and with no Back-Cover
Texts.  A copy of the license is included in the section entitled
``GNU Free Documentation License''.
@end copying

@c oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
@c
@c                           GNAT_UGN Style Guide
@c
@c  1. Always put a @noindent on the line before the first paragraph
@c     after any of these commands:
@c
@c          @chapter
@c          @section
@c          @subsection
@c          @subsubsection
@c          @subsubsubsection
@c
@c          @end smallexample
@c          @end itemize
@c          @end enumerate
@c
@c  2. DO NOT use @example. Use @smallexample instead.
@c     a) DO NOT use highlighting commands (@b{}, @i{}) inside an @smallexample
@c        context.  These can interfere with the readability of the texi
@c        source file.  Instead, use one of the following annotated
@c        @smallexample commands, and preprocess the texi file with the
@c        ada2texi tool (which generates appropriate highlighting):
@c        @smallexample @c ada
@c        @smallexample @c adanocomment
@c        @smallexample @c projectfile
@c     b) The "@c ada" markup will result in boldface for reserved words
@c        and italics for comments
@c     c) The "@c adanocomment" markup will result only in boldface for
@c        reserved words (comments are left alone)
@c     d) The "@c projectfile" markup is like "@c ada" except that the set
@c        of reserved words include the new reserved words for project files
@c
@c  3. Each @chapter, @section, @subsection, @subsubsection, etc.
@c     command must be preceded by two empty lines
@c
@c  4. The @item command should be on a line of its own if it is in an
@c     @itemize or @enumerate command.
@c
@c  5. When talking about ALI files use "ALI" (all uppercase), not "Ali"
@c     or "ali".
@c
@c  6. DO NOT put trailing spaces at the end of a line.  Such spaces will
@c     cause the document build to fail.
@c
@c  7. DO NOT use @cartouche for examples that are longer than around 10 lines.
@c     This command inhibits page breaks, so long examples in a @cartouche can
@c     lead to large, ugly patches of empty space on a page.
@c
@c oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo

@set NOW January 2007
@c This flag is used where the text refers to conditions that exist when the
@c text was entered into the document but which may change over time.
@c Update the setting for the flag, and (if necessary) the text surrounding,
@c the references to the flag, on future doc revisions:
@c search for @value{NOW}.

@set FSFEDITION
@set EDITION GNAT

@set PLATFORM

@c @ovar(ARG)
@c ----------
@c The ARG is an optional argument.  To be used for macro arguments in
@c their documentation (@defmac).
@macro ovar{varname}
@r{[}@var{\varname\}@r{]}@c
@end macro
@c Status as of November 2009:
@c Unfortunately texi2pdf and texi2html treat the trailing "@c"
@c differently, and faulty output is produced by one or the other
@c depending on whether the "@c" is present or absent.
@c As a result, the @ovar macro is not used, and all invocations
@c of the @ovar macro have been expanded inline.


@settitle @value{EDITION} User's Guide
@dircategory GNU Ada tools
@direntry
* @value{EDITION} User's Guide: (gnat_ugn). @value{PLATFORM}
@end direntry

@include gcc-common.texi

@setchapternewpage odd
@syncodeindex fn cp
@c %**end of header

@titlepage
@title @value{EDITION} User's Guide

@sp 2

@subtitle GNAT, The GNU Ada Development Environment
@versionsubtitle
@author AdaCore

@page
@vskip 0pt plus 1filll

@insertcopying

@end titlepage

@ifnottex
@node Top, About This Guide, (dir), (dir)
@top @value{EDITION} User's Guide

@noindent
@value{EDITION} User's Guide @value{PLATFORM}

@noindent
GNAT, The GNU Ada Development Environment@*
GCC version @value{version-GCC}@*

@noindent
AdaCore@*

@menu
* About This Guide::
* Getting Started with GNAT::
* The GNAT Compilation Model::
* Compiling with gcc::
* Binding with gnatbind::
* Linking with gnatlink::
* The GNAT Make Program gnatmake::
* Improving Performance::
* Renaming Files with gnatchop::
* Configuration Pragmas::
* Handling Arbitrary File Naming Conventions with gnatname::
* GNAT Project Manager::
* Tools Supporting Project Files::
* The Cross-Referencing Tools gnatxref and gnatfind::
@ifclear FSFEDITION
* The GNAT Pretty-Printer gnatpp::
* The Ada-to-XML converter gnat2xml::
* The GNAT Metrics Tool gnatmetric::
@end ifclear
* File Name Krunching with gnatkr::
* Preprocessing with gnatprep::
* The GNAT Library Browser gnatls::
* Cleaning Up with gnatclean::
* GNAT and Libraries::
* Using the GNU make Utility::
* Memory Management Issues::
* Stack Related Facilities::
@ifclear FSFEDITION
* Verifying Properties with gnatcheck::
* Creating Sample Bodies with gnatstub::
* Creating Unit Tests with gnattest::
@end ifclear
* Performing Dimensionality Analysis in GNAT::
* Generating Ada Bindings for C and C++ headers::
* Other Utility Programs::
* Code Coverage and Profiling::
* Running and Debugging Ada Programs::
* Platform-Specific Information for the Run-Time Libraries::
* Example of Binder Output File::
* Elaboration Order Handling in GNAT::
* Overflow Check Handling in GNAT::
* Conditional Compilation::
* Inline Assembler::
* Writing Portable Fixed-Point Declarations::
* Compatibility and Porting Guide::
* Microsoft Windows Topics::
* Mac OS Topics::
* GNU Free Documentation License::
* Index::
@end menu
@end ifnottex

@node About This Guide
@unnumbered About This Guide

@noindent
This guide describes the use of @value{EDITION},
a compiler and software development
toolset for the full Ada programming language.
It documents the features of the compiler and tools, and explains
how to use them to build Ada applications.

@value{EDITION} implements Ada 95, Ada 2005 and Ada 2012, and it may also be
invoked in Ada 83 compatibility mode.
By default, @value{EDITION} assumes Ada 2012, but you can override with a
compiler switch (@pxref{Compiling Different Versions of Ada})
to explicitly specify the language version.
Throughout this manual, references to ``Ada'' without a year suffix
apply to all Ada 95/2005/2012 versions of the language.

@ifclear FSFEDITION
For ease of exposition, ``@value{EDITION}'' will be referred to simply as
``GNAT'' in the remainder of this document.
@end ifclear


@menu
* What This Guide Contains::
* What You Should Know before Reading This Guide::
* Related Information::
* Conventions::
@end menu

@node What This Guide Contains
@unnumberedsec What This Guide Contains

@noindent
This guide contains the following chapters:
@itemize @bullet

@item
@ref{Getting Started with GNAT}, describes how to get started compiling
and running Ada programs with the GNAT Ada programming environment.
@item
@ref{The GNAT Compilation Model}, describes the compilation model used
by GNAT.

@item
@ref{Compiling with gcc}, describes how to compile
Ada programs with @command{gcc}, the Ada compiler.

@item
@ref{Binding with gnatbind}, describes how to
perform binding of Ada programs with @code{gnatbind}, the GNAT binding
utility.

@item
@ref{Linking with gnatlink},
describes @command{gnatlink}, a
program that provides for linking using the GNAT run-time library to
construct a program. @command{gnatlink} can also incorporate foreign language
object units into the executable.

@item
@ref{The GNAT Make Program gnatmake}, describes @command{gnatmake}, a
utility that automatically determines the set of sources
needed by an Ada compilation unit, and executes the necessary compilations
binding and link.

@item
@ref{Improving Performance}, shows various techniques for making your
Ada program run faster or take less space and describes the effect of
the compiler's optimization switch.
It also describes
@ifclear FSFEDITION
the @command{gnatelim} tool and
@end ifclear
unused subprogram/data elimination.

@item
@ref{Renaming Files with gnatchop}, describes
@code{gnatchop}, a utility that allows you to preprocess a file that
contains Ada source code, and split it into one or more new files, one
for each compilation unit.

@item
@ref{Configuration Pragmas}, describes the configuration pragmas
handled by GNAT.

@item
@ref{Handling Arbitrary File Naming Conventions with gnatname},
shows how to override the default GNAT file naming conventions,
either for an individual unit or globally.

@item
@ref{GNAT Project Manager}, describes how to use project files
to organize large projects.

@item
@ref{The Cross-Referencing Tools gnatxref and gnatfind}, discusses
@code{gnatxref} and @code{gnatfind}, two tools that provide an easy
way to navigate through sources.

@ifclear FSFEDITION
@item
@ref{The GNAT Pretty-Printer gnatpp}, shows how to produce a reformatted
version of an Ada source file with control over casing, indentation,
comment placement, and other elements of program presentation style.
@end ifclear

@ifclear FSFEDITION
@item
@ref{The Ada-to-XML converter gnat2xml}, shows how to convert Ada
source code into XML.
@end ifclear

@ifclear FSFEDITION
@item
@ref{The GNAT Metrics Tool gnatmetric}, shows how to compute various
metrics for an Ada source file, such as the number of types and subprograms,
and assorted complexity measures.
@end ifclear

@item
@ref{File Name Krunching with gnatkr}, describes the @code{gnatkr}
file name krunching utility, used to handle shortened
file names on operating systems with a limit on the length of names.

@item
@ref{Preprocessing with gnatprep}, describes @code{gnatprep}, a
preprocessor utility that allows a single source file to be used to
generate multiple or parameterized source files by means of macro
substitution.

@item
@ref{The GNAT Library Browser gnatls}, describes @code{gnatls}, a
utility that displays information about compiled units, including dependences
on the corresponding sources files, and consistency of compilations.

@item
@ref{Cleaning Up with gnatclean}, describes @code{gnatclean}, a utility
to delete files that are produced by the compiler, binder and linker.

@item
@ref{GNAT and Libraries}, describes the process of creating and using
Libraries with GNAT. It also describes how to recompile the GNAT run-time
library.

@item
@ref{Using the GNU make Utility}, describes some techniques for using
the GNAT toolset in Makefiles.

@item
@ref{Memory Management Issues}, describes some useful predefined storage pools
and in particular the GNAT Debug Pool facility, which helps detect incorrect
memory references.
@ifclear FSFEDITION
It also describes @command{gnatmem}, a utility that monitors dynamic
allocation and deallocation and helps detect ``memory leaks''.
@end ifclear

@item
@ref{Stack Related Facilities}, describes some useful tools associated with
stack checking and analysis.

@ifclear FSFEDITION
@item
@ref{Verifying Properties with gnatcheck}, discusses @code{gnatcheck},
a utility that checks Ada code against a set of rules.

@item
@ref{Creating Sample Bodies with gnatstub}, discusses @code{gnatstub},
a utility that generates empty but compilable bodies for library units.
@end ifclear

@ifclear FSFEDITION
@item
@ref{Creating Unit Tests with gnattest}, discusses @code{gnattest},
a utility that generates unit testing templates for library units.
@end ifclear

@item
@ref{Performing Dimensionality Analysis in GNAT}, describes the Ada 2012
facilities used in GNAT to declare dimensioned objects, and to verify that
uses of these objects are consistent with their given physical dimensions
(so that meters cannot be assigned to kilograms, and so on).

@item
@ref{Generating Ada Bindings for C and C++ headers}, describes how to
generate automatically Ada bindings from C and C++ headers.

@item
@ref{Other Utility Programs}, discusses several other GNAT utilities,
including @code{gnathtml}.

@item
@ref{Code Coverage and Profiling}, describes how to perform a structural
coverage and profile the execution of Ada programs.

@item
@ref{Running and Debugging Ada Programs}, describes how to run and debug
Ada programs.


@item
@ref{Platform-Specific Information for the Run-Time Libraries},
describes the various run-time
libraries supported by GNAT on various platforms and explains how to
choose a particular library.

@item
@ref{Example of Binder Output File}, shows the source code for the binder
output file for a sample program.

@item
@ref{Elaboration Order Handling in GNAT}, describes how GNAT helps
you deal with elaboration order issues.

@item
@ref{Overflow Check Handling in GNAT}, describes how GNAT helps
you deal with arithmetic overflow issues.

@item
@ref{Conditional Compilation}, describes how to model conditional compilation,
both with Ada in general and with GNAT facilities in particular.

@item
@ref{Inline Assembler}, shows how to use the inline assembly facility
in an Ada program.

@item
@ref{Writing Portable Fixed-Point Declarations}, gives some guidance on
defining portable fixed-point types.

@item
@ref{Compatibility and Porting Guide}, contains sections on compatibility
of GNAT with other Ada development environments (including Ada 83 systems),
to assist in porting code from those environments.

@item
@ref{Microsoft Windows Topics}, presents information relevant to the
Microsoft Windows platform.

@item
@ref{Mac OS Topics}, presents information relevant to Apple's OS X
platform.
@end itemize

@c *************************************************
@node What You Should Know before Reading This Guide
@c *************************************************
@unnumberedsec What You Should Know before Reading This Guide

@cindex Ada 95 Language Reference Manual
@cindex Ada 2005 Language Reference Manual
@noindent
This guide assumes a basic familiarity with the Ada 95 language, as
described in the International Standard ANSI/ISO/IEC-8652:1995, January
1995.
It does not require knowledge of the new features introduced by Ada 2005,
(officially known as ISO/IEC 8652:1995 with Technical Corrigendum 1
and Amendment 1).
Both reference manuals are included in the GNAT documentation
package.

@node Related Information
@unnumberedsec Related Information

@noindent
For further information about related tools, refer to the following
documents:

@itemize @bullet
@item
@xref{Top, GNAT Reference Manual, About This Guide, gnat_rm, GNAT
Reference Manual}, which contains all reference material for the GNAT
implementation of Ada.

@item
@cite{Using the GNAT Programming Studio}, which describes the GPS
Integrated Development Environment.

@item
@cite{GNAT Programming Studio Tutorial}, which introduces the
main GPS features through examples.

@item
@cite{Ada 95 Reference Manual}, which contains reference
material for the Ada 95 programming language.

@item
@cite{Ada 2005 Reference Manual}, which contains reference
material for the Ada 2005 programming language.

@item
@xref{Top,, Debugging with GDB, gdb, Debugging with GDB},
for all details on the use of the GNU source-level debugger.

@item
@xref{Top,, The extensible self-documenting text editor, emacs,
GNU Emacs Manual},
for full information on the extensible editor and programming
environment Emacs.

@end itemize

@c **************
@node Conventions
@unnumberedsec Conventions
@cindex Conventions
@cindex Typographical conventions

@noindent
Following are examples of the typographical and graphic conventions used
in this guide:

@itemize @bullet
@item
@code{Functions}, @command{utility program names}, @code{standard names},
and @code{classes}.

@item
@option{Option flags}

@item
@file{File names}, @samp{button names}, and @samp{field names}.

@item
@code{Variables}, @env{environment variables}, and @var{metasyntactic
variables}.

@item
@emph{Emphasis}.

@item
@r{[}optional information or parameters@r{]}

@item
Examples are described by text
@smallexample
and then shown this way.
@end smallexample
@end itemize

@noindent
Commands that are entered by the user are preceded in this manual by the
characters @w{``@code{$ }''} (dollar sign followed by space). If your system
uses this sequence as a prompt, then the commands will appear exactly as
you see them in the manual. If your system uses some other prompt, then
the command will appear with the @code{$} replaced by whatever prompt
character you are using.

Full file names are shown with the ``@code{/}'' character
as the directory separator; e.g., @file{parent-dir/subdir/myfile.adb}.
If you are using GNAT on a Windows platform, please note that
the ``@code{\}'' character should be used instead.

@c ****************************
@node Getting Started with GNAT
@chapter Getting Started with GNAT

@noindent
This chapter describes some simple ways of using GNAT to build
executable Ada programs.
@ref{Running GNAT}, through @ref{Using the gnatmake Utility},
show how to use the command line environment.
@ref{Introduction to GPS}, provides a brief
introduction to the GNAT Programming Studio, a visually-oriented
Integrated Development Environment for GNAT.
GPS offers a graphical ``look and feel'', support for development in
other programming languages, comprehensive browsing features, and
many other capabilities.
For information on GPS please refer to
@cite{Using the GNAT Programming Studio}.

@menu
* Running GNAT::
* Running a Simple Ada Program::
* Running a Program with Multiple Units::
* Using the gnatmake Utility::
* Introduction to GPS::
@end menu

@node Running GNAT
@section Running GNAT

@noindent
Three steps are needed to create an executable file from an Ada source
file:

@enumerate
@item
The source file(s) must be compiled.
@item
The file(s) must be bound using the GNAT binder.
@item
All appropriate object files must be linked to produce an executable.
@end enumerate

@noindent
All three steps are most commonly handled by using the @command{gnatmake}
utility program that, given the name of the main program, automatically
performs the necessary compilation, binding and linking steps.

@node Running a Simple Ada Program
@section Running a Simple Ada Program

@noindent
Any text editor may be used to prepare an Ada program.
(If @code{Emacs} is
used, the optional Ada mode may be helpful in laying out the program.)
The
program text is a normal text file. We will assume in our initial
example that you have used your editor to prepare the following
standard format text file:

@smallexample @c ada
@cartouche
@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{procedure} Hello @b{is}
@b{begin}
   Put_Line ("Hello WORLD!");
@b{end} Hello;
@end cartouche
@end smallexample

@noindent
This file should be named @file{hello.adb}.
With the normal default file naming conventions, GNAT requires
that each file
contain a single compilation unit whose file name is the
unit name,
with periods replaced by hyphens; the
extension is @file{ads} for a
spec and @file{adb} for a body.
You can override this default file naming convention by use of the
special pragma @code{Source_File_Name} (@pxref{Using Other File Names}).
Alternatively, if you want to rename your files according to this default
convention, which is probably more convenient if you will be using GNAT
for all your compilations, then the @code{gnatchop} utility
can be used to generate correctly-named source files
(@pxref{Renaming Files with gnatchop}).

You can compile the program using the following command (@code{$} is used
as the command prompt in the examples in this document):

@smallexample
$ gcc -c hello.adb
@end smallexample

@noindent
@command{gcc} is the command used to run the compiler. This compiler is
capable of compiling programs in several languages, including Ada and
C. It assumes that you have given it an Ada program if the file extension is
either @file{.ads} or @file{.adb}, and it will then call
the GNAT compiler to compile the specified file.

The @option{-c} switch is required. It tells @command{gcc} to only do a
compilation. (For C programs, @command{gcc} can also do linking, but this
capability is not used directly for Ada programs, so the @option{-c}
switch must always be present.)

This compile command generates a file
@file{hello.o}, which is the object
file corresponding to your Ada program. It also generates
an ``Ada Library Information'' file @file{hello.ali},
which contains additional information used to check
that an Ada program is consistent.
To build an executable file,
use @code{gnatbind} to bind the program
and @command{gnatlink} to link it. The
argument to both @code{gnatbind} and @command{gnatlink} is the name of the
@file{ALI} file, but the default extension of @file{.ali} can
be omitted. This means that in the most common case, the argument
is simply the name of the main program:

@smallexample
$ gnatbind hello
$ gnatlink hello
@end smallexample

@noindent
A simpler method of carrying out these steps is to use
@command{gnatmake},
a master program that invokes all the required
compilation, binding and linking tools in the correct order. In particular,
@command{gnatmake} automatically recompiles any sources that have been
modified since they were last compiled, or sources that depend
on such modified sources, so that ``version skew'' is avoided.
@cindex Version skew (avoided by @command{gnatmake})

@smallexample
$ gnatmake hello.adb
@end smallexample

@noindent
The result is an executable program called @file{hello}, which can be
run by entering:

@smallexample
$ hello
@end smallexample

@noindent
assuming that the current directory is on the search path
for executable programs.

@noindent
and, if all has gone well, you will see

@smallexample
Hello WORLD!
@end smallexample

@noindent
appear in response to this command.

@c ****************************************
@node Running a Program with Multiple Units
@section Running a Program with Multiple Units

@noindent
Consider a slightly more complicated example that has three files: a
main program, and the spec and body of a package:

@smallexample @c ada
@cartouche
@group
@b{package} Greetings @b{is}
   @b{procedure} Hello;
   @b{procedure} Goodbye;
@b{end} Greetings;

@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{package} @b{body} Greetings @b{is}
   @b{procedure} Hello @b{is}
   @b{begin}
      Put_Line ("Hello WORLD!");
   @b{end} Hello;

   @b{procedure} Goodbye @b{is}
   @b{begin}
      Put_Line ("Goodbye WORLD!");
   @b{end} Goodbye;
@b{end} Greetings;
@end group

@group
@b{with} Greetings;
@b{procedure} Gmain @b{is}
@b{begin}
   Greetings.Hello;
   Greetings.Goodbye;
@b{end} Gmain;
@end group
@end cartouche
@end smallexample

@noindent
Following the one-unit-per-file rule, place this program in the
following three separate files:

@table @file
@item greetings.ads
spec of package @code{Greetings}

@item greetings.adb
body of package @code{Greetings}

@item gmain.adb
body of main program
@end table

@noindent
To build an executable version of
this program, we could use four separate steps to compile, bind, and link
the program, as follows:

@smallexample
$ gcc -c gmain.adb
$ gcc -c greetings.adb
$ gnatbind gmain
$ gnatlink gmain
@end smallexample

@noindent
Note that there is no required order of compilation when using GNAT.
In particular it is perfectly fine to compile the main program first.
Also, it is not necessary to compile package specs in the case where
there is an accompanying body; you only need to compile the body. If you want
to submit these files to the compiler for semantic checking and not code
generation, then use the
@option{-gnatc} switch:

@smallexample
$ gcc -c greetings.ads -gnatc
@end smallexample

@noindent
Although the compilation can be done in separate steps as in the
above example, in practice it is almost always more convenient
to use the @command{gnatmake} tool. All you need to know in this case
is the name of the main program's source file. The effect of the above four
commands can be achieved with a single one:

@smallexample
$ gnatmake gmain.adb
@end smallexample

@noindent
In the next section we discuss the advantages of using @command{gnatmake} in
more detail.

@c *****************************
@node Using the gnatmake Utility
@section Using the @command{gnatmake} Utility

@noindent
If you work on a program by compiling single components at a time using
@command{gcc}, you typically keep track of the units you modify. In order to
build a consistent system, you compile not only these units, but also any
units that depend on the units you have modified.
For example, in the preceding case,
if you edit @file{gmain.adb}, you only need to recompile that file. But if
you edit @file{greetings.ads}, you must recompile both
@file{greetings.adb} and @file{gmain.adb}, because both files contain
units that depend on @file{greetings.ads}.

@code{gnatbind} will warn you if you forget one of these compilation
steps, so that it is impossible to generate an inconsistent program as a
result of forgetting to do a compilation. Nevertheless it is tedious and
error-prone to keep track of dependencies among units.
One approach to handle the dependency-bookkeeping is to use a
makefile. However, makefiles present maintenance problems of their own:
if the dependencies change as you change the program, you must make
sure that the makefile is kept up-to-date manually, which is also an
error-prone process.

The @command{gnatmake} utility takes care of these details automatically.
Invoke it using either one of the following forms:

@smallexample
$ gnatmake gmain.adb
$ gnatmake gmain
@end smallexample

@noindent
The argument is the name of the file containing the main program;
you may omit the extension. @command{gnatmake}
examines the environment, automatically recompiles any files that need
recompiling, and binds and links the resulting set of object files,
generating the executable file, @file{gmain}.
In a large program, it
can be extremely helpful to use @command{gnatmake}, because working out by hand
what needs to be recompiled can be difficult.

Note that @command{gnatmake}
takes into account all the Ada rules that
establish dependencies among units. These include dependencies that result
from inlining subprogram bodies, and from
generic instantiation. Unlike some other
Ada make tools, @command{gnatmake} does not rely on the dependencies that were
found by the compiler on a previous compilation, which may possibly
be wrong when sources change. @command{gnatmake} determines the exact set of
dependencies from scratch each time it is run.


@node Introduction to GPS
@section Introduction to GPS
@cindex GPS (GNAT Programming Studio)
@cindex GNAT Programming Studio (GPS)
@noindent
Although the command line interface (@command{gnatmake}, etc.) alone
is sufficient, a graphical Interactive Development
Environment can make it easier for you to compose, navigate, and debug
programs.  This section describes the main features of GPS
(``GNAT Programming Studio''), the GNAT graphical IDE.
You will see how to use GPS to build and debug an executable, and
you will also learn some of the basics of the GNAT ``project'' facility.

GPS enables you to do much more than is presented here;
e.g., you can produce a call graph, interface to a third-party
Version Control System, and inspect the generated assembly language
for a program.
Indeed, GPS also supports languages other than Ada.
Such additional information, and an explanation of all of the GPS menu
items. may be found in the on-line help, which includes
a user's guide and a tutorial (these are also accessible from the GNAT
startup menu).

@menu
* Building a New Program with GPS::
* Simple Debugging with GPS::
@end menu

@node Building a New Program with GPS
@subsection Building a New Program with GPS
@noindent
GPS invokes the GNAT compilation tools using information
contained in a @emph{project} (also known as a @emph{project file}):
a collection of properties such
as source directories, identities of main subprograms, tool switches, etc.,
and their associated values.
See @ref{GNAT Project Manager} for details.
In order to run GPS, you will need to either create a new project
or else open an existing one.

This section will explain how you can use GPS to create a project,
to associate Ada source files with a project, and to build and run
programs.

@enumerate
@item @emph{Creating a project}

Invoke GPS, either from the command line or the platform's IDE.
After it starts, GPS will display a ``Welcome'' screen with three
radio buttons:

@itemize @bullet
@item
@code{Start with default project in directory}

@item
@code{Create new project with wizard}

@item
@code{Open existing project}
@end itemize

@noindent
Select @code{Create new project with wizard} and press @code{OK}.
A new window will appear.  In the text box labeled with
@code{Enter the name of the project to create}, type @file{sample}
as the project name.
In the next box, browse to choose the directory in which you
would like to create the project file.
After selecting an appropriate directory, press @code{Forward}.

A window will appear with the title
@code{Version Control System Configuration}.
Simply press @code{Forward}.

A window will appear with the title
@code{Please select the source directories for this project}.
The directory that you specified for the project file will be selected
by default as the one to use for sources; simply press @code{Forward}.

A window will appear with the title
@code{Please select the build directory for this project}.
The directory that you specified for the project file will be selected
by default for object files and executables;
simply press @code{Forward}.

A window will appear with the title
@code{Please select the main units for this project}.
You will supply this information later, after creating the source file.
Simply press @code{Forward} for now.

A window will appear with the title
@code{Please select the switches to build the project}.
Press @code{Apply}.  This will create a project file named
@file{sample.prj} in the directory that you had specified.

@item @emph{Creating and saving the source file}

After you create the new project, a GPS window will appear, which is
partitioned into two main sections:

@itemize @bullet
@item
A @emph{Workspace area}, initially greyed out, which you will use for
creating and editing source files

@item
Directly below, a @emph{Messages area}, which initially displays a
``Welcome'' message.
(If the Messages area is not visible, drag its border upward to expand it.)
@end itemize

@noindent
Select @code{File} on the menu bar, and then the @code{New} command.
The Workspace area will become white, and you can now
enter the source program explicitly.
Type the following text

@smallexample @c ada
@group
@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{procedure} Hello @b{is}
@b{begin}
  Put_Line("Hello from GPS!");
@b{end} Hello;
@end group
@end smallexample

@noindent
Select @code{File}, then @code{Save As}, and enter the source file name
@file{hello.adb}.
The file will be saved in the same directory you specified as the
location of the default project file.

@item @emph{Updating the project file}

You need to add the new source file to the project.
To do this, select
the @code{Project} menu and then @code{Edit project properties}.
Click the @code{Main files} tab on the left, and then the
@code{Add} button.
Choose @file{hello.adb} from the list, and press @code{Open}.
The project settings window will reflect this action.
Click @code{OK}.

@item @emph{Building and running the program}

In the main GPS window, now choose the @code{Build} menu, then @code{Make},
and select @file{hello.adb}.
The Messages window will display the resulting invocations of @command{gcc},
@command{gnatbind}, and @command{gnatlink}
(reflecting the default switch settings from the
project file that you created) and then a ``successful compilation/build''
message.

To run the program, choose the @code{Build} menu, then @code{Run}, and
select @command{hello}.
An @emph{Arguments Selection} window will appear.
There are no command line arguments, so just click @code{OK}.

The Messages window will now display the program's output (the string
@code{Hello from GPS}), and at the bottom of the GPS window a status
update is displayed (@code{Run: hello}).
Close the GPS window (or select @code{File}, then @code{Exit}) to
terminate this GPS session.
@end enumerate

@node Simple Debugging with GPS
@subsection Simple Debugging with GPS
@noindent
This section illustrates basic debugging techniques (setting breakpoints,
examining/modifying variables, single stepping).

@enumerate
@item @emph{Opening a project}

Start GPS and select @code{Open existing project}; browse to
specify the project file @file{sample.prj} that you had created in the
earlier example.

@item @emph{Creating a source file}

Select @code{File}, then @code{New}, and type in the following program:

@smallexample @c ada
@group
@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{procedure} Example @b{is}
   Line : String (1..80);
   N    : Natural;
@b{begin}
   Put_Line("Type a line of text at each prompt; an empty line to exit");
   @b{loop}
      Put(": ");
      Get_Line (Line, N);
      Put_Line (Line (1..N) );
      @b{exit} @b{when} N=0;
   @b{end} @b{loop};
@b{end} Example;
@end group
@end smallexample

@noindent
Select @code{File}, then @code{Save as}, and enter the file name
@file{example.adb}.

@item @emph{Updating the project file}

Add @code{Example} as a new main unit for the project:
@enumerate a
@item
Select @code{Project}, then @code{Edit Project Properties}.

@item
Select the @code{Main files} tab, click @code{Add}, then
select the file @file{example.adb} from the list, and
click @code{Open}.
You will see the file name appear in the list of main units

@item
Click @code{OK}
@end enumerate

@item @emph{Building/running the executable}

To build the executable
select @code{Build}, then @code{Make}, and then choose @file{example.adb}.

Run the program to see its effect (in the Messages area).
Each line that you enter is displayed; an empty line will
cause the loop to exit and the program to terminate.

@item @emph{Debugging the program}

Note that the @option{-g} switches to @command{gcc} and @command{gnatlink},
which are required for debugging, are on by default when you create
a new project.
Thus unless you intentionally remove these settings, you will be able
to debug any program that you develop using GPS.

@enumerate a
@item @emph{Initializing}

Select @code{Debug}, then @code{Initialize}, then @file{example}

@item @emph{Setting a breakpoint}

After performing the initialization step, you will observe a small
icon to the right of each line number.
This serves as a toggle for breakpoints; clicking the icon will
set a breakpoint at the corresponding line (the icon will change to
a red circle with an ``x''), and clicking it again
will remove the breakpoint / reset the icon.

For purposes of this example, set a breakpoint at line 10 (the
statement @code{Put_Line@ (Line@ (1..N));}

@item @emph{Starting program execution}

Select @code{Debug}, then @code{Run}.  When the
@code{Program Arguments} window appears, click @code{OK}.
A console window will appear; enter some line of text,
e.g.@: @code{abcde}, at the prompt.
The program will pause execution when it gets to the
breakpoint, and the corresponding line is highlighted.

@item @emph{Examining a variable}

Move the mouse over one of the occurrences of the variable @code{N}.
You will see the value (5) displayed, in ``tool tip'' fashion.
Right click on @code{N}, select @code{Debug}, then select @code{Display N}.
You will see information about @code{N} appear in the @code{Debugger Data}
pane, showing the value as 5.

@item @emph{Assigning a new value to a variable}

Right click on the @code{N} in the @code{Debugger Data} pane, and
select @code{Set value of N}.
When the input window appears, enter the value @code{4} and click
@code{OK}.
This value does not automatically appear in the @code{Debugger Data}
pane; to see it, right click again on the @code{N} in the
@code{Debugger Data} pane and select @code{Update value}.
The new value, 4, will appear in red.

@item @emph{Single stepping}

Select @code{Debug}, then @code{Next}.
This will cause the next statement to be executed, in this case the
call of @code{Put_Line} with the string slice.
Notice in the console window that the displayed string is simply
@code{abcd} and not @code{abcde} which you had entered.
This is because the upper bound of the slice is now 4 rather than 5.

@item @emph{Removing a breakpoint}

Toggle the breakpoint icon at line 10.

@item @emph{Resuming execution from a breakpoint}

Select @code{Debug}, then @code{Continue}.
The program will reach the next iteration of the loop, and
wait for input after displaying the prompt.
This time, just hit the @kbd{Enter} key.
The value of @code{N} will be 0, and the program will terminate.
The console window will disappear.
@end enumerate
@end enumerate

@node The GNAT Compilation Model
@chapter The GNAT Compilation Model
@cindex GNAT compilation model
@cindex Compilation model

@menu
* Source Representation::
* Foreign Language Representation::
* File Naming Rules::
* Using Other File Names::
* Alternative File Naming Schemes::
* Generating Object Files::
* Source Dependencies::
* The Ada Library Information Files::
* Binding an Ada Program::
* Mixed Language Programming::
* Building Mixed Ada & C++ Programs::
* Comparison between GNAT and C/C++ Compilation Models::
* Comparison between GNAT and Conventional Ada Library Models::
@end menu

@noindent
This chapter describes the compilation model used by GNAT. Although
similar to that used by other languages, such as C and C++, this model
is substantially different from the traditional Ada compilation models,
which are based on a library. The model is initially described without
reference to the library-based model. If you have not previously used an
Ada compiler, you need only read the first part of this chapter. The
last section describes and discusses the differences between the GNAT
model and the traditional Ada compiler models. If you have used other
Ada compilers, this section will help you to understand those
differences, and the advantages of the GNAT model.

@node Source Representation
@section Source Representation
@cindex Latin-1

@noindent
Ada source programs are represented in standard text files, using
Latin-1 coding. Latin-1 is an 8-bit code that includes the familiar
7-bit ASCII set, plus additional characters used for
representing foreign languages (@pxref{Foreign Language Representation}
for support of non-USA character sets). The format effector characters
are represented using their standard ASCII encodings, as follows:

@table @code
@item VT
@findex VT
Vertical tab, @code{16#0B#}

@item HT
@findex HT
Horizontal tab, @code{16#09#}

@item CR
@findex CR
Carriage return, @code{16#0D#}

@item LF
@findex LF
Line feed, @code{16#0A#}

@item FF
@findex FF
Form feed, @code{16#0C#}
@end table

@noindent
Source files are in standard text file format. In addition, GNAT will
recognize a wide variety of stream formats, in which the end of
physical lines is marked by any of the following sequences:
@code{LF}, @code{CR}, @code{CR-LF}, or @code{LF-CR}. This is useful
in accommodating files that are imported from other operating systems.

@cindex End of source file
@cindex Source file, end
@findex SUB
The end of a source file is normally represented by the physical end of
file. However, the control character @code{16#1A#} (@code{SUB}) is also
recognized as signalling the end of the source file. Again, this is
provided for compatibility with other operating systems where this
code is used to represent the end of file.

Each file contains a single Ada compilation unit, including any pragmas
associated with the unit. For example, this means you must place a
package declaration (a package @dfn{spec}) and the corresponding body in
separate files. An Ada @dfn{compilation} (which is a sequence of
compilation units) is represented using a sequence of files. Similarly,
you will place each subunit or child unit in a separate file.

@node Foreign Language Representation
@section Foreign Language Representation

@noindent
GNAT supports the standard character sets defined in Ada as well as
several other non-standard character sets for use in localized versions
of the compiler (@pxref{Character Set Control}).
@menu
* Latin-1::
* Other 8-Bit Codes::
* Wide_Character Encodings::
* Wide_Wide_Character Encodings::
@end menu

@node Latin-1
@subsection Latin-1
@cindex Latin-1

@noindent
The basic character set is Latin-1. This character set is defined by ISO
standard 8859, part 1. The lower half (character codes @code{16#00#}
@dots{} @code{16#7F#)} is identical to standard ASCII coding, but the upper
half is used to represent additional characters. These include extended letters
used by European languages, such as French accents, the vowels with umlauts
used in German, and the extra letter A-ring used in Swedish.

@findex Ada.Characters.Latin_1
For a complete list of Latin-1 codes and their encodings, see the source
file of library unit @code{Ada.Characters.Latin_1} in file
@file{a-chlat1.ads}.
You may use any of these extended characters freely in character or
string literals. In addition, the extended characters that represent
letters can be used in identifiers.

@node Other 8-Bit Codes
@subsection Other 8-Bit Codes

@noindent
GNAT also supports several other 8-bit coding schemes:

@table @asis
@item ISO 8859-2 (Latin-2)
@cindex Latin-2
@cindex ISO 8859-2
Latin-2 letters allowed in identifiers, with uppercase and lowercase
equivalence.

@item ISO 8859-3 (Latin-3)
@cindex Latin-3
@cindex ISO 8859-3
Latin-3 letters allowed in identifiers, with uppercase and lowercase
equivalence.

@item ISO 8859-4 (Latin-4)
@cindex Latin-4
@cindex ISO 8859-4
Latin-4 letters allowed in identifiers, with uppercase and lowercase
equivalence.

@item ISO 8859-5 (Cyrillic)
@cindex ISO 8859-5
@cindex Cyrillic
ISO 8859-5 letters (Cyrillic) allowed in identifiers, with uppercase and
lowercase equivalence.

@item ISO 8859-15 (Latin-9)
@cindex ISO 8859-15
@cindex Latin-9
ISO 8859-15 (Latin-9) letters allowed in identifiers, with uppercase and
lowercase equivalence

@item IBM PC (code page 437)
@cindex code page 437
This code page is the normal default for PCs in the U.S. It corresponds
to the original IBM PC character set. This set has some, but not all, of
the extended Latin-1 letters, but these letters do not have the same
encoding as Latin-1. In this mode, these letters are allowed in
identifiers with uppercase and lowercase equivalence.

@item IBM PC (code page 850)
@cindex code page 850
This code page is a modification of 437 extended to include all the
Latin-1 letters, but still not with the usual Latin-1 encoding. In this
mode, all these letters are allowed in identifiers with uppercase and
lowercase equivalence.

@item Full Upper 8-bit
Any character in the range 80-FF allowed in identifiers, and all are
considered distinct. In other words, there are no uppercase and lowercase
equivalences in this range. This is useful in conjunction with
certain encoding schemes used for some foreign character sets (e.g.,
the typical method of representing Chinese characters on the PC).

@item No Upper-Half
No upper-half characters in the range 80-FF are allowed in identifiers.
This gives Ada 83 compatibility for identifier names.
@end table

@noindent
For precise data on the encodings permitted, and the uppercase and lowercase
equivalences that are recognized, see the file @file{csets.adb} in
the GNAT compiler sources. You will need to obtain a full source release
of GNAT to obtain this file.

@node Wide_Character Encodings
@subsection Wide_Character Encodings

@noindent
GNAT allows wide character codes to appear in character and string
literals, and also optionally in identifiers, by means of the following
possible encoding schemes:

@table @asis

@item Hex Coding
In this encoding, a wide character is represented by the following five
character sequence:

@smallexample
ESC a b c d
@end smallexample

@noindent
Where @code{a}, @code{b}, @code{c}, @code{d} are the four hexadecimal
characters (using uppercase letters) of the wide character code. For
example, ESC A345 is used to represent the wide character with code
@code{16#A345#}.
This scheme is compatible with use of the full Wide_Character set.

@item Upper-Half Coding
@cindex Upper-Half Coding
The wide character with encoding @code{16#abcd#} where the upper bit is on
(in other words, ``a'' is in the range 8-F) is represented as two bytes,
@code{16#ab#} and @code{16#cd#}. The second byte cannot be a format control
character, but is not required to be in the upper half. This method can
be also used for shift-JIS or EUC, where the internal coding matches the
external coding.

@item Shift JIS Coding
@cindex Shift JIS Coding
A wide character is represented by a two-character sequence,
@code{16#ab#} and
@code{16#cd#}, with the restrictions described for upper-half encoding as
described above. The internal character code is the corresponding JIS
character according to the standard algorithm for Shift-JIS
conversion. Only characters defined in the JIS code set table can be
used with this encoding method.

@item EUC Coding
@cindex EUC Coding
A wide character is represented by a two-character sequence
@code{16#ab#} and
@code{16#cd#}, with both characters being in the upper half. The internal
character code is the corresponding JIS character according to the EUC
encoding algorithm. Only characters defined in the JIS code set table
can be used with this encoding method.

@item UTF-8 Coding
A wide character is represented using
UCS Transformation Format 8 (UTF-8) as defined in Annex R of ISO
10646-1/Am.2. Depending on the character value, the representation
is a one, two, or three byte sequence:
@smallexample
@iftex
@leftskip=.7cm
@end iftex
16#0000#-16#007f#: 2#0@var{xxxxxxx}#
16#0080#-16#07ff#: 2#110@var{xxxxx}# 2#10@var{xxxxxx}#
16#0800#-16#ffff#: 2#1110@var{xxxx}# 2#10@var{xxxxxx}# 2#10@var{xxxxxx}#

@end smallexample

@noindent
where the @var{xxx} bits correspond to the left-padded bits of the
16-bit character value. Note that all lower half ASCII characters
are represented as ASCII bytes and all upper half characters and
other wide characters are represented as sequences of upper-half
(The full UTF-8 scheme allows for encoding 31-bit characters as
6-byte sequences, and in the following section on wide wide
characters, the use of these sequences is documented).

@item Brackets Coding
In this encoding, a wide character is represented by the following eight
character sequence:

@smallexample
[ " a b c d " ]
@end smallexample

@noindent
Where @code{a}, @code{b}, @code{c}, @code{d} are the four hexadecimal
characters (using uppercase letters) of the wide character code. For
example, [``A345''] is used to represent the wide character with code
@code{16#A345#}. It is also possible (though not required) to use the
Brackets coding for upper half characters. For example, the code
@code{16#A3#} can be represented as @code{[``A3'']}.

This scheme is compatible with use of the full Wide_Character set,
and is also the method used for wide character encoding in some standard
ACATS (Ada Conformity Assessment Test Suite) test suite distributions.

@end table

@noindent
Note: Some of these coding schemes do not permit the full use of the
Ada character set. For example, neither Shift JIS, nor EUC allow the
use of the upper half of the Latin-1 set.

@node Wide_Wide_Character Encodings
@subsection Wide_Wide_Character Encodings

@noindent
GNAT allows wide wide character codes to appear in character and string
literals, and also optionally in identifiers, by means of the following
possible encoding schemes:

@table @asis

@item UTF-8 Coding
A wide character is represented using
UCS Transformation Format 8 (UTF-8) as defined in Annex R of ISO
10646-1/Am.2. Depending on the character value, the representation
of character codes with values greater than 16#FFFF# is a
is a four, five, or six byte sequence:

@smallexample
@iftex
@leftskip=.7cm
@end iftex
16#01_0000#-16#10_FFFF#:     11110xxx 10xxxxxx 10xxxxxx
                             10xxxxxx
16#0020_0000#-16#03FF_FFFF#: 111110xx 10xxxxxx 10xxxxxx
                             10xxxxxx 10xxxxxx
16#0400_0000#-16#7FFF_FFFF#: 1111110x 10xxxxxx 10xxxxxx
                             10xxxxxx 10xxxxxx 10xxxxxx
@end smallexample

@noindent
where the @var{xxx} bits correspond to the left-padded bits of the
32-bit character value.

@item Brackets Coding
In this encoding, a wide wide character is represented by the following ten or
twelve byte character sequence:

@smallexample
[ " a b c d e f " ]
[ " a b c d e f g h " ]
@end smallexample

@noindent
Where @code{a-h} are the six or eight hexadecimal
characters (using uppercase letters) of the wide wide character code. For
example, ["1F4567"] is used to represent the wide wide character with code
@code{16#001F_4567#}.

This scheme is compatible with use of the full Wide_Wide_Character set,
and is also the method used for wide wide character encoding in some standard
ACATS (Ada Conformity Assessment Test Suite) test suite distributions.

@end table

@node File Naming Rules
@section File Naming Rules

@noindent
The default file name is determined by the name of the unit that the
file contains. The name is formed by taking the full expanded name of
the unit and replacing the separating dots with hyphens and using
lowercase for all letters.

An exception arises if the file name generated by the above rules starts
with one of the characters
@samp{a}, @samp{g}, @samp{i}, or @samp{s},
and the second character is a
minus. In this case, the character tilde is used in place
of the minus. The reason for this special rule is to avoid clashes with
the standard names for child units of the packages System, Ada,
Interfaces, and GNAT, which use the prefixes
@samp{s-}, @samp{a-}, @samp{i-}, and @samp{g-},
respectively.

The file extension is @file{.ads} for a spec and
@file{.adb} for a body. The following list shows some
examples of these rules.

@table @file
@item main.ads
Main (spec)
@item main.adb
Main (body)
@item arith_functions.ads
Arith_Functions (package spec)
@item arith_functions.adb
Arith_Functions (package body)
@item func-spec.ads
Func.Spec (child package spec)
@item func-spec.adb
Func.Spec (child package body)
@item main-sub.adb
Sub (subunit of Main)
@item a~bad.adb
A.Bad (child package body)
@end table

@noindent
Following these rules can result in excessively long
file names if corresponding
unit names are long (for example, if child units or subunits are
heavily nested). An option is available to shorten such long file names
(called file name ``krunching''). This may be particularly useful when
programs being developed with GNAT are to be used on operating systems
with limited file name lengths. @xref{Using gnatkr}.

Of course, no file shortening algorithm can guarantee uniqueness over
all possible unit names; if file name krunching is used, it is your
responsibility to ensure no name clashes occur. Alternatively you
can specify the exact file names that you want used, as described
in the next section. Finally, if your Ada programs are migrating from a
compiler with a different naming convention, you can use the gnatchop
utility to produce source files that follow the GNAT naming conventions.
(For details @pxref{Renaming Files with gnatchop}.)

Note: in the case of @code{Windows NT/XP} or @code{OpenVMS} operating
systems, case is not significant. So for example on @code{Windows XP}
if the canonical name is @code{main-sub.adb}, you can use the file name
@code{Main-Sub.adb} instead. However, case is significant for other
operating systems, so for example, if you want to use other than
canonically cased file names on a Unix system, you need to follow
the procedures described in the next section.

@node Using Other File Names
@section Using Other File Names
@cindex File names

@noindent
In the previous section, we have described the default rules used by
GNAT to determine the file name in which a given unit resides. It is
often convenient to follow these default rules, and if you follow them,
the compiler knows without being explicitly told where to find all
the files it needs.

However, in some cases, particularly when a program is imported from
another Ada compiler environment, it may be more convenient for the
programmer to specify which file names contain which units. GNAT allows
arbitrary file names to be used by means of the Source_File_Name pragma.
The form of this pragma is as shown in the following examples:
@cindex Source_File_Name pragma

@smallexample @c ada
@cartouche
@b{pragma} Source_File_Name (My_Utilities.Stacks,
  Spec_File_Name => "myutilst_a.ada");
@b{pragma} Source_File_name (My_Utilities.Stacks,
  Body_File_Name => "myutilst.ada");
@end cartouche
@end smallexample

@noindent
As shown in this example, the first argument for the pragma is the unit
name (in this example a child unit). The second argument has the form
of a named association. The identifier
indicates whether the file name is for a spec or a body;
the file name itself is given by a string literal.

The source file name pragma is a configuration pragma, which means that
normally it will be placed in the @file{gnat.adc}
file used to hold configuration
pragmas that apply to a complete compilation environment.
For more details on how the @file{gnat.adc} file is created and used
see @ref{Handling of Configuration Pragmas}.
@cindex @file{gnat.adc}

GNAT allows completely arbitrary file names to be specified using the
source file name pragma. However, if the file name specified has an
extension other than @file{.ads} or @file{.adb} it is necessary to use
a special syntax when compiling the file. The name in this case must be
preceded by the special sequence @option{-x} followed by a space and the name
of the language, here @code{ada}, as in:

@smallexample
$ gcc -c -x ada peculiar_file_name.sim
@end smallexample

@noindent
@command{gnatmake} handles non-standard file names in the usual manner (the
non-standard file name for the main program is simply used as the
argument to gnatmake). Note that if the extension is also non-standard,
then it must be included in the @command{gnatmake} command, it may not
be omitted.

@node Alternative File Naming Schemes
@section Alternative File Naming Schemes
@cindex File naming schemes, alternative
@cindex File names

In the previous section, we described the use of the @code{Source_File_Name}
pragma to allow arbitrary names to be assigned to individual source files.
However, this approach requires one pragma for each file, and especially in
large systems can result in very long @file{gnat.adc} files, and also create
a maintenance problem.

GNAT also provides a facility for specifying systematic file naming schemes
other than the standard default naming scheme previously described. An
alternative scheme for naming is specified by the use of
@code{Source_File_Name} pragmas having the following format:
@cindex Source_File_Name pragma

@smallexample @c ada
@b{pragma} Source_File_Name (
   Spec_File_Name  => FILE_NAME_PATTERN
 @r{[},Casing          => CASING_SPEC@r{]}
 @r{[},Dot_Replacement => STRING_LITERAL@r{]});

@b{pragma} Source_File_Name (
   Body_File_Name  => FILE_NAME_PATTERN
 @r{[},Casing          => CASING_SPEC@r{]}
 @r{[},Dot_Replacement => STRING_LITERAL@r{]});

@b{pragma} Source_File_Name (
   Subunit_File_Name  => FILE_NAME_PATTERN
 @r{[},Casing             => CASING_SPEC@r{]}
 @r{[},Dot_Replacement    => STRING_LITERAL@r{]});

FILE_NAME_PATTERN ::= STRING_LITERAL
CASING_SPEC ::= Lowercase | Uppercase | Mixedcase
@end smallexample

@noindent
The @code{FILE_NAME_PATTERN} string shows how the file name is constructed.
It contains a single asterisk character, and the unit name is substituted
systematically for this asterisk. The optional parameter
@code{Casing} indicates
whether the unit name is to be all upper-case letters, all lower-case letters,
or mixed-case. If no
@code{Casing} parameter is used, then the default is all
lower-case.

The optional @code{Dot_Replacement} string is used to replace any periods
that occur in subunit or child unit names. If no @code{Dot_Replacement}
argument is used then separating dots appear unchanged in the resulting
file name.
Although the above syntax indicates that the
@code{Casing} argument must appear
before the @code{Dot_Replacement} argument, but it
is also permissible to write these arguments in the opposite order.

As indicated, it is possible to specify different naming schemes for
bodies, specs, and subunits. Quite often the rule for subunits is the
same as the rule for bodies, in which case, there is no need to give
a separate @code{Subunit_File_Name} rule, and in this case the
@code{Body_File_name} rule is used for subunits as well.

The separate rule for subunits can also be used to implement the rather
unusual case of a compilation environment (e.g.@: a single directory) which
contains a subunit and a child unit with the same unit name. Although
both units cannot appear in the same partition, the Ada Reference Manual
allows (but does not require) the possibility of the two units coexisting
in the same environment.

The file name translation works in the following steps:

@itemize @bullet

@item
If there is a specific @code{Source_File_Name} pragma for the given unit,
then this is always used, and any general pattern rules are ignored.

@item
If there is a pattern type @code{Source_File_Name} pragma that applies to
the unit, then the resulting file name will be used if the file exists. If
more than one pattern matches, the latest one will be tried first, and the
first attempt resulting in a reference to a file that exists will be used.

@item
If no pattern type @code{Source_File_Name} pragma that applies to the unit
for which the corresponding file exists, then the standard GNAT default
naming rules are used.

@end itemize

@noindent
As an example of the use of this mechanism, consider a commonly used scheme
in which file names are all lower case, with separating periods copied
unchanged to the resulting file name, and specs end with @file{.1.ada}, and
bodies end with @file{.2.ada}. GNAT will follow this scheme if the following
two pragmas appear:

@smallexample @c ada
@b{pragma} Source_File_Name
  (Spec_File_Name => "*.1.ada");
@b{pragma} Source_File_Name
  (Body_File_Name => "*.2.ada");
@end smallexample

@noindent
The default GNAT scheme is actually implemented by providing the following
default pragmas internally:

@smallexample @c ada
@b{pragma} Source_File_Name
  (Spec_File_Name => "*.ads", Dot_Replacement => "-");
@b{pragma} Source_File_Name
  (Body_File_Name => "*.adb", Dot_Replacement => "-");
@end smallexample

@noindent
Our final example implements a scheme typically used with one of the
Ada 83 compilers, where the separator character for subunits was ``__''
(two underscores), specs were identified by adding @file{_.ADA}, bodies
by adding @file{.ADA}, and subunits by
adding @file{.SEP}. All file names were
upper case. Child units were not present of course since this was an
Ada 83 compiler, but it seems reasonable to extend this scheme to use
the same double underscore separator for child units.

@smallexample @c ada
@b{pragma} Source_File_Name
  (Spec_File_Name => "*_.ADA",
   Dot_Replacement => "__",
   Casing = Uppercase);
@b{pragma} Source_File_Name
  (Body_File_Name => "*.ADA",
   Dot_Replacement => "__",
   Casing = Uppercase);
@b{pragma} Source_File_Name
  (Subunit_File_Name => "*.SEP",
   Dot_Replacement => "__",
   Casing = Uppercase);
@end smallexample

@node Generating Object Files
@section Generating Object Files

@noindent
An Ada program consists of a set of source files, and the first step in
compiling the program is to generate the corresponding object files.
These are generated by compiling a subset of these source files.
The files you need to compile are the following:

@itemize @bullet
@item
If a package spec has no body, compile the package spec to produce the
object file for the package.

@item
If a package has both a spec and a body, compile the body to produce the
object file for the package. The source file for the package spec need
not be compiled in this case because there is only one object file, which
contains the code for both the spec and body of the package.

@item
For a subprogram, compile the subprogram body to produce the object file
for the subprogram. The spec, if one is present, is as usual in a
separate file, and need not be compiled.

@item
@cindex Subunits
In the case of subunits, only compile the parent unit. A single object
file is generated for the entire subunit tree, which includes all the
subunits.

@item
Compile child units independently of their parent units
(though, of course, the spec of all the ancestor unit must be present in order
to compile a child unit).

@item
@cindex Generics
Compile generic units in the same manner as any other units. The object
files in this case are small dummy files that contain at most the
flag used for elaboration checking. This is because GNAT always handles generic
instantiation by means of macro expansion. However, it is still necessary to
compile generic units, for dependency checking and elaboration purposes.
@end itemize

@noindent
The preceding rules describe the set of files that must be compiled to
generate the object files for a program. Each object file has the same
name as the corresponding source file, except that the extension is
@file{.o} as usual.

You may wish to compile other files for the purpose of checking their
syntactic and semantic correctness. For example, in the case where a
package has a separate spec and body, you would not normally compile the
spec. However, it is convenient in practice to compile the spec to make
sure it is error-free before compiling clients of this spec, because such
compilations will fail if there is an error in the spec.

GNAT provides an option for compiling such files purely for the
purposes of checking correctness; such compilations are not required as
part of the process of building a program. To compile a file in this
checking mode, use the @option{-gnatc} switch.

@node Source Dependencies
@section Source Dependencies

@noindent
A given object file clearly depends on the source file which is compiled
to produce it. Here we are using @dfn{depends} in the sense of a typical
@code{make} utility; in other words, an object file depends on a source
file if changes to the source file require the object file to be
recompiled.
In addition to this basic dependency, a given object may depend on
additional source files as follows:

@itemize @bullet
@item
If a file being compiled @code{with}'s a unit @var{X}, the object file
depends on the file containing the spec of unit @var{X}. This includes
files that are @code{with}'ed implicitly either because they are parents
of @code{with}'ed child units or they are run-time units required by the
language constructs used in a particular unit.

@item
If a file being compiled instantiates a library level generic unit, the
object file depends on both the spec and body files for this generic
unit.

@item
If a file being compiled instantiates a generic unit defined within a
package, the object file depends on the body file for the package as
well as the spec file.

@item
@findex Inline
@cindex @option{-gnatn} switch
If a file being compiled contains a call to a subprogram for which
pragma @code{Inline} applies and inlining is activated with the
@option{-gnatn} switch, the object file depends on the file containing the
body of this subprogram as well as on the file containing the spec. Note
that for inlining to actually occur as a result of the use of this switch,
it is necessary to compile in optimizing mode.

@cindex @option{-gnatN} switch
The use of @option{-gnatN} activates  inlining optimization
that is performed by the front end of the compiler. This inlining does
not require that the code generation be optimized. Like @option{-gnatn},
the use of this switch generates additional dependencies.

When using a gcc-based back end (in practice this means using any version
of GNAT other than the JGNAT, .NET or GNAAMP versions), then the use of
@option{-gnatN} is deprecated, and the use of @option{-gnatn} is preferred.
Historically front end inlining was more extensive than the gcc back end
inlining, but that is no longer the case.

@item
If an object file @file{O} depends on the proper body of a subunit through
inlining or instantiation, it depends on the parent unit of the subunit.
This means that any modification of the parent unit or one of its subunits
affects the compilation of @file{O}.

@item
The object file for a parent unit depends on all its subunit body files.

@item
The previous two rules meant that for purposes of computing dependencies and
recompilation, a body and all its subunits are treated as an indivisible whole.

@noindent
These rules are applied transitively: if unit @code{A} @code{with}'s
unit @code{B}, whose elaboration calls an inlined procedure in package
@code{C}, the object file for unit @code{A} will depend on the body of
@code{C}, in file @file{c.adb}.

The set of dependent files described by these rules includes all the
files on which the unit is semantically dependent, as dictated by the
Ada language standard. However, it is a superset of what the
standard describes, because it includes generic, inline, and subunit
dependencies.

An object file must be recreated by recompiling the corresponding source
file if any of the source files on which it depends are modified. For
example, if the @code{make} utility is used to control compilation,
the rule for an Ada object file must mention all the source files on
which the object file depends, according to the above definition.
The determination of the necessary
recompilations is done automatically when one uses @command{gnatmake}.
@end itemize

@node The Ada Library Information Files
@section The Ada Library Information Files
@cindex Ada Library Information files
@cindex @file{ALI} files

@noindent
Each compilation actually generates two output files. The first of these
is the normal object file that has a @file{.o} extension. The second is a
text file containing full dependency information. It has the same
name as the source file, but an @file{.ali} extension.
This file is known as the Ada Library Information (@file{ALI}) file.
The following information is contained in the @file{ALI} file.

@itemize @bullet
@item
Version information (indicates which version of GNAT was used to compile
the unit(s) in question)

@item
Main program information (including priority and time slice settings,
as well as the wide character encoding used during compilation).

@item
List of arguments used in the @command{gcc} command for the compilation

@item
Attributes of the unit, including configuration pragmas used, an indication
of whether the compilation was successful, exception model used etc.

@item
A list of relevant restrictions applying to the unit (used for consistency)
checking.

@item
Categorization information (e.g.@: use of pragma @code{Pure}).

@item
Information on all @code{with}'ed units, including presence of
@code{Elaborate} or @code{Elaborate_All} pragmas.

@item
Information from any @code{Linker_Options} pragmas used in the unit

@item
Information on the use of @code{Body_Version} or @code{Version}
attributes in the unit.

@item
Dependency information. This is a list of files, together with
time stamp and checksum information. These are files on which
the unit depends in the sense that recompilation is required
if any of these units are modified.

@item
Cross-reference data. Contains information on all entities referenced
in the unit. Used by tools like @code{gnatxref} and @code{gnatfind} to
provide cross-reference information.

@end itemize

@noindent
For a full detailed description of the format of the @file{ALI} file,
see the source of the body of unit @code{Lib.Writ}, contained in file
@file{lib-writ.adb} in the GNAT compiler sources.

@node Binding an Ada Program
@section Binding an Ada Program

@noindent
When using languages such as C and C++, once the source files have been
compiled the only remaining step in building an executable program
is linking the object modules together. This means that it is possible to
link an inconsistent version of a program, in which two units have
included different versions of the same header.

The rules of Ada do not permit such an inconsistent program to be built.
For example, if two clients have different versions of the same package,
it is illegal to build a program containing these two clients.
These rules are enforced by the GNAT binder, which also determines an
elaboration order consistent with the Ada rules.

The GNAT binder is run after all the object files for a program have
been created. It is given the name of the main program unit, and from
this it determines the set of units required by the program, by reading the
corresponding ALI files. It generates error messages if the program is
inconsistent or if no valid order of elaboration exists.

If no errors are detected, the binder produces a main program, in Ada by
default, that contains calls to the elaboration procedures of those
compilation unit that require them, followed by
a call to the main program. This Ada program is compiled to generate the
object file for the main program. The name of
the Ada file is @file{b~@var{xxx}.adb} (with the corresponding spec
@file{b~@var{xxx}.ads}) where @var{xxx} is the name of the
main program unit.

Finally, the linker is used to build the resulting executable program,
using the object from the main program from the bind step as well as the
object files for the Ada units of the program.

@node Mixed Language Programming
@section Mixed Language Programming
@cindex Mixed Language Programming

@noindent
This section describes how to develop a mixed-language program,
specifically one that comprises units in both Ada and C.

@menu
* Interfacing to C::
* Calling Conventions::
@end menu

@node Interfacing to C
@subsection Interfacing to C
@noindent
Interfacing Ada with a foreign language such as C involves using
compiler directives to import and/or export entity definitions in each
language---using @code{extern} statements in C, for instance, and the
@code{Import}, @code{Export}, and @code{Convention} pragmas in Ada.
A full treatment of these topics is provided in Appendix B, section 1
of the Ada Reference Manual.

There are two ways to build a program using GNAT that contains some Ada
sources and some foreign language sources, depending on whether or not
the main subprogram is written in Ada.  Here is a source example with
the main subprogram in Ada:

@smallexample
/* file1.c */
#include <stdio.h>

void print_num (int num)
@{
  printf ("num is %d.\n", num);
  return;
@}

/* file2.c */

/* num_from_Ada is declared in my_main.adb */
extern int num_from_Ada;

int get_num (void)
@{
  return num_from_Ada;
@}
@end smallexample

@smallexample @c ada
--  my_main.adb
procedure My_Main is

   --  Declare then export an Integer entity called num_from_Ada
   My_Num : Integer := 10;
   pragma Export (C, My_Num, "num_from_Ada");

   --  Declare an Ada function spec for Get_Num, then use
   --  C function get_num for the implementation.
   function Get_Num return Integer;
   pragma Import (C, Get_Num, "get_num");

   --  Declare an Ada procedure spec for Print_Num, then use
   --  C function print_num for the implementation.
   procedure Print_Num (Num : Integer);
   pragma Import (C, Print_Num, "print_num";

begin
   Print_Num (Get_Num);
end My_Main;
@end smallexample

@enumerate
@item
To build this example, first compile the foreign language files to
generate object files:
@smallexample
gcc -c file1.c
gcc -c file2.c
@end smallexample

@item
Then, compile the Ada units to produce a set of object files and ALI
files:
@smallexample
gnatmake -c my_main.adb
@end smallexample

@item
Run the Ada binder on the Ada main program:
@smallexample
gnatbind my_main.ali
@end smallexample

@item
Link the Ada main program, the Ada objects and the other language
objects:
@smallexample
gnatlink my_main.ali file1.o file2.o
@end smallexample
@end enumerate

The last three steps can be grouped in a single command:
@smallexample
gnatmake my_main.adb -largs file1.o file2.o
@end smallexample

@cindex Binder output file
@noindent
If the main program is in a language other than Ada, then you may have
more than one entry point into the Ada subsystem. You must use a special
binder option to generate callable routines that initialize and
finalize the Ada units (@pxref{Binding with Non-Ada Main Programs}).
Calls to the initialization and finalization routines must be inserted
in the main program, or some other appropriate point in the code. The
call to initialize the Ada units must occur before the first Ada
subprogram is called, and the call to finalize the Ada units must occur
after the last Ada subprogram returns. The binder will place the
initialization and finalization subprograms into the
@file{b~@var{xxx}.adb} file where they can be accessed by your C
sources.  To illustrate, we have the following example:

@smallexample
/* main.c */
extern void adainit (void);
extern void adafinal (void);
extern int add (int, int);
extern int sub (int, int);

int main (int argc, char *argv[])
@{
  int a = 21, b = 7;

  adainit();

  /* Should print "21 + 7 = 28" */
  printf ("%d + %d = %d\n", a, b, add (a, b));
  /* Should print "21 - 7 = 14" */
  printf ("%d - %d = %d\n", a, b, sub (a, b));

  adafinal();
@}
@end smallexample

@smallexample @c ada
--  unit1.ads
package Unit1 is
   function Add (A, B : Integer) return Integer;
   pragma Export (C, Add, "add");
end Unit1;

--  unit1.adb
package body Unit1 is
   function Add (A, B : Integer) return Integer is
   begin
      return A + B;
   end Add;
end Unit1;

--  unit2.ads
package Unit2 is
   function Sub (A, B : Integer) return Integer;
   pragma Export (C, Sub, "sub");
end Unit2;

--  unit2.adb
package body Unit2 is
   function Sub (A, B : Integer) return Integer is
   begin
      return A - B;
   end Sub;
end Unit2;
@end smallexample

@enumerate
@item
The build procedure for this application is similar to the last
example's.  First, compile the foreign language files to generate object
files:
@smallexample
gcc -c main.c
@end smallexample

@item
Next, compile the Ada units to produce a set of object files and ALI
files:
@smallexample
gnatmake -c unit1.adb
gnatmake -c unit2.adb
@end smallexample

@item
Run the Ada binder on every generated ALI file.  Make sure to use the
@option{-n} option to specify a foreign main program:
@smallexample
gnatbind -n unit1.ali unit2.ali
@end smallexample

@item
Link the Ada main program, the Ada objects and the foreign language
objects. You need only list the last ALI file here:
@smallexample
gnatlink unit2.ali main.o -o exec_file
@end smallexample

This procedure yields a binary executable called @file{exec_file}.
@end enumerate

@noindent
Depending on the circumstances (for example when your non-Ada main object
does not provide symbol @code{main}), you may also need to instruct the
GNAT linker not to include the standard startup objects by passing the
@option{-nostartfiles} switch to @command{gnatlink}.

@node Calling Conventions
@subsection Calling Conventions
@cindex Foreign Languages
@cindex Calling Conventions
GNAT follows standard calling sequence conventions and will thus interface
to any other language that also follows these conventions. The following
Convention identifiers are recognized by GNAT:

@table @code
@cindex Interfacing to Ada
@cindex Other Ada compilers
@cindex Convention Ada
@item Ada
This indicates that the standard Ada calling sequence will be
used and all Ada data items may be passed without any limitations in the
case where GNAT is used to generate both the caller and callee. It is also
possible to mix GNAT generated code and code generated by another Ada
compiler. In this case, the data types should be restricted to simple
cases, including primitive types. Whether complex data types can be passed
depends on the situation. Probably it is safe to pass simple arrays, such
as arrays of integers or floats. Records may or may not work, depending
on whether both compilers lay them out identically. Complex structures
involving variant records, access parameters, tasks, or protected types,
are unlikely to be able to be passed.

Note that in the case of GNAT running
on a platform that supports HP Ada 83, a higher degree of compatibility
can be guaranteed, and in particular records are laid out in an identical
manner in the two compilers. Note also that if output from two different
compilers is mixed, the program is responsible for dealing with elaboration
issues. Probably the safest approach is to write the main program in the
version of Ada other than GNAT, so that it takes care of its own elaboration
requirements, and then call the GNAT-generated adainit procedure to ensure
elaboration of the GNAT components. Consult the documentation of the other
Ada compiler for further details on elaboration.

However, it is not possible to mix the tasking run time of GNAT and
HP Ada 83, All the tasking operations must either be entirely within
GNAT compiled sections of the program, or entirely within HP Ada 83
compiled sections of the program.

@cindex Interfacing to Assembly
@cindex Convention Assembler
@item Assembler
Specifies assembler as the convention. In practice this has the
same effect as convention Ada (but is not equivalent in the sense of being
considered the same convention).

@cindex Convention Asm
@findex Asm
@item Asm
Equivalent to Assembler.

@cindex Interfacing to COBOL
@cindex Convention COBOL
@findex COBOL
@item COBOL
Data will be passed according to the conventions described
in section B.4 of the Ada Reference Manual.

@findex C
@cindex Interfacing to C
@cindex Convention C
@item C
Data will be passed according to the conventions described
in section B.3 of the Ada Reference Manual.

A note on interfacing to a C ``varargs'' function:
@findex C varargs function
@cindex Interfacing to C varargs function
@cindex varargs function interfaces

@itemize @bullet
@item
In C, @code{varargs} allows a function to take a variable number of
arguments. There is no direct equivalent in this to Ada. One
approach that can be used is to create a C wrapper for each
different profile and then interface to this C wrapper. For
example, to print an @code{int} value using @code{printf},
create a C function @code{printfi} that takes two arguments, a
pointer to a string and an int, and calls @code{printf}.
Then in the Ada program, use pragma @code{Import} to
interface to @code{printfi}.

@item
It may work on some platforms to directly interface to
a @code{varargs} function by providing a specific Ada profile
for a particular call. However, this does not work on
all platforms, since there is no guarantee that the
calling sequence for a two argument normal C function
is the same as for calling a @code{varargs} C function with
the same two arguments.
@end itemize

@cindex Convention Default
@findex Default
@item Default
Equivalent to C.

@cindex Convention External
@findex External
@item External
Equivalent to C.

@findex C++
@cindex Interfacing to C++
@cindex Convention C++
@item C_Plus_Plus (or CPP)
This stands for C++. For most purposes this is identical to C.
See the separate description of the specialized GNAT pragmas relating to
C++ interfacing for further details.

@findex Fortran
@cindex Interfacing to Fortran
@cindex Convention Fortran
@item Fortran
Data will be passed according to the conventions described
in section B.5 of the Ada Reference Manual.

@item Intrinsic
This applies to an intrinsic operation, as defined in the Ada
Reference Manual. If a pragma Import (Intrinsic) applies to a subprogram,
this means that the body of the subprogram is provided by the compiler itself,
usually by means of an efficient code sequence, and that the user does not
supply an explicit body for it. In an application program, the pragma may
be applied to the following sets of names:

@itemize @bullet
@item
Rotate_Left, Rotate_Right, Shift_Left, Shift_Right,
Shift_Right_Arithmetic.  The corresponding subprogram declaration must have
two formal parameters. The
first one must be a signed integer type or a modular type with a binary
modulus, and the second parameter must be of type Natural.
The return type must be the same as the type of the first argument. The size
of this type can only be 8, 16, 32, or 64.

@item
Binary arithmetic operators: ``+'', ``-'', ``*'', ``/''
The corresponding operator declaration must have parameters and result type
that have the same root numeric type (for example, all three are long_float
types). This simplifies the definition of operations that use type checking
to perform dimensional checks:

@smallexample @c ada
@b{type} Distance @b{is} @b{new} Long_Float;
@b{type} Time     @b{is} @b{new} Long_Float;
@b{type} Velocity @b{is} @b{new} Long_Float;
@b{function} "/" (D : Distance; T : Time)
  @b{return} Velocity;
@b{pragma} Import (Intrinsic, "/");
@end smallexample

@noindent
This common idiom is often programmed with a generic definition and an
explicit body. The pragma makes it simpler to introduce such declarations.
It incurs no overhead in compilation time or code size, because it is
implemented as a single machine instruction.

@item
General subprogram entities, to bind an Ada subprogram declaration to
a compiler builtin by name with back-ends where such interfaces are
available. A typical example is the set of ``__builtin'' functions
exposed by the GCC back-end, as in the following example:

@smallexample @c ada
   @b{function} builtin_sqrt (F : Float) @b{return} Float;
   @b{pragma} Import (Intrinsic, builtin_sqrt, "__builtin_sqrtf");
@end smallexample

Most of the GCC builtins are accessible this way, and as for other
import conventions (e.g. C), it is the user's responsibility to ensure
that the Ada subprogram profile matches the underlying builtin
expectations.
@end itemize

@noindent

@findex Stdcall
@cindex Convention Stdcall
@item Stdcall
This is relevant only to Windows XP/2000/NT implementations of GNAT,
and specifies that the @code{Stdcall} calling sequence will be used,
as defined by the NT API. Nevertheless, to ease building
cross-platform bindings this convention will be handled as a @code{C} calling
convention on non-Windows platforms.

@findex DLL
@cindex Convention DLL
@item DLL
This is equivalent to @code{Stdcall}.

@findex Win32
@cindex Convention Win32
@item Win32
This is equivalent to @code{Stdcall}.

@findex Stubbed
@cindex Convention Stubbed
@item Stubbed
This is a special convention that indicates that the compiler
should provide a stub body that raises @code{Program_Error}.
@end table

@noindent
GNAT additionally provides a useful pragma @code{Convention_Identifier}
that can be used to parameterize conventions and allow additional synonyms
to be specified. For example if you have legacy code in which the convention
identifier Fortran77 was used for Fortran, you can use the configuration
pragma:

@smallexample @c ada
@b{pragma} Convention_Identifier (Fortran77, Fortran);
@end smallexample

@noindent
And from now on the identifier Fortran77 may be used as a convention
identifier (for example in an @code{Import} pragma) with the same
meaning as Fortran.

@node Building Mixed Ada & C++ Programs
@section Building Mixed Ada and C++ Programs

@noindent
A programmer inexperienced with mixed-language development may find that
building an application containing both Ada and C++ code can be a
challenge.  This section gives a few
hints that should make this task easier. The first section addresses
the differences between interfacing with C and interfacing with C++.
The second section
looks into the delicate problem of linking the complete application from
its Ada and C++ parts. The last section gives some hints on how the GNAT
run-time library can be adapted in order to allow inter-language dispatching
with a new C++ compiler.

@menu
* Interfacing to C++::
* Linking a Mixed C++ & Ada Program::
* A Simple Example::
* Interfacing with C++ constructors::
* Interfacing with C++ at the Class Level::
@end menu

@node Interfacing to C++
@subsection Interfacing to C++

@noindent
GNAT supports interfacing with the G++ compiler (or any C++ compiler
generating code that is compatible with the G++ Application Binary
Interface ---see http://www.codesourcery.com/archives/cxx-abi).

@noindent
Interfacing can be done at 3 levels: simple data, subprograms, and
classes. In the first two cases, GNAT offers a specific @code{Convention
C_Plus_Plus} (or @code{CPP}) that behaves exactly like @code{Convention C}.
Usually, C++ mangles the names of subprograms. To generate proper mangled
names automatically, see @ref{Generating Ada Bindings for C and C++ headers}).
This problem can also be addressed manually in two ways:

@itemize @bullet
@item
by modifying the C++ code in order to force a C convention using
the @code{extern "C"} syntax.

@item
by figuring out the mangled name (using e.g. @command{nm}) and using it as the
Link_Name argument of the pragma import.
@end itemize

@noindent
Interfacing at the class level can be achieved by using the GNAT specific
pragmas such as @code{CPP_Constructor}.  @xref{Interfacing to C++,,,
gnat_rm, GNAT Reference Manual}, for additional information.

@node Linking a Mixed C++ & Ada Program
@subsection Linking a Mixed C++ & Ada Program

@noindent
Usually the linker of the C++ development system must be used to link
mixed applications because most C++ systems will resolve elaboration
issues (such as calling constructors on global class instances)
transparently during the link phase. GNAT has been adapted to ease the
use of a foreign linker for the last phase. Three cases can be
considered:
@enumerate

@item
Using GNAT and G++ (GNU C++ compiler) from the same GCC installation:
The C++ linker can simply be called by using the C++ specific driver
called @code{g++}.

Note that if the C++ code uses inline functions, you will need to
compile your C++ code with the @code{-fkeep-inline-functions} switch in
order to provide an existing function implementation that the Ada code can
link with.

@smallexample
$ g++ -c -fkeep-inline-functions file1.C
$ g++ -c -fkeep-inline-functions file2.C
$ gnatmake ada_unit -largs file1.o file2.o --LINK=g++
@end smallexample

@item
Using GNAT and G++ from two different GCC installations: If both
compilers are on the @env{PATH}, the previous method may be used. It is
important to note that environment variables such as
@env{C_INCLUDE_PATH}, @env{GCC_EXEC_PREFIX}, @env{BINUTILS_ROOT}, and
@env{GCC_ROOT} will affect both compilers
at the same time and may make one of the two compilers operate
improperly if set during invocation of the wrong compiler.  It is also
very important that the linker uses the proper @file{libgcc.a} GCC
library -- that is, the one from the C++ compiler installation. The
implicit link command as suggested in the @command{gnatmake} command
from the former example can be replaced by an explicit link command with
the full-verbosity option in order to verify which library is used:
@smallexample
$ gnatbind ada_unit
$ gnatlink -v -v ada_unit file1.o file2.o --LINK=c++
@end smallexample
If there is a problem due to interfering environment variables, it can
be worked around by using an intermediate script. The following example
shows the proper script to use when GNAT has not been installed at its
default location and g++ has been installed at its default location:

@smallexample
$ cat ./my_script
#!/bin/sh
unset BINUTILS_ROOT
unset GCC_ROOT
c++ $*
$ gnatlink -v -v ada_unit file1.o file2.o --LINK=./my_script
@end smallexample

@item
Using a non-GNU C++ compiler: The commands previously described can be
used to insure that the C++ linker is used. Nonetheless, you need to add
a few more parameters to the link command line, depending on the exception
mechanism used.

If the @code{setjmp/longjmp} exception mechanism is used, only the paths
to the libgcc libraries are required:

@smallexample
$ cat ./my_script
#!/bin/sh
CC $* `gcc -print-file-name=libgcc.a` `gcc -print-file-name=libgcc_eh.a`
$ gnatlink ada_unit file1.o file2.o --LINK=./my_script
@end smallexample

Where CC is the name of the non-GNU C++ compiler.

If the @code{zero cost} exception mechanism is used, and the platform
supports automatic registration of exception tables (e.g.@: Solaris),
paths to more objects are required:

@smallexample
$ cat ./my_script
#!/bin/sh
CC `gcc -print-file-name=crtbegin.o` $* \
`gcc -print-file-name=libgcc.a` `gcc -print-file-name=libgcc_eh.a` \
`gcc -print-file-name=crtend.o`
$ gnatlink ada_unit file1.o file2.o --LINK=./my_script
@end smallexample

If the @code{zero cost} exception mechanism is used, and the platform
doesn't support automatic registration of exception tables (e.g.@: HP-UX
or AIX), the simple approach described above will not work and
a pre-linking phase using GNAT will be necessary.

@end enumerate

Another alternative is to use the @command{gprbuild} multi-language builder
which has a large knowledge base and knows how to link Ada and C++ code
together automatically in most cases.

@node A Simple Example
@subsection  A Simple Example
@noindent
The following example, provided as part of the GNAT examples, shows how
to achieve procedural interfacing between Ada and C++ in both
directions. The C++ class A has two methods. The first method is exported
to Ada by the means of an extern C wrapper function. The second method
calls an Ada subprogram. On the Ada side, The C++ calls are modelled by
a limited record with a layout comparable to the C++ class. The Ada
subprogram, in turn, calls the C++ method. So, starting from the C++
main program, the process passes back and forth between the two
languages.

@noindent
Here are the compilation commands:
@smallexample
$ gnatmake -c simple_cpp_interface
$ g++ -c cpp_main.C
$ g++ -c ex7.C
$ gnatbind -n simple_cpp_interface
$ gnatlink simple_cpp_interface -o cpp_main --LINK=g++
      -lstdc++ ex7.o cpp_main.o
@end smallexample

@noindent
Here are the corresponding sources:
@smallexample

//cpp_main.C

#include "ex7.h"

extern "C" @{
  void adainit (void);
  void adafinal (void);
  void method1 (A *t);
@}

void method1 (A *t)
@{
  t->method1 ();
@}

int main ()
@{
  A obj;
  adainit ();
  obj.method2 (3030);
  adafinal ();
@}

//ex7.h

class Origin @{
 public:
  int o_value;
@};
class A : public Origin @{
 public:
  void method1 (void);
  void method2 (int v);
  A();
  int   a_value;
@};

//ex7.C

#include "ex7.h"
#include <stdio.h>

extern "C" @{ void ada_method2 (A *t, int v);@}

void A::method1 (void)
@{
  a_value = 2020;
  printf ("in A::method1, a_value = %d \n",a_value);

@}

void A::method2 (int v)
@{
   ada_method2 (this, v);
   printf ("in A::method2, a_value = %d \n",a_value);

@}

A::A(void)
@{
   a_value = 1010;
  printf ("in A::A, a_value = %d \n",a_value);
@}
@end smallexample

@smallexample @c ada
--@i{ Ada sources}
@b{package} @b{body} Simple_Cpp_Interface @b{is}

   @b{procedure} Ada_Method2 (This : @b{in} @b{out} A; V : Integer) @b{is}
   @b{begin}
      Method1 (This);
      This.A_Value := V;
   @b{end} Ada_Method2;

@b{end} Simple_Cpp_Interface;

@b{with} System;
@b{package} Simple_Cpp_Interface @b{is}
   @b{type} A @b{is} @b{limited}
      @b{record}
         Vptr    : System.Address;
         O_Value : Integer;
         A_Value : Integer;
      @b{end} @b{record};
   @b{pragma} Convention (C, A);

   @b{procedure} Method1 (This : @b{in} @b{out} A);
   @b{pragma} Import (C, Method1);

   @b{procedure} Ada_Method2 (This : @b{in} @b{out} A; V : Integer);
   @b{pragma} Export (C, Ada_Method2);

@b{end} Simple_Cpp_Interface;
@end smallexample

@node Interfacing with C++ constructors
@subsection Interfacing with C++ constructors
@noindent

In order to interface with C++ constructors GNAT provides the
@code{pragma CPP_Constructor} (@xref{Interfacing to C++,,,
gnat_rm, GNAT Reference Manual}, for additional information).
In this section we present some common uses of C++ constructors
in mixed-languages programs in GNAT.

Let us assume that we need to interface with the following
C++ class:

@smallexample
@b{class} Root @{
@b{public}:
  int  a_value;
  int  b_value;
  @b{virtual} int Get_Value ();
  Root();              // Default constructor
  Root(int v);         // 1st non-default constructor
  Root(int v, int w);  // 2nd non-default constructor
@};
@end smallexample

For this purpose we can write the following package spec (further
information on how to build this spec is available in
@ref{Interfacing with C++ at the Class Level} and
@ref{Generating Ada Bindings for C and C++ headers}).

@smallexample @c ada
@b{with} Interfaces.C; @b{use} Interfaces.C;
@b{package} Pkg_Root @b{is}
  @b{type} Root @b{is} @b{tagged} @b{limited} @b{record}
     A_Value : int;
     B_Value : int;
  @b{end} @b{record};
  @b{pragma} Import (CPP, Root);

  @b{function} Get_Value (Obj : Root) @b{return} int;
  @b{pragma} Import (CPP, Get_Value);

  @b{function} Constructor @b{return} Root;
  @b{pragma} Cpp_Constructor (Constructor, "_ZN4RootC1Ev");

  @b{function} Constructor (v : Integer) @b{return} Root;
  @b{pragma} Cpp_Constructor (Constructor, "_ZN4RootC1Ei");

  @b{function} Constructor (v, w : Integer) @b{return} Root;
  @b{pragma} Cpp_Constructor (Constructor, "_ZN4RootC1Eii");
@b{end} Pkg_Root;
@end smallexample

On the Ada side the constructor is represented by a function (whose
name is arbitrary) that returns the classwide type corresponding to
the imported C++ class. Although the constructor is described as a
function, it is typically a procedure with an extra implicit argument
(the object being initialized) at the implementation level. GNAT
issues the appropriate call, whatever it is, to get the object
properly initialized.

Constructors can only appear in the following contexts:

@itemize @bullet
@item
On the right side of an initialization of an object of type @var{T}.
@item
On the right side of an initialization of a record component of type @var{T}.
@item
In an Ada 2005 limited aggregate.
@item
In an Ada 2005 nested limited aggregate.
@item
In an Ada 2005 limited aggregate that initializes an object built in
place by an extended return statement.
@end itemize

@noindent
In a declaration of an object whose type is a class imported from C++,
either the default C++ constructor is implicitly called by GNAT, or
else the required C++ constructor must be explicitly called in the
expression that initializes the object. For example:

@smallexample @c ada
  Obj1 : Root;
  Obj2 : Root := Constructor;
  Obj3 : Root := Constructor (v => 10);
  Obj4 : Root := Constructor (30, 40);
@end smallexample

The first two declarations are equivalent: in both cases the default C++
constructor is invoked (in the former case the call to the constructor is
implicit, and in the latter case the call is explicit in the object
declaration). @code{Obj3} is initialized by the C++ non-default constructor
that takes an integer argument, and @code{Obj4} is initialized by the
non-default C++ constructor that takes two integers.

Let us derive the imported C++ class in the Ada side. For example:

@smallexample @c ada
  @b{type} DT @b{is} @b{new} Root @b{with} @b{record}
     C_Value : Natural := 2009;
  @b{end} @b{record};
@end smallexample

In this case the components DT inherited from the C++ side must be
initialized by a C++ constructor, and the additional Ada components
of type DT are initialized by GNAT. The initialization of such an
object is done either by default, or by means of a function returning
an aggregate of type DT, or by means of an extension aggregate.

@smallexample @c ada
  Obj5 : DT;
  Obj6 : DT := Function_Returning_DT (50);
  Obj7 : DT := (Constructor (30,40) @b{with} C_Value => 50);
@end smallexample

The declaration of @code{Obj5} invokes the default constructors: the
C++ default constructor of the parent type takes care of the initialization
of the components inherited from Root, and GNAT takes care of the default
initialization of the additional Ada components of type DT (that is,
@code{C_Value} is initialized to value 2009). The order of invocation of
the constructors is consistent with the order of elaboration required by
Ada and C++. That is, the constructor of the parent type is always called
before the constructor of the derived type.

Let us now consider a record that has components whose type is imported
from C++. For example:

@smallexample @c ada
  @b{type} Rec1 @b{is} @b{limited} @b{record}
     Data1 : Root := Constructor (10);
     Value : Natural := 1000;
  @b{end} @b{record};

  @b{type} Rec2 (D : Integer := 20) @b{is} @b{limited} @b{record}
     Rec   : Rec1;
     Data2 : Root := Constructor (D, 30);
  @b{end} @b{record};
@end smallexample

The initialization of an object of type @code{Rec2} will call the
non-default C++ constructors specified for the imported components.
For example:

@smallexample @c ada
  Obj8 : Rec2 (40);
@end smallexample

Using Ada 2005 we can use limited aggregates to initialize an object
invoking C++ constructors that differ from those specified in the type
declarations. For example:

@smallexample @c ada
  Obj9 : Rec2 := (Rec => (Data1 => Constructor (15, 16),
                          @b{others} => <>),
                  @b{others} => <>);
@end smallexample

The above declaration uses an Ada 2005 limited aggregate to
initialize @code{Obj9}, and the C++ constructor that has two integer
arguments is invoked to initialize the @code{Data1} component instead
of the constructor specified in the declaration of type @code{Rec1}. In
Ada 2005 the box in the aggregate indicates that unspecified components
are initialized using the expression (if any) available in the component
declaration. That is, in this case discriminant @code{D} is initialized
to value @code{20}, @code{Value} is initialized to value 1000, and the
non-default C++ constructor that handles two integers takes care of
initializing component @code{Data2} with values @code{20,30}.

In Ada 2005 we can use the extended return statement to build the Ada
equivalent to C++ non-default constructors. For example:

@smallexample @c ada
  @b{function} Constructor (V : Integer) @b{return} Rec2 @b{is}
  @b{begin}
     @b{return} Obj : Rec2 := (Rec => (Data1  => Constructor (V, 20),
                                   @b{others} => <>),
                           @b{others} => <>) @b{do}
        --@i{  Further actions required for construction of}
        --@i{  objects of type Rec2}
        ...
     @b{end} @b{record};
  @b{end} Constructor;
@end smallexample

In this example the extended return statement construct is used to
build in place the returned object whose components are initialized
by means of a limited aggregate. Any further action associated with
the constructor can be placed inside the construct.

@node Interfacing with C++ at the Class Level
@subsection Interfacing with C++ at the Class Level
@noindent
In this section we demonstrate the GNAT features for interfacing with
C++ by means of an example making use of Ada 2005 abstract interface
types. This example consists of a classification of animals; classes
have been used to model our main classification of animals, and
interfaces provide support for the management of secondary
classifications. We first demonstrate a case in which the types and
constructors are defined on the C++ side and imported from the Ada
side, and latter the reverse case.

The root of our derivation will be the @code{Animal} class, with a
single private attribute (the @code{Age} of the animal), a constructor,
and two public primitives to set and get the value of this attribute.

@smallexample
@b{class} Animal @{
 @b{public}:
   @b{virtual} void Set_Age (int New_Age);
   @b{virtual} int Age ();
   Animal() @{Age_Count = 0;@};
 @b{private}:
   int Age_Count;
@};
@end smallexample

Abstract interface types are defined in C++ by means of classes with pure
virtual functions and no data members. In our example we will use two
interfaces that provide support for the common management of @code{Carnivore}
and @code{Domestic} animals:

@smallexample
@b{class} Carnivore @{
@b{public}:
   @b{virtual} int Number_Of_Teeth () = 0;
@};

@b{class} Domestic @{
@b{public}:
   @b{virtual void} Set_Owner (char* Name) = 0;
@};
@end smallexample

Using these declarations, we can now say that a @code{Dog} is an animal that is
both Carnivore and Domestic, that is:

@smallexample
@b{class} Dog : Animal, Carnivore, Domestic @{
 @b{public}:
   @b{virtual} int  Number_Of_Teeth ();
   @b{virtual} void Set_Owner (char* Name);

   Dog(); // Constructor
 @b{private}:
   int  Tooth_Count;
   char *Owner;
@};
@end smallexample

In the following examples we will assume that the previous declarations are
located in a file named @code{animals.h}. The following package demonstrates
how to import these C++ declarations from the Ada side:

@smallexample @c ada
@b{with} Interfaces.C.Strings; @b{use} Interfaces.C.Strings;
@b{package} Animals @b{is}
  @b{type} Carnivore @b{is} @b{limited} interface;
  @b{pragma} Convention (C_Plus_Plus, Carnivore);
  @b{function} Number_Of_Teeth (X : Carnivore)
     @b{return} Natural @b{is} @b{abstract};

  @b{type} Domestic @b{is} @b{limited} interface;
  @b{pragma} Convention (C_Plus_Plus, Domestic);
  @b{procedure} Set_Owner
    (X    : @b{in} @b{out} Domestic;
     Name : Chars_Ptr) @b{is} @b{abstract};

  @b{type} Animal @b{is} @b{tagged} @b{limited} @b{record}
    Age : Natural;
  @b{end} @b{record};
  @b{pragma} Import (C_Plus_Plus, Animal);

  @b{procedure} Set_Age (X : @b{in} @b{out} Animal; Age : Integer);
  @b{pragma} Import (C_Plus_Plus, Set_Age);

  @b{function} Age (X : Animal) @b{return} Integer;
  @b{pragma} Import (C_Plus_Plus, Age);

  @b{function} New_Animal @b{return} Animal;
  @b{pragma} CPP_Constructor (New_Animal);
  @b{pragma} Import (CPP, New_Animal, "_ZN6AnimalC1Ev");

  @b{type} Dog @b{is} @b{new} Animal @b{and} Carnivore @b{and} Domestic @b{with} @b{record}
    Tooth_Count : Natural;
    Owner       : String (1 .. 30);
  @b{end} @b{record};
  @b{pragma} Import (C_Plus_Plus, Dog);

  @b{function} Number_Of_Teeth (A : Dog) @b{return} Natural;
  @b{pragma} Import (C_Plus_Plus, Number_Of_Teeth);

  @b{procedure} Set_Owner (A : @b{in} @b{out} Dog; Name : Chars_Ptr);
  @b{pragma} Import (C_Plus_Plus, Set_Owner);

  @b{function} New_Dog @b{return} Dog;
  @b{pragma} CPP_Constructor (New_Dog);
  @b{pragma} Import (CPP, New_Dog, "_ZN3DogC2Ev");
@b{end} Animals;
@end smallexample

Thanks to the compatibility between GNAT run-time structures and the C++ ABI,
interfacing with these C++ classes is easy. The only requirement is that all
the primitives and components must be declared exactly in the same order in
the two languages.

Regarding the abstract interfaces, we must indicate to the GNAT compiler by
means of a @code{pragma Convention (C_Plus_Plus)}, the convention used to pass
the arguments to the called primitives will be the same as for C++. For the
imported classes we use @code{pragma Import} with convention @code{C_Plus_Plus}
to indicate that they have been defined on the C++ side; this is required
because the dispatch table associated with these tagged types will be built
in the C++ side and therefore will not contain the predefined Ada primitives
which Ada would otherwise expect.

As the reader can see there is no need to indicate the C++ mangled names
associated with each subprogram because it is assumed that all the calls to
these primitives will be dispatching calls. The only exception is the
constructor, which must be registered with the compiler by means of
@code{pragma CPP_Constructor} and needs to provide its associated C++
mangled name because the Ada compiler generates direct calls to it.

With the above packages we can now declare objects of type Dog on the Ada side
and dispatch calls to the corresponding subprograms on the C++ side. We can
also extend the tagged type Dog with further fields and primitives, and
override some of its C++ primitives on the Ada side. For example, here we have
a type derivation defined on the Ada side that inherits all the dispatching
primitives of the ancestor from the C++ side.

@smallexample
@b{with} Animals; @b{use} Animals;
@b{package} Vaccinated_Animals @b{is}
  @b{type} Vaccinated_Dog @b{is new} Dog @b{with null record};
  @b{function} Vaccination_Expired (A : Vaccinated_Dog) @b{return} Boolean;
@b{end} Vaccinated_Animals;
@end smallexample

It is important to note that, because of the ABI compatibility, the programmer
does not need to add any further information to indicate either the object
layout or the dispatch table entry associated with each dispatching operation.

Now let us define all the types and constructors on the Ada side and export
them to C++, using the same hierarchy of our previous example:

@smallexample @c ada
@b{with} Interfaces.C.Strings;
@b{use} Interfaces.C.Strings;
@b{package} Animals @b{is}
  @b{type} Carnivore @b{is} @b{limited} interface;
  @b{pragma} Convention (C_Plus_Plus, Carnivore);
  @b{function} Number_Of_Teeth (X : Carnivore)
     @b{return} Natural @b{is} @b{abstract};

  @b{type} Domestic @b{is} @b{limited} interface;
  @b{pragma} Convention (C_Plus_Plus, Domestic);
  @b{procedure} Set_Owner
    (X    : @b{in} @b{out} Domestic;
     Name : Chars_Ptr) @b{is} @b{abstract};

  @b{type} Animal @b{is} @b{tagged} @b{record}
    Age : Natural;
  @b{end} @b{record};
  @b{pragma} Convention (C_Plus_Plus, Animal);

  @b{procedure} Set_Age (X : @b{in} @b{out} Animal; Age : Integer);
  @b{pragma} Export (C_Plus_Plus, Set_Age);

  @b{function} Age (X : Animal) @b{return} Integer;
  @b{pragma} Export (C_Plus_Plus, Age);

  @b{function} New_Animal @b{return} Animal'Class;
  @b{pragma} Export (C_Plus_Plus, New_Animal);

  @b{type} Dog @b{is} @b{new} Animal @b{and} Carnivore @b{and} Domestic @b{with} @b{record}
    Tooth_Count : Natural;
    Owner       : String (1 .. 30);
  @b{end} @b{record};
  @b{pragma} Convention (C_Plus_Plus, Dog);

  @b{function} Number_Of_Teeth (A : Dog) @b{return} Natural;
  @b{pragma} Export (C_Plus_Plus, Number_Of_Teeth);

  @b{procedure} Set_Owner (A : @b{in} @b{out} Dog; Name : Chars_Ptr);
  @b{pragma} Export (C_Plus_Plus, Set_Owner);

  @b{function} New_Dog @b{return} Dog'Class;
  @b{pragma} Export (C_Plus_Plus, New_Dog);
@b{end} Animals;
@end smallexample

Compared with our previous example the only differences are the use of
@code{pragma Convention} (instead of @code{pragma Import}), and the use of
@code{pragma Export} to indicate to the GNAT compiler that the primitives will
be available to C++. Thanks to the ABI compatibility, on the C++ side there is
nothing else to be done; as explained above, the only requirement is that all
the primitives and components are declared in exactly the same order.

For completeness, let us see a brief C++ main program that uses the
declarations available in @code{animals.h} (presented in our first example) to
import and use the declarations from the Ada side, properly initializing and
finalizing the Ada run-time system along the way:

@smallexample
@b{#include} "animals.h"
@b{#include} <iostream>
@b{using namespace} std;

void Check_Carnivore (Carnivore *obj) @{@dots{}@}
void Check_Domestic (Domestic *obj)   @{@dots{}@}
void Check_Animal (Animal *obj)       @{@dots{}@}
void Check_Dog (Dog *obj)             @{@dots{}@}

@b{extern} "C" @{
  void adainit (void);
  void adafinal (void);
  Dog* new_dog ();
@}

void test ()
@{
  Dog *obj = new_dog();  // Ada constructor
  Check_Carnivore (obj); // Check secondary DT
  Check_Domestic (obj);  // Check secondary DT
  Check_Animal (obj);    // Check primary DT
  Check_Dog (obj);       // Check primary DT
@}

int main ()
@{
  adainit ();  test();  adafinal ();
  return 0;
@}
@end smallexample

@node Comparison between GNAT and C/C++ Compilation Models
@section Comparison between GNAT and C/C++ Compilation Models

@noindent
The GNAT model of compilation is close to the C and C++ models. You can
think of Ada specs as corresponding to header files in C. As in C, you
don't need to compile specs; they are compiled when they are used. The
Ada @code{with} is similar in effect to the @code{#include} of a C
header.

One notable difference is that, in Ada, you may compile specs separately
to check them for semantic and syntactic accuracy. This is not always
possible with C headers because they are fragments of programs that have
less specific syntactic or semantic rules.

The other major difference is the requirement for running the binder,
which performs two important functions. First, it checks for
consistency. In C or C++, the only defense against assembling
inconsistent programs lies outside the compiler, in a makefile, for
example. The binder satisfies the Ada requirement that it be impossible
to construct an inconsistent program when the compiler is used in normal
mode.

@cindex Elaboration order control
The other important function of the binder is to deal with elaboration
issues. There are also elaboration issues in C++ that are handled
automatically. This automatic handling has the advantage of being
simpler to use, but the C++ programmer has no control over elaboration.
Where @code{gnatbind} might complain there was no valid order of
elaboration, a C++ compiler would simply construct a program that
malfunctioned at run time.

@node Comparison between GNAT and Conventional Ada Library Models
@section Comparison between GNAT and Conventional Ada Library Models

@noindent
This section is intended for Ada programmers who have
used an Ada compiler implementing the traditional Ada library
model, as described in the Ada Reference Manual.

@cindex GNAT library
In GNAT, there is no ``library'' in the normal sense. Instead, the set of
source files themselves acts as the library. Compiling Ada programs does
not generate any centralized information, but rather an object file and
a ALI file, which are of interest only to the binder and linker.
In a traditional system, the compiler reads information not only from
the source file being compiled, but also from the centralized library.
This means that the effect of a compilation depends on what has been
previously compiled. In particular:

@itemize @bullet
@item
When a unit is @code{with}'ed, the unit seen by the compiler corresponds
to the version of the unit most recently compiled into the library.

@item
Inlining is effective only if the necessary body has already been
compiled into the library.

@item
Compiling a unit may obsolete other units in the library.
@end itemize

@noindent
In GNAT, compiling one unit never affects the compilation of any other
units because the compiler reads only source files. Only changes to source
files can affect the results of a compilation. In particular:

@itemize @bullet
@item
When a unit is @code{with}'ed, the unit seen by the compiler corresponds
to the source version of the unit that is currently accessible to the
compiler.

@item
@cindex Inlining
Inlining requires the appropriate source files for the package or
subprogram bodies to be available to the compiler. Inlining is always
effective, independent of the order in which units are complied.

@item
Compiling a unit never affects any other compilations. The editing of
sources may cause previous compilations to be out of date if they
depended on the source file being modified.
@end itemize

@noindent
The most important result of these differences is that order of compilation
is never significant in GNAT. There is no situation in which one is
required to do one compilation before another. What shows up as order of
compilation requirements in the traditional Ada library becomes, in
GNAT, simple source dependencies; in other words, there is only a set
of rules saying what source files must be present when a file is
compiled.


@c *************************
@node Compiling with gcc
@chapter Compiling with @command{gcc}

@noindent
This chapter discusses how to compile Ada programs using the @command{gcc}
command. It also describes the set of switches
that can be used to control the behavior of the compiler.
@menu
* Compiling Programs::
* Switches for gcc::
* Search Paths and the Run-Time Library (RTL)::
* Order of Compilation Issues::
* Examples::
@end menu

@node Compiling Programs
@section Compiling Programs

@noindent
The first step in creating an executable program is to compile the units
of the program using the @command{gcc} command. You must compile the
following files:

@itemize @bullet
@item
the body file (@file{.adb}) for a library level subprogram or generic
subprogram

@item
the spec file (@file{.ads}) for a library level package or generic
package that has no body

@item
the body file (@file{.adb}) for a library level package
or generic package that has a body

@end itemize

@noindent
You need @emph{not} compile the following files

@itemize @bullet

@item
the spec of a library unit which has a body

@item
subunits
@end itemize

@noindent
because they are compiled as part of compiling related units. GNAT
package specs
when the corresponding body is compiled, and subunits when the parent is
compiled.

@cindex cannot generate code
If you attempt to compile any of these files, you will get one of the
following error messages (where @var{fff} is the name of the file you
compiled):

@smallexample
cannot generate code for file @var{fff} (package spec)
to check package spec, use -gnatc

cannot generate code for file @var{fff} (missing subunits)
to check parent unit, use -gnatc

cannot generate code for file @var{fff} (subprogram spec)
to check subprogram spec, use -gnatc

cannot generate code for file @var{fff} (subunit)
to check subunit, use -gnatc
@end smallexample

@noindent
As indicated by the above error messages, if you want to submit
one of these files to the compiler to check for correct semantics
without generating code, then use the @option{-gnatc} switch.

The basic command for compiling a file containing an Ada unit is

@smallexample
@c $ gcc -c @ovar{switches} @file{file name}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gcc -c @r{[}@var{switches}@r{]} @file{file name}
@end smallexample

@noindent
where @var{file name} is the name of the Ada file (usually
having an extension
@file{.ads} for a spec or @file{.adb} for a body).
You specify the
@option{-c} switch to tell @command{gcc} to compile, but not link, the file.
The result of a successful compilation is an object file, which has the
same name as the source file but an extension of @file{.o} and an Ada
Library Information (ALI) file, which also has the same name as the
source file, but with @file{.ali} as the extension. GNAT creates these
two output files in the current directory, but you may specify a source
file in any directory using an absolute or relative path specification
containing the directory information.

@findex gnat1
@command{gcc} is actually a driver program that looks at the extensions of
the file arguments and loads the appropriate compiler. For example, the
GNU C compiler is @file{cc1}, and the Ada compiler is @file{gnat1}.
These programs are in directories known to the driver program (in some
configurations via environment variables you set), but need not be in
your path. The @command{gcc} driver also calls the assembler and any other
utilities needed to complete the generation of the required object
files.

It is possible to supply several file names on the same @command{gcc}
command. This causes @command{gcc} to call the appropriate compiler for
each file. For example, the following command lists two separate
files to be compiled:

@smallexample
$ gcc -c x.adb y.adb
@end smallexample

@noindent
calls @code{gnat1} (the Ada compiler) twice to compile @file{x.adb} and
@file{y.adb}.
The compiler generates two object files @file{x.o} and @file{y.o}
and the two ALI files @file{x.ali} and @file{y.ali}.
Any switches apply to all the files listed,

@node Switches for gcc
@section Switches for @command{gcc}

@noindent
The @command{gcc} command accepts switches that control the
compilation process. These switches are fully described in this section.
First we briefly list all the switches, in alphabetical order, then we
describe the switches in more detail in functionally grouped sections.

More switches exist for GCC than those documented here, especially
for specific targets. However, their use is not recommended as
they may change code generation in ways that are incompatible with
the Ada run-time library, or can cause inconsistencies between
compilation units.

@menu
* Output and Error Message Control::
* Warning Message Control::
* Debugging and Assertion Control::
* Validity Checking::
* Style Checking::
* Run-Time Checks::
* Using gcc for Syntax Checking::
* Using gcc for Semantic Checking::
* Compiling Different Versions of Ada::
* Character Set Control::
* File Naming Control::
* Subprogram Inlining Control::
* Auxiliary Output Control::
* Debugging Control::
* Exception Handling Control::
* Units to Sources Mapping Files::
* Integrated Preprocessing::
* Code Generation Control::
@end menu

@table @option
@c !sort!
@cindex @option{-b} (@command{gcc})
@item -b @var{target}
Compile your program to run on @var{target}, which is the name of a
system configuration. You must have a GNAT cross-compiler built if
@var{target} is not the same as your host system.

@item -B@var{dir}
@cindex @option{-B} (@command{gcc})
Load compiler executables (for example, @code{gnat1}, the Ada compiler)
from @var{dir} instead of the default location. Only use this switch
when multiple versions of the GNAT compiler are available.
@xref{Directory Options,, Options for Directory Search, gcc, Using the
GNU Compiler Collection (GCC)}, for further details. You would normally
use the @option{-b} or @option{-V} switch instead.

@item -c
@cindex @option{-c} (@command{gcc})
Compile. Always use this switch when compiling Ada programs.

Note: for some other languages when using @command{gcc}, notably in
the case of C and C++, it is possible to use
use @command{gcc} without a @option{-c} switch to
compile and link in one step. In the case of GNAT, you
cannot use this approach, because the binder must be run
and @command{gcc} cannot be used to run the GNAT binder.

@item -fcallgraph-info@r{[}=su,da@r{]}
@cindex @option{-fcallgraph-info} (@command{gcc})
Makes the compiler output callgraph information for the program, on a
per-file basis. The information is generated in the VCG format.  It can
be decorated with additional, per-node and/or per-edge information, if a
list of comma-separated markers is additionally specified. When the
@var{su} marker is specified, the callgraph is decorated with stack usage information; it is equivalent to @option{-fstack-usage}. When the @var{da}
marker is specified, the callgraph is decorated with information about
dynamically allocated objects.

@item -fdump-scos
@cindex @option{-fdump-scos} (@command{gcc})
Generates SCO (Source Coverage Obligation) information in the ALI file.
This information is used by advanced coverage tools. See unit @file{SCOs}
in the compiler sources for details in files @file{scos.ads} and
@file{scos.adb}.

@item -fdump-xref
@cindex @option{-fdump-xref} (@command{gcc})
Generates cross reference information in GLI files for C and C++ sources.
The GLI files have the same syntax as the ALI files for Ada, and can be used
for source navigation in IDEs and on the command line using e.g. gnatxref
and the @option{--ext=gli} switch.

@item -flto@r{[}=n@r{]}
@cindex @option{-flto} (@command{gcc})
Enables Link Time Optimization. This switch must be used in conjunction
with the traditional @option{-Ox} switches and instructs the compiler to
defer most optimizations until the link stage. The advantage of this
approach is that the compiler can do a whole-program analysis and choose
the best interprocedural optimization strategy based on a complete view
of the program, instead of a fragmentary view with the usual approach.
This can also speed up the compilation of big programs and reduce the
size of the executable, compared with a traditional per-unit compilation
with inlining across modules enabled by the @option{-gnatn} switch.
The drawback of this approach is that it may require more memory and that
the debugging information generated by -g with it might be hardly usable.
The switch, as well as the accompanying @option{-Ox} switches, must be
specified both for the compilation and the link phases.
If the @var{n} parameter is specified, the optimization and final code
generation at link time are executed using @var{n} parallel jobs by
means of an installed @command{make} program.

@item -fno-inline
@cindex @option{-fno-inline} (@command{gcc})
Suppresses all inlining, even if other optimization or inlining
switches are set.  This includes suppression of inlining that
results from the use of the pragma @code{Inline_Always}.
Any occurrences of pragma @code{Inline} or @code{Inline_Always}
are ignored, and @option{-gnatn} and @option{-gnatN} have no
effects if this switch is present.  Note that inlining can also
be suppressed on a finer-grained basis with pragma @code{No_Inline}.

@item -fno-inline-functions
@cindex @option{-fno-inline-functions} (@command{gcc})
Suppresses automatic inlining of subprograms, which is enabled
if @option{-O3} is used.

@item -fno-inline-small-functions
@cindex @option{-fno-inline-small-functions} (@command{gcc})
Suppresses automatic inlining of small subprograms, which is enabled
if @option{-O2} is used.

@item -fno-inline-functions-called-once
@cindex @option{-fno-inline-functions-called-once} (@command{gcc})
Suppresses inlining of subprograms local to the unit and called once
from within it, which is enabled if @option{-O1} is used.

@item -fno-ivopts
@cindex @option{-fno-ivopts} (@command{gcc})
Suppresses high-level loop induction variable optimizations, which are
enabled if @option{-O1} is used. These optimizations are generally
profitable but, for some specific cases of loops with numerous uses
of the iteration variable that follow a common pattern, they may end
up destroying the regularity that could be exploited at a lower level
and thus producing inferior code.

@item -fno-strict-aliasing
@cindex @option{-fno-strict-aliasing} (@command{gcc})
Causes the compiler to avoid assumptions regarding non-aliasing
of objects of different types. See
@ref{Optimization and Strict Aliasing} for details.

@item -fstack-check
@cindex @option{-fstack-check} (@command{gcc})
Activates stack checking.
See @ref{Stack Overflow Checking} for details.

@item -fstack-usage
@cindex @option{-fstack-usage} (@command{gcc})
Makes the compiler output stack usage information for the program, on a
per-subprogram basis. See @ref{Static Stack Usage Analysis} for details.

@item -g
@cindex @option{-g} (@command{gcc})
Generate debugging information. This information is stored in the object
file and copied from there to the final executable file by the linker,
where it can be read by the debugger. You must use the
@option{-g} switch if you plan on using the debugger.

@item -gnat83
@cindex @option{-gnat83} (@command{gcc})
Enforce Ada 83 restrictions.

@item -gnat95
@cindex @option{-gnat95} (@command{gcc})
Enforce Ada 95 restrictions.

Note: for compatibility with some Ada 95 compilers which support only
the @code{overriding} keyword of Ada 2005, the @option{-gnatd.D} switch can
be used along with @option{-gnat95} to achieve a similar effect with GNAT.

@option{-gnatd.D} instructs GNAT to consider @code{overriding} as a keyword
and handle its associated semantic checks, even in Ada 95 mode.

@item -gnat05
@cindex @option{-gnat05} (@command{gcc})
Allow full Ada 2005 features.

@item -gnat2005
@cindex @option{-gnat2005} (@command{gcc})
Allow full Ada 2005 features (same as @option{-gnat05})

@item -gnat12
@cindex @option{-gnat12} (@command{gcc})

@item -gnat2012
@cindex @option{-gnat2012} (@command{gcc})
Allow full Ada 2012 features (same as @option{-gnat12})

@item -gnata
@cindex @option{-gnata} (@command{gcc})
Assertions enabled. @code{Pragma Assert} and @code{pragma Debug} to be
activated. Note that these pragmas can also be controlled using the
configuration pragmas @code{Assertion_Policy} and @code{Debug_Policy}.
It also activates pragmas @code{Check}, @code{Precondition}, and
@code{Postcondition}. Note that these pragmas can also be controlled
using the configuration pragma @code{Check_Policy}. In Ada 2012, it
also activates all assertions defined in the RM as aspects: preconditions,
postconditions, type invariants and (sub)type predicates. In all Ada modes,
corresponding pragmas for type invariants and (sub)type predicates are
also activated. The default is that all these assertions are disabled,
and have no effect, other than being checked for syntactic validity, and
in the case of subtype predicates, constructions such as membership tests
still test predicates even if assertions are turned off.

@item -gnatA
@cindex @option{-gnatA} (@command{gcc})
Avoid processing @file{gnat.adc}. If a @file{gnat.adc} file is present,
it will be ignored.

@item -gnatb
@cindex @option{-gnatb} (@command{gcc})
Generate brief messages to @file{stderr} even if verbose mode set.

@item -gnatB
@cindex @option{-gnatB} (@command{gcc})
Assume no invalid (bad) values except for 'Valid attribute use
(@pxref{Validity Checking}).

@item -gnatc
@cindex @option{-gnatc} (@command{gcc})
Check syntax and semantics only (no code generation attempted). When the
compiler is invoked by @command{gnatmake}, if the switch @option{-gnatc} is
only given to the compiler (after @option{-cargs} or in package Compiler of
the project file, @command{gnatmake} will fail because it will not find the
object file after compilation. If @command{gnatmake} is called with
@option{-gnatc} as a builder switch (before @option{-cargs} or in package
Builder of the project file) then @command{gnatmake} will not fail because
it will not look for the object files after compilation, and it will not try
to build and link. This switch may not be given if a previous @code{-gnatR}
switch has been given, since @code{-gnatR} requires that the code generator
be called to complete determination of representation information.

@item -gnatC
@cindex @option{-gnatC} (@command{gcc})
Generate CodePeer intermediate format (no code generation attempted).
This switch will generate an intermediate representation suitable for
use by CodePeer (@file{.scil} files). This switch is not compatible with
code generation (it will, among other things, disable some switches such
as -gnatn, and enable others such as -gnata).

@item -gnatd
@cindex @option{-gnatd} (@command{gcc})
Specify debug options for the compiler. The string of characters after
the @option{-gnatd} specify the specific debug options. The possible
characters are 0-9, a-z, A-Z, optionally preceded by a dot. See
compiler source file @file{debug.adb} for details of the implemented
debug options. Certain debug options are relevant to applications
programmers, and these are documented at appropriate points in this
users guide.

@item -gnatD
@cindex @option{-gnatD[nn]} (@command{gcc})
Create expanded source files for source level debugging. This switch
also suppress generation of cross-reference information
(see @option{-gnatx}). Note that this switch is not allowed if a previous
-gnatR switch has been given, since these two switches are not compatible.

@item -gnateA
@cindex @option{-gnateA} (@command{gcc})
Check that the actual parameters of a subprogram call are not aliases of one
another. To qualify as aliasing, the actuals must denote objects of a composite
type, their memory locations must be identical or overlapping, and at least one
of the corresponding formal parameters must be of mode OUT or IN OUT.

@smallexample
type Rec_Typ is record
   Data : Integer := 0;
end record;

function Self (Val : Rec_Typ) return Rec_Typ is
begin
   return Val;
end Self;

procedure Detect_Aliasing (Val_1 : in out Rec_Typ; Val_2 : Rec_Typ) is
begin
   null;
end Detect_Aliasing;

Obj : Rec_Typ;

Detect_Aliasing (Obj, Obj);
Detect_Aliasing (Obj, Self (Obj));
@end smallexample

In the example above, the first call to @code{Detect_Aliasing} fails with a
@code{Program_Error} at runtime because the actuals for @code{Val_1} and
@code{Val_2} denote the same object. The second call executes without raising
an exception because @code{Self(Obj)} produces an anonymous object which does
not share the memory location of @code{Obj}.

@item -gnatec=@var{path}
@cindex @option{-gnatec} (@command{gcc})
Specify a configuration pragma file
(the equal sign is optional)
(@pxref{The Configuration Pragmas Files}).

@item -gnateC
@cindex @option{-gnateC} (@command{gcc})
Generate CodePeer messages in a compiler-like format. This switch is only
effective if @option{-gnatcC} is also specified and requires an installation
of CodePeer.

@item -gnated
@cindex @option{-gnated} (@command{gcc})
Disable atomic synchronization

@item -gnateDsymbol@r{[}=@var{value}@r{]}
@cindex @option{-gnateD} (@command{gcc})
Defines a symbol, associated with @var{value}, for preprocessing.
(@pxref{Integrated Preprocessing}).

@item -gnateE
@cindex @option{-gnateE} (@command{gcc})
Generate extra information in exception messages. In particular, display
extra column information and the value and range associated with index and
range check failures, and extra column information for access checks.
In cases where the compiler is able to determine at compile time that
a check will fail, it gives a warning, and the extra information is not
produced at run time.

@item -gnatef
@cindex @option{-gnatef} (@command{gcc})
Display full source path name in brief error messages.

@item -gnateF
@cindex @option{-gnateF} (@command{gcc})
Check for overflow on all floating-point operations, including those
for unconstrained predefined types. See description of pragma
@code{Check_Float_Overflow} in GNAT RM.

@item -gnateG
@cindex @option{-gnateG} (@command{gcc})
Save result of preprocessing in a text file.

@item -gnatei@var{nnn}
@cindex @option{-gnatei} (@command{gcc})
Set maximum number of instantiations during compilation of a single unit to
@var{nnn}. This may be useful in increasing the default maximum of 8000 for
the rare case when a single unit legitimately exceeds this limit.

@item -gnateI@var{nnn}
@cindex @option{-gnateI} (@command{gcc})
Indicates that the source is a multi-unit source and that the index of the
unit to compile is @var{nnn}. @var{nnn} needs to be a positive number and need
to be a valid index in the multi-unit source.

@item -gnatel
@cindex @option{-gnatel} (@command{gcc})
This switch can be used with the static elaboration model to issue info
messages showing
where implicit @code{pragma Elaborate} and @code{pragma Elaborate_All}
are generated. This is useful in diagnosing elaboration circularities
caused by these implicit pragmas when using the static elaboration
model. See See the section in this guide on elaboration checking for
further details. These messages are not generated by default, and are
intended only for temporary use when debugging circularity problems.

@item -gnateL
@cindex @option{-gnatel} (@command{gcc})
This switch turns off the info messages about implicit elaboration pragmas.

@item -gnatem=@var{path}
@cindex @option{-gnatem} (@command{gcc})
Specify a mapping file
(the equal sign is optional)
(@pxref{Units to Sources Mapping Files}).

@item -gnatep=@var{file}
@cindex @option{-gnatep} (@command{gcc})
Specify a preprocessing data file
(the equal sign is optional)
(@pxref{Integrated Preprocessing}).

@item -gnateP
@cindex @option{-gnateP} (@command{gcc})
Turn categorization dependency errors into warnings.
Ada requires that units that WITH one another have compatible categories, for
example a Pure unit cannot WITH a Preelaborate unit. If this switch is used,
these errors become warnings (which can be ignored, or suppressed in the usual
manner). This can be useful in some specialized circumstances such as the
temporary use of special test software.

@item -gnateS
@cindex @option{-gnateS} (@command{gcc})
Synonym of @option{-fdump-scos}, kept for backwards compatibility.

@item -gnatet=@var{path}
@cindex @option{-gnatet=file} (@command{gcc})
Generate target dependent information. The format of the output file is
described in the section about switch @option{-gnateT}.

@item -gnateT=@var{path}
@cindex @option{-gnateT} (@command{gcc})
Read target dependent information, such as endianness or sizes and alignments
of base type. If this switch is passed, the default target dependent
information of the compiler is replaced by the one read from the input file.
This is used by tools other than the compiler, e.g. to do
semantic analysis of programs that will run on some other target than
the machine on which the tool is run.

The following target dependent values should be defined,
where @code{Nat} denotes a natural integer value, @code{Pos} denotes a
positive integer value, and fields marked with a question mark are
boolean fields, where a value of 0 is False, and a value of 1 is True:

@smallexample
Bits_BE                    : Nat; -- Bits stored big-endian?
Bits_Per_Unit              : Pos; -- Bits in a storage unit
Bits_Per_Word              : Pos; -- Bits in a word
Bytes_BE                   : Nat; -- Bytes stored big-endian?
Char_Size                  : Pos; -- Standard.Character'Size
Double_Float_Alignment     : Nat; -- Alignment of double float
Double_Scalar_Alignment    : Nat; -- Alignment of double length scalar
Double_Size                : Pos; -- Standard.Long_Float'Size
Float_Size                 : Pos; -- Standard.Float'Size
Float_Words_BE             : Nat; -- Float words stored big-endian?
Int_Size                   : Pos; -- Standard.Integer'Size
Long_Double_Size           : Pos; -- Standard.Long_Long_Float'Size
Long_Long_Size             : Pos; -- Standard.Long_Long_Integer'Size
Long_Size                  : Pos; -- Standard.Long_Integer'Size
Maximum_Alignment          : Pos; -- Maximum permitted alignment
Max_Unaligned_Field        : Pos; -- Maximum size for unaligned bit field
Pointer_Size               : Pos; -- System.Address'Size
Short_Enums                : Nat; -- Short foreign convention enums?
Short_Size                 : Pos; -- Standard.Short_Integer'Size
Strict_Alignment           : Nat; -- Strict alignment?
System_Allocator_Alignment : Nat; -- Alignment for malloc calls
Wchar_T_Size               : Pos; -- Interfaces.C.wchar_t'Size
Words_BE                   : Nat; -- Words stored big-endian?
@end smallexample

The format of the input file is as follows. First come the values of
the variables defined above, with one line per value:

@smallexample
name  value
@end smallexample

where @code{name} is the name of the parameter, spelled out in full,
and cased as in the above list, and @code{value} is an unsigned decimal
integer. Two or more blanks separates the name from the value.

All the variables must be present, in alphabetical order (i.e. the
same order as the list above).

Then there is a blank line to separate the two parts of the file. Then
come the lines showing the floating-point types to be registered, with
one line per registered mode:

@smallexample
name  digs float_rep size alignment
@end smallexample

where @code{name} is the string name of the type (which can have
single spaces embedded in the name (e.g. long double), @code{digs} is
the number of digits for the floating-point type, @code{float_rep} is
the float representation (I/V/A for IEEE-754-Binary, Vax_Native,
AAMP), @code{size} is the size in bits, @code{alignment} is the
alignment in bits. The name is followed by at least two blanks, fields
are separated by at least one blank, and a LF character immediately
follows the alignment field.

Here is an example of a target parameterization file:

@smallexample
Bits_BE                       0
Bits_Per_Unit                 8
Bits_Per_Word                64
Bytes_BE                      0
Char_Size                     8
Double_Float_Alignment        0
Double_Scalar_Alignment       0
Double_Size                  64
Float_Size                   32
Float_Words_BE                0
Int_Size                     64
Long_Double_Size            128
Long_Long_Size               64
Long_Size                    64
Maximum_Alignment            16
Max_Unaligned_Field          64
Pointer_Size                 64
Short_Size                   16
Strict_Alignment              0
System_Allocator_Alignment   16
Wchar_T_Size                 32
Words_BE                      0

float         15  I  64  64
double        15  I  64  64
long double   18  I  80 128
TF            33  I 128 128
@end smallexample

@item -gnateu
@cindex @option{-gnateu} (@command{gcc})
Ignore unrecognized validity, warning, and style switches that
appear after this switch is given. This may be useful when
compiling sources developed on a later version of the compiler
with an earlier version. Of course the earlier version must
support this switch.

@item -gnateV
@cindex @option{-gnateV} (@command{gcc})
Check that all actual parameters of a subprogram call are valid according to
the rules of validity checking (@pxref{Validity Checking}).

@item -gnateY
@cindex @option{-gnateY} (@command{gcc})
Ignore all STYLE_CHECKS pragmas. Full legality checks
are still carried out, but the pragmas have no effect
on what style checks are active. This allows all style
checking options to be controlled from the command line.

@item -gnatE
@cindex @option{-gnatE} (@command{gcc})
Full dynamic elaboration checks.

@item -gnatf
@cindex @option{-gnatf} (@command{gcc})
Full errors. Multiple errors per line, all undefined references, do not
attempt to suppress cascaded errors.

@item -gnatF
@cindex @option{-gnatF} (@command{gcc})
Externals names are folded to all uppercase.

@item -gnatg
@cindex @option{-gnatg} (@command{gcc})
Internal GNAT implementation mode. This should not be used for
applications programs, it is intended only for use by the compiler
and its run-time library. For documentation, see the GNAT sources.
Note that @option{-gnatg} implies
@option{-gnatw.ge} and
@option{-gnatyg}
so that all standard warnings and all standard style options are turned on.
All warnings and style messages are treated as errors.

@item -gnatG=nn
@cindex @option{-gnatG[nn]} (@command{gcc})
List generated expanded code in source form.

@item -gnath
@cindex @option{-gnath} (@command{gcc})
Output usage information. The output is written to @file{stdout}.

@item -gnati@var{c}
@cindex @option{-gnati} (@command{gcc})
Identifier character set
(@var{c}=1/2/3/4/8/9/p/f/n/w).
For details of the possible selections for @var{c},
see @ref{Character Set Control}.

@item -gnatI
@cindex @option{-gnatI} (@command{gcc})
Ignore representation clauses. When this switch is used,
representation clauses are treated as comments. This is useful
when initially porting code where you want to ignore rep clause
problems, and also for compiling foreign code (particularly
for use with ASIS). The representation clauses that are ignored
are: enumeration_representation_clause, record_representation_clause,
and attribute_definition_clause for the following attributes:
Address, Alignment, Bit_Order, Component_Size, Machine_Radix,
Object_Size, Size, Small, Stream_Size, and Value_Size.
Note that this option should be used only for compiling -- the
code is likely to malfunction at run time.

Note that when @code{-gnatct} is used to generate trees for input
into @code{ASIS} tools, these representation clauses are removed
from the tree and ignored. This means that the tool will not see them.

@item -gnatjnn
@cindex @option{-gnatjnn} (@command{gcc})
Reformat error messages to fit on nn character lines

@item -gnatk=@var{n}
@cindex @option{-gnatk} (@command{gcc})
Limit file names to @var{n} (1-999) characters (@code{k} = krunch).

@item -gnatl
@cindex @option{-gnatl} (@command{gcc})
Output full source listing with embedded error messages.

@item -gnatL
@cindex @option{-gnatL} (@command{gcc})
Used in conjunction with -gnatG or -gnatD to intersperse original
source lines (as comment lines with line numbers) in the expanded
source output.

@item -gnatm=@var{n}
@cindex @option{-gnatm} (@command{gcc})
Limit number of detected error or warning messages to @var{n}
where @var{n} is in the range 1..999999. The default setting if
no switch is given is 9999. If the number of warnings reaches this
limit, then a message is output and further warnings are suppressed,
but the compilation is continued. If the number of error messages
reaches this limit, then a message is output and the compilation
is abandoned. The equal sign here is optional. A value of zero
means that no limit applies.

@item -gnatn[12]
@cindex @option{-gnatn} (@command{gcc})
Activate inlining for subprograms for which pragma @code{Inline} is
specified. This inlining is performed by the GCC back-end. An optional
digit sets the inlining level: 1 for moderate inlining across modules
or 2 for full inlining across modules. If no inlining level is specified,
the compiler will pick it based on the optimization level.

@item -gnatN
@cindex @option{-gnatN} (@command{gcc})
Activate front end inlining for subprograms for which
pragma @code{Inline} is specified. This inlining is performed
by the front end and will be visible in the
@option{-gnatG} output.

When using a gcc-based back end (in practice this means using any version
of GNAT other than the JGNAT, .NET or GNAAMP versions), then the use of
@option{-gnatN} is deprecated, and the use of @option{-gnatn} is preferred.
Historically front end inlining was more extensive than the gcc back end
inlining, but that is no longer the case.

@item -gnato0
@cindex @option{-gnato0} (@command{gcc})
Suppresses overflow checking. This causes the behavior of the compiler to
match the default for older versions where overflow checking was suppressed
by default. This is equivalent to having
@code{pragma Suppress (Overflow_Mode)} in a configuration pragma file.

@item -gnato??
@cindex @option{-gnato??} (@command{gcc})
Set default mode for handling generation of code to avoid intermediate
arithmetic overflow. Here `@code{??}' is two digits, a
single digit, or nothing. Each digit is one of the digits `@code{1}'
through `@code{3}':

@itemize @bullet
@item   @code{1}:
all intermediate overflows checked against base type (@code{STRICT})
@item   @code{2}:
minimize intermediate overflows (@code{MINIMIZED})
@item   @code{3}:
eliminate intermediate overflows (@code{ELIMINATED})
@end itemize

If only one digit appears then it applies to all
cases; if two digits are given, then the first applies outside
assertions, and the second within assertions.

If no digits follow the @option{-gnato}, then it is equivalent to
@option{-gnato11},
causing all intermediate overflows to be handled in strict mode.

This switch also causes arithmetic overflow checking to be performed
(as though @code{pragma Unsuppress (Overflow_Mode)} had been specified.

The default if no option @option{-gnato} is given is that overflow handling
is in @code{STRICT} mode (computations done using the base type), and that
overflow checking is enabled.

Note that division by zero is a separate check that is not
controlled by this switch (division by zero checking is on by default).

See also @ref{Specifying the Desired Mode}.

@item -gnatp
@cindex @option{-gnatp} (@command{gcc})
Suppress all checks. See @ref{Run-Time Checks} for details. This switch
has no effect if cancelled by a subsequent @option{-gnat-p} switch.

@item -gnat-p
@cindex @option{-gnat-p} (@command{gcc})
Cancel effect of previous @option{-gnatp} switch.

@item -gnatP
@cindex @option{-gnatP} (@command{gcc})
Enable polling. This is required on some systems (notably Windows NT) to
obtain asynchronous abort and asynchronous transfer of control capability.
@xref{Pragma Polling,,, gnat_rm, GNAT Reference Manual}, for full
details.

@item -gnatq
@cindex @option{-gnatq} (@command{gcc})
Don't quit. Try semantics, even if parse errors.

@item -gnatQ
@cindex @option{-gnatQ} (@command{gcc})
Don't quit. Generate @file{ALI} and tree files even if illegalities.
Note that code generation is still suppressed in the presence of any
errors, so even with @option{-gnatQ} no object file is generated.

@item -gnatr
@cindex @option{-gnatr} (@command{gcc})
Treat pragma Restrictions as Restriction_Warnings.

@item -gnatR@r{[}0@r{/}1@r{/}2@r{/}3@r{[}s@r{]]}
@cindex @option{-gnatR} (@command{gcc})
Output representation information for declared types and objects.
Note that this switch is not allowed if a previous @code{-gnatD} switch has
been given, since these two switches are not compatible.

@item -gnatRm[s]
Output convention and parameter passing mechanisms for all subprograms.

@item -gnats
@cindex @option{-gnats} (@command{gcc})
Syntax check only.

@item -gnatS
@cindex @option{-gnatS} (@command{gcc})
Print package Standard.

@item -gnatt
@cindex @option{-gnatt} (@command{gcc})
Generate tree output file.

@item -gnatT@var{nnn}
@cindex @option{-gnatT} (@command{gcc})
All compiler tables start at @var{nnn} times usual starting size.

@item -gnatu
@cindex @option{-gnatu} (@command{gcc})
List units for this compilation.

@item -gnatU
@cindex @option{-gnatU} (@command{gcc})
Tag all error messages with the unique string ``error:''

@item -gnatv
@cindex @option{-gnatv} (@command{gcc})
Verbose mode. Full error output with source lines to @file{stdout}.

@item -gnatV
@cindex @option{-gnatV} (@command{gcc})
Control level of validity checking (@pxref{Validity Checking}).

@item -gnatw@var{xxx}
@cindex @option{-gnatw} (@command{gcc})
Warning mode where
@var{xxx} is a string of option letters that denotes
the exact warnings that
are enabled or disabled (@pxref{Warning Message Control}).

@item -gnatW@var{e}
@cindex @option{-gnatW} (@command{gcc})
Wide character encoding method
(@var{e}=n/h/u/s/e/8).

@item -gnatx
@cindex @option{-gnatx} (@command{gcc})
Suppress generation of cross-reference information.

@item -gnatX
@cindex @option{-gnatX} (@command{gcc})
Enable GNAT implementation extensions and latest Ada version.

@item -gnaty
@cindex @option{-gnaty} (@command{gcc})
Enable built-in style checks (@pxref{Style Checking}).

@item -gnatz@var{m}
@cindex @option{-gnatz} (@command{gcc})
Distribution stub generation and compilation
(@var{m}=r/c for receiver/caller stubs).

@item -I@var{dir}
@cindex @option{-I} (@command{gcc})
@cindex RTL
Direct GNAT to search the @var{dir} directory for source files needed by
the current compilation
(@pxref{Search Paths and the Run-Time Library (RTL)}).

@item -I-
@cindex @option{-I-} (@command{gcc})
@cindex RTL
Except for the source file named in the command line, do not look for source
files in the directory containing the source file named in the command line
(@pxref{Search Paths and the Run-Time Library (RTL)}).

@item -mbig-switch
@cindex @option{-mbig-switch} (@command{gcc})
@cindex @code{case} statement (effect of @option{-mbig-switch} option)
This standard gcc switch causes the compiler to use larger offsets in its
jump table representation for @code{case} statements.
This may result in less efficient code, but is sometimes necessary
(for example on HP-UX targets)
@cindex HP-UX and @option{-mbig-switch} option
in order to compile large and/or nested @code{case} statements.

@item -o @var{file}
@cindex @option{-o} (@command{gcc})
This switch is used in @command{gcc} to redirect the generated object file
and its associated ALI file. Beware of this switch with GNAT, because it may
cause the object file and ALI file to have different names which in turn
may confuse the binder and the linker.

@item -nostdinc
@cindex @option{-nostdinc} (@command{gcc})
Inhibit the search of the default location for the GNAT Run Time
Library (RTL) source files.

@item -nostdlib
@cindex @option{-nostdlib} (@command{gcc})
Inhibit the search of the default location for the GNAT Run Time
Library (RTL) ALI files.

@c @item -O@ovar{n}
@c Expanding @ovar macro inline (explanation in macro def comments)
@item -O@r{[}@var{n}@r{]}
@cindex @option{-O} (@command{gcc})
@var{n} controls the optimization level.

@table @asis
@item n = 0
No optimization, the default setting if no @option{-O} appears

@item n = 1
Normal optimization, the default if you specify @option{-O} without
an operand. A good compromise between code quality and compilation
time.

@item n = 2
Extensive optimization, may improve execution time, possibly at the cost of
substantially increased compilation time.

@item n = 3
Same as @option{-O2}, and also includes inline expansion for small subprograms
in the same unit.

@item n = s
Optimize space usage
@end table

@noindent
See also @ref{Optimization Levels}.


@item -pass-exit-codes
@cindex @option{-pass-exit-codes} (@command{gcc})
Catch exit codes from the compiler and use the most meaningful as
exit status.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gcc})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item -S
@cindex @option{-S} (@command{gcc})
Used in place of @option{-c} to
cause the assembler source file to be
generated, using @file{.s} as the extension,
instead of the object file.
This may be useful if you need to examine the generated assembly code.

@item -fverbose-asm
@cindex @option{-fverbose-asm} (@command{gcc})
Used in conjunction with @option{-S}
to cause the generated assembly code file to be annotated with variable
names, making it significantly easier to follow.

@item -v
@cindex @option{-v} (@command{gcc})
Show commands generated by the @command{gcc} driver. Normally used only for
debugging purposes or if you need to be sure what version of the
compiler you are executing.

@item -V @var{ver}
@cindex @option{-V} (@command{gcc})
Execute @var{ver} version of the compiler. This is the @command{gcc}
version, not the GNAT version.

@item -w
@cindex @option{-w} (@command{gcc})
Turn off warnings generated by the back end of the compiler. Use of
this switch also causes the default for front end warnings to be set
to suppress (as though @option{-gnatws} had appeared at the start of
the options).

@end table

@c Combining qualifiers does not work on VMS
You may combine a sequence of GNAT switches into a single switch. For
example, the combined switch

@cindex Combining GNAT switches
@smallexample
-gnatofi3
@end smallexample

@noindent
is equivalent to specifying the following sequence of switches:

@smallexample
-gnato -gnatf -gnati3
@end smallexample

@noindent
The following restrictions apply to the combination of switches
in this manner:

@itemize @bullet
@item
The switch @option{-gnatc} if combined with other switches must come
first in the string.

@item
The switch @option{-gnats} if combined with other switches must come
first in the string.

@item
The switches

@option{-gnatzc} and @option{-gnatzr} may not be combined with any other
switches, and only one of them may appear in the command line.

@item
The switch @option{-gnat-p} may not be combined with any other switch.

@item
Once a ``y'' appears in the string (that is a use of the @option{-gnaty}
switch), then all further characters in the switch are interpreted
as style modifiers (see description of @option{-gnaty}).

@item
Once a ``d'' appears in the string (that is a use of the @option{-gnatd}
switch), then all further characters in the switch are interpreted
as debug flags (see description of @option{-gnatd}).

@item
Once a ``w'' appears in the string (that is a use of the @option{-gnatw}
switch), then all further characters in the switch are interpreted
as warning mode modifiers (see description of @option{-gnatw}).

@item
Once a ``V'' appears in the string (that is a use of the @option{-gnatV}
switch), then all further characters in the switch are interpreted
as validity checking options (@pxref{Validity Checking}).

@item
Option ``em'', ``ec'', ``ep'', ``l='' and ``R'' must be the last options in
a combined list of options.
@end itemize

@node Output and Error Message Control
@subsection Output and Error Message Control
@findex stderr

@noindent
The standard default format for error messages is called ``brief format''.
Brief format messages are written to @file{stderr} (the standard error
file) and have the following form:

@smallexample
e.adb:3:04: Incorrect spelling of keyword "function"
e.adb:4:20: ";" should be "is"
@end smallexample

@noindent
The first integer after the file name is the line number in the file,
and the second integer is the column number within the line.
@code{GPS} can parse the error messages
and point to the referenced character.
The following switches provide control over the error message
format:

@table @option
@c !sort!
@item -gnatv
@cindex @option{-gnatv} (@command{gcc})
@findex stdout
The v stands for verbose.
The effect of this setting is to write long-format error
messages to @file{stdout} (the standard output file.
The same program compiled with the
@option{-gnatv} switch would generate:

@smallexample
@cartouche
3. funcion X (Q : Integer)
   |
>>> Incorrect spelling of keyword "function"
4. return Integer;
                 |
>>> ";" should be "is"
@end cartouche
@end smallexample

@noindent
The vertical bar indicates the location of the error, and the @samp{>>>}
prefix can be used to search for error messages. When this switch is
used the only source lines output are those with errors.

@item -gnatl
@cindex @option{-gnatl} (@command{gcc})
The @code{l} stands for list.
This switch causes a full listing of
the file to be generated. In the case where a body is
compiled, the corresponding spec is also listed, along
with any subunits. Typical output from compiling a package
body @file{p.adb} might look like:

@smallexample @c ada
@cartouche
 Compiling: p.adb

     1. @b{package} @b{body} p @b{is}
     2.    @b{procedure} a;
     3.    @b{procedure} a @b{is} @b{separate};
     4. @b{begin}
     5.    @b{null}
               |
        >>> missing ";"

     6. @b{end};

Compiling: p.ads

     1. @b{package} p @b{is}
     2.    @b{pragma} Elaborate_Body
                                |
        >>> missing ";"

     3. @b{end} p;

Compiling: p-a.adb

     1. @b{separate} p
                |
        >>> missing "("

     2. @b{procedure} a @b{is}
     3. @b{begin}
     4.    @b{null}
               |
        >>> missing ";"

     5. @b{end};
@end cartouche
@end smallexample

@noindent
@findex stderr
When you specify the @option{-gnatv} or @option{-gnatl} switches and
standard output is redirected, a brief summary is written to
@file{stderr} (standard error) giving the number of error messages and
warning messages generated.

@item -gnatl=file
@cindex @option{-gnatl=fname} (@command{gcc})
This has the same effect as @option{-gnatl} except that the output is
written to a file instead of to standard output. If the given name
@file{fname} does not start with a period, then it is the full name
of the file to be written. If @file{fname} is an extension, it is
appended to the name of the file being compiled. For example, if
file @file{xyz.adb} is compiled with @option{-gnatl=.lst},
then the output is written to file xyz.adb.lst.

@item -gnatU
@cindex @option{-gnatU} (@command{gcc})
This switch forces all error messages to be preceded by the unique
string ``error:''. This means that error messages take a few more
characters in space, but allows easy searching for and identification
of error messages.

@item -gnatb
@cindex @option{-gnatb} (@command{gcc})
The @code{b} stands for brief.
This switch causes GNAT to generate the
brief format error messages to @file{stderr} (the standard error
file) as well as the verbose
format message or full listing (which as usual is written to
@file{stdout} (the standard output file).

@item -gnatm=@var{n}
@cindex @option{-gnatm} (@command{gcc})
The @code{m} stands for maximum.
@var{n} is a decimal integer in the
range of 1 to 999999 and limits the number of error or warning
messages to be generated. For example, using
@option{-gnatm2} might yield

@smallexample
e.adb:3:04: Incorrect spelling of keyword "function"
e.adb:5:35: missing ".."
fatal error: maximum number of errors detected
compilation abandoned
@end smallexample

@noindent
The default setting if
no switch is given is 9999. If the number of warnings reaches this
limit, then a message is output and further warnings are suppressed,
but the compilation is continued. If the number of error messages
reaches this limit, then a message is output and the compilation
is abandoned. A value of zero means that no limit applies.

@noindent
Note that the equal sign is optional, so the switches
@option{-gnatm2} and @option{-gnatm=2} are equivalent.

@item -gnatf
@cindex @option{-gnatf} (@command{gcc})
@cindex Error messages, suppressing
The @code{f} stands for full.
Normally, the compiler suppresses error messages that are likely to be
redundant. This switch causes all error
messages to be generated. In particular, in the case of
references to undefined variables. If a given variable is referenced
several times, the normal format of messages is
@smallexample
e.adb:7:07: "V" is undefined (more references follow)
@end smallexample

@noindent
where the parenthetical comment warns that there are additional
references to the variable @code{V}. Compiling the same program with the
@option{-gnatf} switch yields

@smallexample
e.adb:7:07: "V" is undefined
e.adb:8:07: "V" is undefined
e.adb:8:12: "V" is undefined
e.adb:8:16: "V" is undefined
e.adb:9:07: "V" is undefined
e.adb:9:12: "V" is undefined
@end smallexample

@noindent
The @option{-gnatf} switch also generates additional information for
some error messages.  Some examples are:

@itemize @bullet
@item
Details on possibly non-portable unchecked conversion
@item
List possible interpretations for ambiguous calls
@item
Additional details on incorrect parameters
@end itemize

@item -gnatjnn
@cindex @option{-gnatjnn} (@command{gcc})
In normal operation mode (or if @option{-gnatj0} is used), then error messages
with continuation lines are treated as though the continuation lines were
separate messages (and so a warning with two continuation lines counts as
three warnings, and is listed as three separate messages).

If the @option{-gnatjnn} switch is used with a positive value for nn, then
messages are output in a different manner. A message and all its continuation
lines are treated as a unit, and count as only one warning or message in the
statistics totals. Furthermore, the message is reformatted so that no line
is longer than nn characters.

@item -gnatq
@cindex @option{-gnatq} (@command{gcc})
The @code{q} stands for quit (really ``don't quit'').
In normal operation mode, the compiler first parses the program and
determines if there are any syntax errors. If there are, appropriate
error messages are generated and compilation is immediately terminated.
This switch tells
GNAT to continue with semantic analysis even if syntax errors have been
found. This may enable the detection of more errors in a single run. On
the other hand, the semantic analyzer is more likely to encounter some
internal fatal error when given a syntactically invalid tree.

@item -gnatQ
@cindex @option{-gnatQ} (@command{gcc})
In normal operation mode, the @file{ALI} file is not generated if any
illegalities are detected in the program. The use of @option{-gnatQ} forces
generation of the @file{ALI} file. This file is marked as being in
error, so it cannot be used for binding purposes, but it does contain
reasonably complete cross-reference information, and thus may be useful
for use by tools (e.g., semantic browsing tools or integrated development
environments) that are driven from the @file{ALI} file. This switch
implies @option{-gnatq}, since the semantic phase must be run to get a
meaningful ALI file.

In addition, if @option{-gnatt} is also specified, then the tree file is
generated even if there are illegalities. It may be useful in this case
to also specify @option{-gnatq} to ensure that full semantic processing
occurs. The resulting tree file can be processed by ASIS, for the purpose
of providing partial information about illegal units, but if the error
causes the tree to be badly malformed, then ASIS may crash during the
analysis.

When @option{-gnatQ} is used and the generated @file{ALI} file is marked as
being in error, @command{gnatmake} will attempt to recompile the source when it
finds such an @file{ALI} file, including with switch @option{-gnatc}.

Note that @option{-gnatQ} has no effect if @option{-gnats} is specified,
since ALI files are never generated if @option{-gnats} is set.

@end table

@node Warning Message Control
@subsection Warning Message Control
@cindex Warning messages
@noindent
In addition to error messages, which correspond to illegalities as defined
in the Ada Reference Manual, the compiler detects two kinds of warning
situations.

First, the compiler considers some constructs suspicious and generates a
warning message to alert you to a possible error. Second, if the
compiler detects a situation that is sure to raise an exception at
run time, it generates a warning message. The following shows an example
of warning messages:
@smallexample
e.adb:4:24: warning: creation of object may raise Storage_Error
e.adb:10:17: warning: static value out of range
e.adb:10:17: warning: "Constraint_Error" will be raised at run time
@end smallexample

@noindent
GNAT considers a large number of situations as appropriate
for the generation of warning messages. As always, warnings are not
definite indications of errors. For example, if you do an out-of-range
assignment with the deliberate intention of raising a
@code{Constraint_Error} exception, then the warning that may be
issued does not indicate an error. Some of the situations for which GNAT
issues warnings (at least some of the time) are given in the following
list. This list is not complete, and new warnings are often added to
subsequent versions of GNAT. The list is intended to give a general idea
of the kinds of warnings that are generated.

@itemize @bullet
@item
Possible infinitely recursive calls

@item
Out-of-range values being assigned

@item
Possible order of elaboration problems

@item
Size not a multiple of alignment for a record type

@item
Assertions (pragma Assert) that are sure to fail

@item
Unreachable code

@item
Address clauses with possibly unaligned values, or where an attempt is
made to overlay a smaller variable with a larger one.

@item
Fixed-point type declarations with a null range

@item
Direct_IO or Sequential_IO instantiated with a type that has access values

@item
Variables that are never assigned a value

@item
Variables that are referenced before being initialized

@item
Task entries with no corresponding @code{accept} statement

@item
Duplicate accepts for the same task entry in a @code{select}

@item
Objects that take too much storage

@item
Unchecked conversion between types of differing sizes

@item
Missing @code{return} statement along some execution path in a function

@item
Incorrect (unrecognized) pragmas

@item
Incorrect external names

@item
Allocation from empty storage pool

@item
Potentially blocking operation in protected type

@item
Suspicious parenthesization of expressions

@item
Mismatching bounds in an aggregate

@item
Attempt to return local value by reference

@item
Premature instantiation of a generic body

@item
Attempt to pack aliased components

@item
Out of bounds array subscripts

@item
Wrong length on string assignment

@item
Violations of style rules if style checking is enabled

@item
Unused @code{with} clauses

@item
@code{Bit_Order} usage that does not have any effect

@item
@code{Standard.Duration} used to resolve universal fixed expression

@item
Dereference of possibly null value

@item
Declaration that is likely to cause storage error

@item
Internal GNAT unit @code{with}'ed by application unit

@item
Values known to be out of range at compile time

@item
Unreferenced or unmodified variables. Note that a special
exemption applies to variables which contain any of the substrings
@code{DISCARD, DUMMY, IGNORE, JUNK, UNUSED}, in any casing. Such variables
are considered likely to be intentionally used in a situation where
otherwise a warning would be given, so warnings of this kind are
always suppressed for such variables.

@item
Address overlays that could clobber memory

@item
Unexpected initialization when address clause present

@item
Bad alignment for address clause

@item
Useless type conversions

@item
Redundant assignment statements and other redundant constructs

@item
Useless exception handlers

@item
Accidental hiding of name by child unit

@item
Access before elaboration detected at compile time

@item
A range in a @code{for} loop that is known to be null or might be null

@end itemize

@noindent
The following section lists compiler switches that are available
to control the handling of warning messages. It is also possible
to exercise much finer control over what warnings are issued and
suppressed using the GNAT pragma Warnings, @xref{Pragma Warnings,,,
gnat_rm, GNAT Reference manual}.

@table @option
@c !sort!
@item -gnatwa
@emph{Activate most optional warnings.}
@cindex @option{-gnatwa} (@command{gcc})
This switch activates most optional warning messages.  See the remaining list
in this section for details on optional warning messages that can be
individually controlled.  The warnings that are not turned on by this
switch are:

@itemize
@item @option{-gnatwd} (implicit dereferencing)
@item @option{-gnatw.d} (tag warnings with -gnatw switch)
@item @option{-gnatwh} (hiding)
@item @option{-gnatw.h} (holes in record layouts)
@item @option{-gnatw.k} (redefinition of names in standard)
@item @option{-gnatwl} (elaboration warnings)
@item @option{-gnatw.l} (inherited aspects)
@item @option{-gnatw.n} (atomic synchronization)
@item @option{-gnatwo} (address clause overlay)
@item @option{-gnatw.o} (values set by out parameters ignored)
@item @option{-gnatw.s} (overridden size clause)
@item @option{-gnatwt} (tracking of deleted conditional code)
@item @option{-gnatw.u} (unordered enumeration)
@item @option{-gnatw.w} (use of Warnings Off)
@item @option{-gnatw.y} (reasons for package needing body)
@end itemize

All other optional warnings are turned on.

@item -gnatwA
@emph{Suppress all optional errors.}
@cindex @option{-gnatwA} (@command{gcc})
This switch suppresses all optional warning messages, see remaining list
in this section for details on optional warning messages that can be
individually controlled. Note that unlike switch @option{-gnatws}, the
use of switch @option{-gnatwA} does not suppress warnings that are
normally given unconditionally and cannot be individually controlled
(for example, the warning about a missing exit path in a function).
Also, again unlike switch @option{-gnatws}, warnings suppressed by
the use of switch @option{-gnatwA} can be individually turned back
on. For example the use of switch @option{-gnatwA} followed by
switch @option{-gnatwd} will suppress all optional warnings except
the warnings for implicit dereferencing.

@item -gnatw.a
@emph{Activate warnings on failing assertions.}
@cindex @option{-gnatw.a} (@command{gcc})
@cindex Assert failures
This switch activates warnings for assertions where the compiler can tell at
compile time that the assertion will fail. Note that this warning is given
even if assertions are disabled. The default is that such warnings are
generated.

@item -gnatw.A
@emph{Suppress warnings on failing assertions.}
@cindex @option{-gnatw.A} (@command{gcc})
@cindex Assert failures
This switch suppresses warnings for assertions where the compiler can tell at
compile time that the assertion will fail.

@item -gnatwb
@emph{Activate warnings on bad fixed values.}
@cindex @option{-gnatwb} (@command{gcc})
@cindex Bad fixed values
@cindex Fixed-point Small value
@cindex Small value
This switch activates warnings for static fixed-point expressions whose
value is not an exact multiple of Small. Such values are implementation
dependent, since an implementation is free to choose either of the multiples
that surround the value. GNAT always chooses the closer one, but this is not
required behavior, and it is better to specify a value that is an exact
multiple, ensuring predictable execution. The default is that such warnings
are not generated.

@item -gnatwB
@emph{Suppress warnings on bad fixed values.}
@cindex @option{-gnatwB} (@command{gcc})
This switch suppresses warnings for static fixed-point expressions whose
value is not an exact multiple of Small.

@item -gnatw.b
@emph{Activate warnings on biased representation.}
@cindex @option{-gnatw.b} (@command{gcc})
@cindex Biased representation
This switch activates warnings when a size clause, value size clause, component
clause, or component size clause forces the use of biased representation for an
integer type (e.g. representing a range of 10..11 in a single bit by using 0/1
to represent 10/11). The default is that such warnings are generated.

@item -gnatw.B
@emph{Suppress warnings on biased representation.}
@cindex @option{-gnatwB} (@command{gcc})
This switch suppresses warnings for representation clauses that force the use
of biased representation.

@item -gnatwc
@emph{Activate warnings on conditionals.}
@cindex @option{-gnatwc} (@command{gcc})
@cindex Conditionals, constant
This switch activates warnings for conditional expressions used in
tests that are known to be True or False at compile time. The default
is that such warnings are not generated.
Note that this warning does
not get issued for the use of boolean variables or constants whose
values are known at compile time, since this is a standard technique
for conditional compilation in Ada, and this would generate too many
false positive warnings.

This warning option also activates a special test for comparisons using
the operators ``>='' and`` <=''.
If the compiler can tell that only the equality condition is possible,
then it will warn that the ``>'' or ``<'' part of the test
is useless and that the operator could be replaced by ``=''.
An example would be comparing a @code{Natural} variable <= 0.

This warning option also generates warnings if
one or both tests is optimized away in a membership test for integer
values if the result can be determined at compile time. Range tests on
enumeration types are not included, since it is common for such tests
to include an end point.

This warning can also be turned on using @option{-gnatwa}.

@item -gnatwC
@emph{Suppress warnings on conditionals.}
@cindex @option{-gnatwC} (@command{gcc})
This switch suppresses warnings for conditional expressions used in
tests that are known to be True or False at compile time.

@item -gnatw.c
@emph{Activate warnings on missing component clauses.}
@cindex @option{-gnatw.c} (@command{gcc})
@cindex Component clause, missing
This switch activates warnings for record components where a record
representation clause is present and has component clauses for the
majority, but not all, of the components. A warning is given for each
component for which no component clause is present.

@item -gnatw.C
@emph{Suppress warnings on missing component clauses.}
@cindex @option{-gnatwC} (@command{gcc})
This switch suppresses warnings for record components that are
missing a component clause in the situation described above.

@item -gnatwd
@emph{Activate warnings on implicit dereferencing.}
@cindex @option{-gnatwd} (@command{gcc})
If this switch is set, then the use of a prefix of an access type
in an indexed component, slice, or selected component without an
explicit @code{.all} will generate a warning. With this warning
enabled, access checks occur only at points where an explicit
@code{.all} appears in the source code (assuming no warnings are
generated as a result of this switch). The default is that such
warnings are not generated.

@item -gnatwD
@emph{Suppress warnings on implicit dereferencing.}
@cindex @option{-gnatwD} (@command{gcc})
@cindex Implicit dereferencing
@cindex Dereferencing, implicit
This switch suppresses warnings for implicit dereferences in
indexed components, slices, and selected components.

@item -gnatw.d
@emph{Activate tagging of warning and info messages.}
@cindex @option{-gnatw.d} (@command{gcc})
If this switch is set, then warning messages are tagged, with one of the
following strings:

@table @option

@item [-gnatw?]
Used to tag warnings controlled by the switch @option{-gnatwx} where x
is a letter a-z.

@item [-gnatw.?]
Used to tag warnings controlled by the switch @option{-gnatw.x} where x
is a letter a-z.

@item [-gnatel]
Used to tag elaboration information (info) messages generated when the
static model of elaboration is used and the @option{-gnatel} switch is set.

@item [restriction warning]
Used to tag warning messages for restriction violations, activated by use
of the pragma @option{Restriction_Warnings}.

@item [warning-as-error]
Used to tag warning messages that have been converted to error messages by
use of the pragma Warning_As_Error. Note that such warnings are prefixed by
the string "error: " rather than "warning: ".

@item [enabled by default]
Used to tag all other warnings that are always given by default, unless
warnings are completely suppressed using pragma @option{Warnings(Off)} or
the switch @option{-gnatws}.

@end table

@item -gnatw.D
@emph{Deactivate tagging of warning and info messages messages.}
@cindex @option{-gnatw.d} (@command{gcc})
If this switch is set, then warning messages return to the default
mode in which warnings and info messages are not tagged as described above for
@code{-gnatw.d}.

@item -gnatwe
@emph{Treat warnings and style checks as errors.}
@cindex @option{-gnatwe} (@command{gcc})
@cindex Warnings, treat as error
This switch causes warning messages and style check messages to be
treated as errors.
The warning string still appears, but the warning messages are counted
as errors, and prevent the generation of an object file. Note that this
is the only -gnatw switch that affects the handling of style check messages.
Note also that this switch has no effect on info (information) messages, which
are not treated as errors if this switch is present.

@item -gnatw.e
@emph{Activate every optional warning}
@cindex @option{-gnatw.e} (@command{gcc})
@cindex Warnings, activate every optional warning
This switch activates all optional warnings, including those which
are not activated by @code{-gnatwa}. The use of this switch is not
recommended for normal use. If you turn this switch on, it is almost
certain that you will get large numbers of useless warnings. The
warnings that are excluded from @code{-gnatwa} are typically highly
specialized warnings that are suitable for use only in code that has
been specifically designed according to specialized coding rules.

@item -gnatwf
@emph{Activate warnings on unreferenced formals.}
@cindex @option{-gnatwf} (@command{gcc})
@cindex Formals, unreferenced
This switch causes a warning to be generated if a formal parameter
is not referenced in the body of the subprogram. This warning can
also be turned on using @option{-gnatwu}. The
default is that these warnings are not generated.

@item -gnatwF
@emph{Suppress warnings on unreferenced formals.}
@cindex @option{-gnatwF} (@command{gcc})
This switch suppresses warnings for unreferenced formal
parameters. Note that the
combination @option{-gnatwu} followed by @option{-gnatwF} has the
effect of warning on unreferenced entities other than subprogram
formals.

@item -gnatwg
@emph{Activate warnings on unrecognized pragmas.}
@cindex @option{-gnatwg} (@command{gcc})
@cindex Pragmas, unrecognized
This switch causes a warning to be generated if an unrecognized
pragma is encountered. Apart from issuing this warning, the
pragma is ignored and has no effect. The default
is that such warnings are issued (satisfying the Ada Reference
Manual requirement that such warnings appear).

@item -gnatwG
@emph{Suppress warnings on unrecognized pragmas.}
@cindex @option{-gnatwG} (@command{gcc})
This switch suppresses warnings for unrecognized pragmas.

@item -gnatw.g
@emph{Warnings used for GNAT sources}
@cindex @option{-gnatw.g} (@command{gcc})
This switch sets the warning categories that are used by the standard
GNAT style. Currently this is equivalent to
@option{-gnatwAao.sI.C.V.X}
but more warnings may be added in the future without advanced notice.

@item -gnatwh
@emph{Activate warnings on hiding.}
@cindex @option{-gnatwh} (@command{gcc})
@cindex Hiding of Declarations
This switch activates warnings on hiding declarations.
A declaration is considered hiding
if it is for a non-overloadable entity, and it declares an entity with the
same name as some other entity that is directly or use-visible. The default
is that such warnings are not generated.

@item -gnatwH
@emph{Suppress warnings on hiding.}
@cindex @option{-gnatwH} (@command{gcc})
This switch suppresses warnings on hiding declarations.

@item -gnatw.h
@emph{Activate warnings on holes/gaps in records.}
@cindex @option{-gnatw.h} (@command{gcc})
@cindex Record Representation (gaps)
This switch activates warnings on component clauses in record
representation clauses that leave holes (gaps) in the record layout.
If this warning option is active, then record representation clauses
should specify a contiguous layout, adding unused fill fields if needed.

@item -gnatw.H
@emph{Suppress warnings on holes/gaps in records.}
@cindex @option{-gnatw.H} (@command{gcc})
This switch suppresses warnings on component clauses in record
representation clauses that leave holes (haps) in the record layout.

@item -gnatwi
@emph{Activate warnings on implementation units.}
@cindex @option{-gnatwi} (@command{gcc})
This switch activates warnings for a @code{with} of an internal GNAT
implementation unit, defined as any unit from the @code{Ada},
@code{Interfaces}, @code{GNAT},
 or @code{System}
hierarchies that is not
documented in either the Ada Reference Manual or the GNAT
Programmer's Reference Manual. Such units are intended only
for internal implementation purposes and should not be @code{with}'ed
by user programs. The default is that such warnings are generated

@item -gnatwI
@emph{Disable warnings on implementation units.}
@cindex @option{-gnatwI} (@command{gcc})
This switch disables warnings for a @code{with} of an internal GNAT
implementation unit.

@item -gnatw.i
@emph{Activate warnings on overlapping actuals.}
@cindex @option{-gnatw.i} (@command{gcc})
This switch enables a warning on statically detectable overlapping actuals in
a subprogram call, when one of the actuals is an in-out parameter, and the
types of the actuals are not by-copy types. This warning is off by default.

@item -gnatw.I
@emph{Disable warnings on overlapping actuals.}
@cindex @option{-gnatw.I} (@command{gcc})
This switch disables warnings on overlapping actuals in a call..

@item -gnatwj
@emph{Activate warnings on obsolescent features (Annex J).}
@cindex @option{-gnatwj} (@command{gcc})
@cindex Features, obsolescent
@cindex Obsolescent features
If this warning option is activated, then warnings are generated for
calls to subprograms marked with @code{pragma Obsolescent} and
for use of features in Annex J of the Ada Reference Manual. In the
case of Annex J, not all features are flagged. In particular use
of the renamed packages (like @code{Text_IO}) and use of package
@code{ASCII} are not flagged, since these are very common and
would generate many annoying positive warnings. The default is that
such warnings are not generated.

In addition to the above cases, warnings are also generated for
GNAT features that have been provided in past versions but which
have been superseded (typically by features in the new Ada standard).
For example, @code{pragma Ravenscar} will be flagged since its
function is replaced by @code{pragma Profile(Ravenscar)}, and
@code{pragma Interface_Name} will be flagged since its function
is replaced by @code{pragma Import}.

Note that this warning option functions differently from the
restriction @code{No_Obsolescent_Features} in two respects.
First, the restriction applies only to annex J features.
Second, the restriction does flag uses of package @code{ASCII}.

@item -gnatwJ
@emph{Suppress warnings on obsolescent features (Annex J).}
@cindex @option{-gnatwJ} (@command{gcc})
This switch disables warnings on use of obsolescent features.

@item -gnatwk
@emph{Activate warnings on variables that could be constants.}
@cindex @option{-gnatwk} (@command{gcc})
This switch activates warnings for variables that are initialized but
never modified, and then could be declared constants. The default is that
such warnings are not given.

@item -gnatwK
@emph{Suppress warnings on variables that could be constants.}
@cindex @option{-gnatwK} (@command{gcc})
This switch disables warnings on variables that could be declared constants.

@item -gnatw.k
@emph{Activate warnings on redefinition of names in standard.}
@cindex @option{-gnatw.k} (@command{gcc})
This switch activates warnings for declarations that declare a name that
is defined in package Standard. Such declarations can be confusing,
especially since the names in package Standard continue to be directly
visible, meaning that use visibiliy on such redeclared names does not
work as expected. Names of discriminants and components in records are
not included in this check.

@item -gnatw.K
@emph{Suppress warnings on redefinition of names in standard.}
@cindex @option{-gnatwK} (@command{gcc})
This switch activates warnings for declarations that declare a name that
is defined in package Standard.

@item -gnatwl
@emph{Activate warnings for elaboration pragmas.}
@cindex @option{-gnatwl} (@command{gcc})
@cindex Elaboration, warnings
This switch activates warnings for possible elaboration problems,
including suspicious use
of @code{Elaborate} pragmas, when using the static elaboration model, and
possible situations that may raise @code{Program_Error} when using the
dynamic elaboration model.
See the section in this guide on elaboration checking for further details.
The default is that such warnings
are not generated.

@item -gnatwL
@emph{Suppress warnings for elaboration pragmas.}
@cindex @option{-gnatwL} (@command{gcc})
This switch suppresses warnings for possible elaboration problems.

@item -gnatw.l
@emph{List inherited aspects.}
@cindex @option{-gnatw.l} (@command{gcc})
This switch causes the compiler to list inherited invariants,
preconditions, and postconditions from Type_Invariant'Class, Invariant'Class,
Pre'Class, and Post'Class aspects. Also list inherited subtype predicates.

@item -gnatw.L
@emph{Suppress listing of inherited aspects.}
@cindex @option{-gnatw.L} (@command{gcc})
This switch suppresses listing of inherited aspects.

@item -gnatwm
@emph{Activate warnings on modified but unreferenced variables.}
@cindex @option{-gnatwm} (@command{gcc})
This switch activates warnings for variables that are assigned (using
an initialization value or with one or more assignment statements) but
whose value is never read. The warning is suppressed for volatile
variables and also for variables that are renamings of other variables
or for which an address clause is given.
The default is that these warnings are not given.

@item -gnatwM
@emph{Disable warnings on modified but unreferenced variables.}
@cindex @option{-gnatwM} (@command{gcc})
This switch disables warnings for variables that are assigned or
initialized, but never read.

@item -gnatw.m
@emph{Activate warnings on suspicious modulus values.}
@cindex @option{-gnatw.m} (@command{gcc})
This switch activates warnings for modulus values that seem suspicious.
The cases caught are where the size is the same as the modulus (e.g.
a modulus of 7 with a size of 7 bits), and modulus values of 32 or 64
with no size clause. The guess in both cases is that 2**x was intended
rather than x. In addition expressions of the form 2*x for small x
generate a warning (the almost certainly accurate guess being that
2**x was intended). The default is that these warnings are given.

@item -gnatw.M
@emph{Disable warnings on suspicious modulus values.}
@cindex @option{-gnatw.M} (@command{gcc})
This switch disables warnings for suspicious modulus values.

@item -gnatwn
@emph{Set normal warnings mode.}
@cindex @option{-gnatwn} (@command{gcc})
This switch sets normal warning mode, in which enabled warnings are
issued and treated as warnings rather than errors. This is the default
mode. the switch @option{-gnatwn} can be used to cancel the effect of
an explicit @option{-gnatws} or
@option{-gnatwe}. It also cancels the effect of the
implicit @option{-gnatwe} that is activated by the
use of @option{-gnatg}.

@item -gnatw.n
@emph{Activate warnings on atomic synchronization.}
@cindex @option{-gnatw.n} (@command{gcc})
@cindex Atomic Synchronization, warnings
This switch actives warnings when an access to an atomic variable
requires the generation of atomic synchronization code. These
warnings are off by default.
@item -gnatw.N
@emph{Suppress warnings on atomic synchronization.}
@cindex @option{-gnatw.n} (@command{gcc})
@cindex Atomic Synchronization, warnings
This switch suppresses warnings when an access to an atomic variable
requires the generation of atomic synchronization code.

@item -gnatwo
@emph{Activate warnings on address clause overlays.}
@cindex @option{-gnatwo} (@command{gcc})
@cindex Address Clauses, warnings
This switch activates warnings for possibly unintended initialization
effects of defining address clauses that cause one variable to overlap
another. The default is that such warnings are generated.

@item -gnatwO
@emph{Suppress warnings on address clause overlays.}
@cindex @option{-gnatwO} (@command{gcc})
This switch suppresses warnings on possibly unintended initialization
effects of defining address clauses that cause one variable to overlap
another.

@item -gnatw.o
@emph{Activate warnings on modified but unreferenced out parameters.}
@cindex @option{-gnatw.o} (@command{gcc})
This switch activates warnings for variables that are modified by using
them as actuals for a call to a procedure with an out mode formal, where
the resulting assigned value is never read. It is applicable in the case
where there is more than one out mode formal. If there is only one out
mode formal, the warning is issued by default (controlled by -gnatwu).
The warning is suppressed for volatile
variables and also for variables that are renamings of other variables
or for which an address clause is given.
The default is that these warnings are not given.

@item -gnatw.O
@emph{Disable warnings on modified but unreferenced out parameters.}
@cindex @option{-gnatw.O} (@command{gcc})
This switch suppresses warnings for variables that are modified by using
them as actuals for a call to a procedure with an out mode formal, where
the resulting assigned value is never read.

@item -gnatwp
@emph{Activate warnings on ineffective pragma Inlines.}
@cindex @option{-gnatwp} (@command{gcc})
@cindex Inlining, warnings
This switch activates warnings for failure of front end inlining
(activated by @option{-gnatN}) to inline a particular call. There are
many reasons for not being able to inline a call, including most
commonly that the call is too complex to inline. The default is
that such warnings are not given.
Warnings on ineffective inlining by the gcc back-end can be activated
separately, using the gcc switch -Winline.

@item -gnatwP
@emph{Suppress warnings on ineffective pragma Inlines.}
@cindex @option{-gnatwP} (@command{gcc})
This switch suppresses warnings on ineffective pragma Inlines. If the
inlining mechanism cannot inline a call, it will simply ignore the
request silently.

@item -gnatw.p
@emph{Activate warnings on parameter ordering.}
@cindex @option{-gnatw.p} (@command{gcc})
@cindex Parameter order, warnings
This switch activates warnings for cases of suspicious parameter
ordering when the list of arguments are all simple identifiers that
match the names of the formals, but are in a different order. The
warning is suppressed if any use of named parameter notation is used,
so this is the appropriate way to suppress a false positive (and
serves to emphasize that the "misordering" is deliberate). The
default is that such warnings are not given.

@item -gnatw.P
@emph{Suppress warnings on parameter ordering.}
@cindex @option{-gnatw.P} (@command{gcc})
This switch suppresses warnings on cases of suspicious parameter
ordering.

@item -gnatwq
@emph{Activate warnings on questionable missing parentheses.}
@cindex @option{-gnatwq} (@command{gcc})
@cindex Parentheses, warnings
This switch activates warnings for cases where parentheses are not used and
the result is potential ambiguity from a readers point of view. For example
(not a > b) when a and b are modular means ((not a) > b) and very likely the
programmer intended (not (a > b)). Similarly (-x mod 5) means (-(x mod 5)) and
quite likely ((-x) mod 5) was intended. In such situations it seems best to
follow the rule of always parenthesizing to make the association clear, and
this warning switch warns if such parentheses are not present. The default
is that these warnings are given.

@item -gnatwQ
@emph{Suppress warnings on questionable missing parentheses.}
@cindex @option{-gnatwQ} (@command{gcc})
This switch suppresses warnings for cases where the association is not
clear and the use of parentheses is preferred.

@item -gnatwr
@emph{Activate warnings on redundant constructs.}
@cindex @option{-gnatwr} (@command{gcc})
This switch activates warnings for redundant constructs. The following
is the current list of constructs regarded as redundant:

@itemize @bullet
@item
Assignment of an item to itself.
@item
Type conversion that converts an expression to its own type.
@item
Use of the attribute @code{Base} where @code{typ'Base} is the same
as @code{typ}.
@item
Use of pragma @code{Pack} when all components are placed by a record
representation clause.
@item
Exception handler containing only a reraise statement (raise with no
operand) which has no effect.
@item
Use of the operator abs on an operand that is known at compile time
to be non-negative
@item
Comparison of boolean expressions to an explicit True value.
@end itemize

The default is that warnings for redundant constructs are not given.

@item -gnatwR
@emph{Suppress warnings on redundant constructs.}
@cindex @option{-gnatwR} (@command{gcc})
This switch suppresses warnings for redundant constructs.

@item -gnatw.r
@emph{Activate warnings for object renaming function.}
@cindex @option{-gnatw.r} (@command{gcc})
This switch activates warnings for an object renaming that renames a
function call, which is equivalent to a constant declaration (as
opposed to renaming the function itself).  The default is that these
warnings are given.

@item -gnatw.R
@emph{Suppress warnings for object renaming function.}
@cindex @option{-gnatwT} (@command{gcc})
This switch suppresses warnings for object renaming function.

@item -gnatws
@emph{Suppress all warnings.}
@cindex @option{-gnatws} (@command{gcc})
This switch completely suppresses the
output of all warning messages from the GNAT front end, including
both warnings that can be controlled by switches described in this
section, and those that are normally given unconditionally. The
effect of this suppress action can only be cancelled by a subsequent
use of the switch @option{-gnatwn}.

Note that switch @option{-gnatws} does not suppress
warnings from the @command{gcc} back end.
To suppress these back end warnings as well, use the switch @option{-w}
in addition to @option{-gnatws}. Also this switch has no effect on the
handling of style check messages.

@item -gnatw.s
@emph{Activate warnings on overridden size clauses.}
@cindex @option{-gnatw.s} (@command{gcc})
@cindex Record Representation (component sizes)
This switch activates warnings on component clauses in record
representation clauses where the length given overrides that
specified by an explicit size clause for the component type. A
warning is similarly given in the array case if a specified
component size overrides an explicit size clause for the array
component type.

@item -gnatw.S
@emph{Suppress warnings on overridden size clauses.}
@cindex @option{-gnatw.S} (@command{gcc})
This switch suppresses warnings on component clauses in record
representation clauses that override size clauses, and similar
warnings when an array component size overrides a size clause.

@item -gnatwt
@emph{Activate warnings for tracking of deleted conditional code.}
@cindex @option{-gnatwt} (@command{gcc})
@cindex Deactivated code, warnings
@cindex Deleted code, warnings
This switch activates warnings for tracking of code in conditionals (IF and
CASE statements) that is detected to be dead code which cannot be executed, and
which is removed by the front end. This warning is off by default. This may be
useful for detecting deactivated code in certified applications.

@item -gnatwT
@emph{Suppress warnings for tracking of deleted conditional code.}
@cindex @option{-gnatwT} (@command{gcc})
This switch suppresses warnings for tracking of deleted conditional code.

@item -gnatw.t
@emph{Activate warnings on suspicious contracts.}
@cindex @option{-gnatw.t} (@command{gcc})
This switch activates warnings on suspicious postconditions (whether a
pragma @code{Postcondition} or a @code{Post} aspect in Ada 2012)
and suspicious contract cases (pragma @code{Contract_Cases}). A
function postcondition or contract case is suspicious when no postcondition
or contract case for this function mentions the result of the function.
A procedure postcondition or contract case is suspicious when it only
refers to the pre-state of the procedure, because in that case it should
rather be expressed as a precondition. The default is that such warnings
are not generated.

@item -gnatw.T
@emph{Suppress warnings on suspicious contracts.}
@cindex @option{-gnatw.T} (@command{gcc})
This switch suppresses warnings on suspicious postconditions.

@item -gnatwu
@emph{Activate warnings on unused entities.}
@cindex @option{-gnatwu} (@command{gcc})
This switch activates warnings to be generated for entities that
are declared but not referenced, and for units that are @code{with}'ed
and not
referenced. In the case of packages, a warning is also generated if
no entities in the package are referenced. This means that if a with'ed
package is referenced but the only references are in @code{use}
clauses or @code{renames}
declarations, a warning is still generated. A warning is also generated
for a generic package that is @code{with}'ed but never instantiated.
In the case where a package or subprogram body is compiled, and there
is a @code{with} on the corresponding spec
that is only referenced in the body,
a warning is also generated, noting that the
@code{with} can be moved to the body. The default is that
such warnings are not generated.
This switch also activates warnings on unreferenced formals
(it includes the effect of @option{-gnatwf}).

@item -gnatwU
@emph{Suppress warnings on unused entities.}
@cindex @option{-gnatwU} (@command{gcc})
This switch suppresses warnings for unused entities and packages.
It also turns off warnings on unreferenced formals (and thus includes
the effect of @option{-gnatwF}).

@item -gnatw.u
@emph{Activate warnings on unordered enumeration types.}
@cindex @option{-gnatw.u} (@command{gcc})
This switch causes enumeration types to be considered as conceptually
unordered, unless an explicit pragma @code{Ordered} is given for the type.
The effect is to generate warnings in clients that use explicit comparisons
or subranges, since these constructs both treat objects of the type as
ordered. (A @emph{client} is defined as a unit that is other than the unit in
which the type is declared, or its body or subunits.) Please refer to
the description of pragma @code{Ordered} in the
@cite{@value{EDITION} Reference Manual} for further details.
The default is that such warnings are not generated.

@item -gnatw.U
@emph{Deactivate warnings on unordered enumeration types.}
@cindex @option{-gnatw.U} (@command{gcc})
This switch causes all enumeration types to be considered as ordered, so
that no warnings are given for comparisons or subranges for any type.

@item -gnatwv
@emph{Activate warnings on unassigned variables.}
@cindex @option{-gnatwv} (@command{gcc})
@cindex Unassigned variable warnings
This switch activates warnings for access to variables which
may not be properly initialized. The default is that
such warnings are generated.

@item -gnatwV
@emph{Suppress warnings on unassigned variables.}
@cindex @option{-gnatwV} (@command{gcc})
This switch suppresses warnings for access to variables which
may not be properly initialized.
For variables of a composite type, the warning can also be suppressed in
Ada 2005 by using a default initialization with a box. For example, if
Table is an array of records whose components are only partially uninitialized,
then the following code:

@smallexample @c ada
   Tab : Table := (@b{others} => <>);
@end smallexample

will suppress warnings on subsequent statements that access components
of variable Tab.

@item -gnatw.v
@emph{Activate info messages for non-default bit order.}
@cindex @option{-gnatw.v} (@command{gcc})
@cindex bit order warnings
This switch activates messages (labeled "info", they are not warnings,
just informational messages) about the effects of non-default bit-order
on records to which a component clause is applied. The effect of specifying
non-default bit ordering is a bit subtle (and changed with Ada 2005), so
these messages, which are given by default, are useful in understanding the
exact consequences of using this feature.

@item -gnatw.V
@emph{Suppress info messages for non-default bit order.}
@cindex @option{-gnatw.V} (@command{gcc})
This switch suppresses information messages for the effects of specifying
non-default bit order on record components with component clauses.

@item -gnatww
@emph{Activate warnings on wrong low bound assumption.}
@cindex @option{-gnatww} (@command{gcc})
@cindex String indexing warnings
This switch activates warnings for indexing an unconstrained string parameter
with a literal or S'Length. This is a case where the code is assuming that the
low bound is one, which is in general not true (for example when a slice is
passed). The default is that such warnings are generated.

@item -gnatwW
@emph{Suppress warnings on wrong low bound assumption.}
@cindex @option{-gnatwW} (@command{gcc})
This switch suppresses warnings for indexing an unconstrained string parameter
with a literal or S'Length. Note that this warning can also be suppressed
in a particular case by adding an
assertion that the lower bound is 1,
as shown in the following example.

@smallexample @c ada
   @b{procedure} K (S : String) @b{is}
      @b{pragma} Assert (S'First = 1);
      @dots{}
@end smallexample

@item -gnatw.w
@emph{Activate warnings on Warnings Off pragmas}
@cindex @option{-gnatw.w} (@command{gcc})
@cindex Warnings Off control
This switch activates warnings for use of @code{pragma Warnings (Off, entity)}
where either the pragma is entirely useless (because it suppresses no
warnings), or it could be replaced by @code{pragma Unreferenced} or
@code{pragma Unmodified}.
Also activates warnings for the case of
Warnings (Off, String), where either there is no matching
Warnings (On, String), or the Warnings (Off) did not suppress any warning.
The default is that these warnings are not given.

@item -gnatw.W
@emph{Suppress warnings on unnecessary Warnings Off pragmas}
@cindex @option{-gnatw.W} (@command{gcc})
This switch suppresses warnings for use of @code{pragma Warnings (Off, ...)}.

@item -gnatwx
@emph{Activate warnings on Export/Import pragmas.}
@cindex @option{-gnatwx} (@command{gcc})
@cindex Export/Import pragma warnings
This switch activates warnings on Export/Import pragmas when
the compiler detects a possible conflict between the Ada and
foreign language calling sequences. For example, the use of
default parameters in a convention C procedure is dubious
because the C compiler cannot supply the proper default, so
a warning is issued. The default is that such warnings are
generated.

@item -gnatwX
@emph{Suppress warnings on Export/Import pragmas.}
@cindex @option{-gnatwX} (@command{gcc})
This switch suppresses warnings on Export/Import pragmas.
The sense of this is that you are telling the compiler that
you know what you are doing in writing the pragma, and it
should not complain at you.

@item -gnatw.x
@emph{Activate warnings for No_Exception_Propagation mode.}
@cindex @option{-gnatwm} (@command{gcc})
This switch activates warnings for exception usage when pragma Restrictions
(No_Exception_Propagation) is in effect. Warnings are given for implicit or
explicit exception raises which are not covered by a local handler, and for
exception handlers which do not cover a local raise. The default is that these
warnings are not given.

@item -gnatw.X
@emph{Disable warnings for No_Exception_Propagation mode.}
This switch disables warnings for exception usage when pragma Restrictions
(No_Exception_Propagation) is in effect.

@item -gnatwy
@emph{Activate warnings for Ada compatibility issues.}
@cindex @option{-gnatwy} (@command{gcc})
@cindex Ada compatibility issues warnings
For the most part, newer versions of Ada are upwards compatible
with older versions. For example, Ada 2005 programs will almost
always work when compiled as Ada 2012.
However there are some exceptions (for example the fact that
@code{some} is now a reserved word in Ada 2012). This
switch activates several warnings to help in identifying
and correcting such incompatibilities. The default is that
these warnings are generated. Note that at one point Ada 2005
was called Ada 0Y, hence the choice of character.

@item -gnatwY
@emph{Disable warnings for Ada compatibility issues.}
@cindex @option{-gnatwY} (@command{gcc})
@cindex Ada compatibility issues warnings
This switch suppresses the warnings intended to help in identifying
incompatibilities between Ada language versions.

@item -gnatw.y
@emph{Activate information messages for why package spec needs body}
@cindex @option{-gnatw.y} (@command{gcc})
@cindex Package spec needing body
There are a number of cases in which a package spec needs a body.
For example, the use of pragma Elaborate_Body, or the declaration
of a procedure specification requiring a completion. This switch
causes information messages to be output showing why a package
specification requires a body. This can be useful in the case of
a large package specification which is unexpectedly requiring a
body. The default is that such information messages are not output.

@item -gnatw.Y
@emph{Disable information messages for why package spec needs body}
@cindex @option{-gnatw.Y} (@command{gcc})
@cindex No information messages for why package spec needs body
This switch suppresses the output of information messages showing why
a package specification needs a body.

@item -gnatwz
@emph{Activate warnings on unchecked conversions.}
@cindex @option{-gnatwz} (@command{gcc})
@cindex Unchecked_Conversion warnings
This switch activates warnings for unchecked conversions
where the types are known at compile time to have different
sizes. The default
is that such warnings are generated. Warnings are also
generated for subprogram pointers with different conventions,
and, on VMS only, for data pointers with different conventions.

@item -gnatwZ
@emph{Suppress warnings on unchecked conversions.}
@cindex @option{-gnatwZ} (@command{gcc})
This switch suppresses warnings for unchecked conversions
where the types are known at compile time to have different
sizes or conventions.

@item -gnatw.z
@emph{Activate warnings for size not a multiple of alignment.}
@cindex @option{-gnatw.z} (@command{gcc})
@cindex Size/Alignment warnings
This switch activates warnings for cases of record types with
specified @code{Size} and @code{Alignment} attributes where the
size is not a multiple of the alignment, resulting in an object
size that is greater than the specified size. The default
is that such warnings are generated.

@item -gnatw.Z
@emph{Suppress warnings for size not a multiple of alignment.}
@cindex @option{-gnatw.Z} (@command{gcc})
@cindex Size/Alignment warnings
This switch suppresses warnings for cases of record types with
specified @code{Size} and @code{Alignment} attributes where the
size is not a multiple of the alignment, resulting in an object
size that is greater than the specified size.
The warning can also be
suppressed by giving an explicit @code{Object_Size} value.

@item -Wunused
@cindex @option{-Wunused}
The warnings controlled by the @option{-gnatw} switch are generated by
the front end of the compiler. The @option{GCC} back end can provide
additional warnings and they are controlled by the @option{-W} switch.
For example, @option{-Wunused} activates back end
warnings for entities that are declared but not referenced.

@item -Wuninitialized
@cindex @option{-Wuninitialized}
Similarly, @option{-Wuninitialized} activates
the back end warning for uninitialized variables. This switch must be
used in conjunction with an optimization level greater than zero.

@item -Wstack-usage=@var{len}
@cindex @option{-Wstack-usage}
Warn if the stack usage of a subprogram might be larger than @var{len} bytes.
See @ref{Static Stack Usage Analysis} for details.

@item -Wall
@cindex @option{-Wall}
This switch enables most warnings from the @option{GCC} back end.
The code generator detects a number of warning situations that are missed
by the @option{GNAT} front end, and this switch can be used to activate them.
The use of this switch also sets the default front end warning mode to
@option{-gnatwa}, that is, most front end warnings activated as well.

@item -w
@cindex @option{-w}
Conversely, this switch suppresses warnings from the @option{GCC} back end.
The use of this switch also sets the default front end warning mode to
@option{-gnatws}, that is, front end warnings suppressed as well.

@item -Werror
@cindex @option{-Werror}
This switch causes warnings from the @option{GCC} back end to be treated as
errors.  The warning string still appears, but the warning messages are
counted as errors, and prevent the generation of an object file.

@end table

@noindent
A string of warning parameters can be used in the same parameter. For example:

@smallexample
-gnatwaGe
@end smallexample

@noindent
will turn on all optional warnings except for unrecognized pragma warnings,
and also specify that warnings should be treated as errors.

When no switch @option{-gnatw} is used, this is equivalent to:

@table @option
@c !sort!
@item -gnatw.a
@item -gnatwB
@item -gnatw.b
@item -gnatwC
@item -gnatw.C
@item -gnatwD
@item -gnatwF
@item -gnatwg
@item -gnatwH
@item -gnatwi
@item -gnatw.I
@item -gnatwJ
@item -gnatwK
@item -gnatwL
@item -gnatw.L
@item -gnatwM
@item -gnatw.m
@item -gnatwn
@item -gnatwo
@item -gnatw.O
@item -gnatwP
@item -gnatw.P
@item -gnatwq
@item -gnatwR
@item -gnatw.R
@item -gnatw.S
@item -gnatwT
@item -gnatw.T
@item -gnatwU
@item -gnatwv
@item -gnatww
@item -gnatw.W
@item -gnatwx
@item -gnatw.X
@item -gnatwy
@item -gnatwz

@end table

@node Debugging and Assertion Control
@subsection Debugging and Assertion Control

@table @option
@item -gnata
@cindex @option{-gnata} (@command{gcc})
@findex Assert
@findex Debug
@cindex Assertions

@noindent
The pragmas @code{Assert} and @code{Debug} normally have no effect and
are ignored. This switch, where @samp{a} stands for assert, causes
@code{Assert} and @code{Debug} pragmas to be activated.

The pragmas have the form:

@smallexample
@cartouche
   @b{pragma} Assert (@var{Boolean-expression} @r{[},
                      @var{static-string-expression}@r{]})
   @b{pragma} Debug (@var{procedure call})
@end cartouche
@end smallexample

@noindent
The @code{Assert} pragma causes @var{Boolean-expression} to be tested.
If the result is @code{True}, the pragma has no effect (other than
possible side effects from evaluating the expression). If the result is
@code{False}, the exception @code{Assert_Failure} declared in the package
@code{System.Assertions} is
raised (passing @var{static-string-expression}, if present, as the
message associated with the exception). If no string expression is
given the default is a string giving the file name and line number
of the pragma.

The @code{Debug} pragma causes @var{procedure} to be called. Note that
@code{pragma Debug} may appear within a declaration sequence, allowing
debugging procedures to be called between declarations.

@end table

@node Validity Checking
@subsection Validity Checking
@findex Validity Checking

@noindent
The Ada Reference Manual defines the concept of invalid values (see
RM 13.9.1). The primary source of invalid values is uninitialized
variables. A scalar variable that is left uninitialized may contain
an invalid value; the concept of invalid does not apply to access or
composite types.

It is an error to read an invalid value, but the RM does not require
run-time checks to detect such errors, except for some minimal
checking to prevent erroneous execution (i.e. unpredictable
behavior). This corresponds to the @option{-gnatVd} switch below,
which is the default. For example, by default, if the expression of a
case statement is invalid, it will raise Constraint_Error rather than
causing a wild jump, and if an array index on the left-hand side of an
assignment is invalid, it will raise Constraint_Error rather than
overwriting an arbitrary memory location.

The @option{-gnatVa} may be used to enable additional validity checks,
which are not required by the RM. These checks are often very
expensive (which is why the RM does not require them). These checks
are useful in tracking down uninitialized variables, but they are
not usually recommended for production builds, and in particular
we do not recommend using these extra validity checking options in
combination with optimization, since this can confuse the optimizer.
If performance is a consideration, leading to the need to optimize,
then the validity checking options should not be used.

The other @option{-gnatV@var{x}} switches below allow finer-grained
control; you can enable whichever validity checks you desire. However,
for most debugging purposes, @option{-gnatVa} is sufficient, and the
default @option{-gnatVd} (i.e. standard Ada behavior) is usually
sufficient for non-debugging use.

The @option{-gnatB} switch tells the compiler to assume that all
values are valid (that is, within their declared subtype range)
except in the context of a use of the Valid attribute. This means
the compiler can generate more efficient code, since the range
of values is better known at compile time. However, an uninitialized
variable can cause wild jumps and memory corruption in this mode.

The @option{-gnatV@var{x}} switch allows control over the validity
checking mode as described below.
The @code{x} argument is a string of letters that
indicate validity checks that are performed or not performed in addition
to the default checks required by Ada as described above.

@table @option
@c !sort!
@item -gnatVa
@emph{All validity checks.}
@cindex @option{-gnatVa} (@command{gcc})
All validity checks are turned on.
That is, @option{-gnatVa} is
equivalent to @option{gnatVcdfimorst}.

@item -gnatVc
@emph{Validity checks for copies.}
@cindex @option{-gnatVc} (@command{gcc})
The right hand side of assignments, and the initializing values of
object declarations are validity checked.

@item -gnatVd
@emph{Default (RM) validity checks.}
@cindex @option{-gnatVd} (@command{gcc})
Some validity checks are done by default following normal Ada semantics
(RM 13.9.1 (9-11)).
A check is done in case statements that the expression is within the range
of the subtype. If it is not, Constraint_Error is raised.
For assignments to array components, a check is done that the expression used
as index is within the range. If it is not, Constraint_Error is raised.
Both these validity checks may be turned off using switch @option{-gnatVD}.
They are turned on by default. If @option{-gnatVD} is specified, a subsequent
switch @option{-gnatVd} will leave the checks turned on.
Switch @option{-gnatVD} should be used only if you are sure that all such
expressions have valid values. If you use this switch and invalid values
are present, then the program is erroneous, and wild jumps or memory
overwriting may occur.

@item -gnatVe
@emph{Validity checks for elementary components.}
@cindex @option{-gnatVe} (@command{gcc})
In the absence of this switch, assignments to record or array components are
not validity checked, even if validity checks for assignments generally
(@option{-gnatVc}) are turned on. In Ada, assignment of composite values do not
require valid data, but assignment of individual components does. So for
example, there is a difference between copying the elements of an array with a
slice assignment, compared to assigning element by element in a loop. This
switch allows you to turn off validity checking for components, even when they
are assigned component by component.

@item -gnatVf
@emph{Validity checks for floating-point values.}
@cindex @option{-gnatVf} (@command{gcc})
In the absence of this switch, validity checking occurs only for discrete
values. If @option{-gnatVf} is specified, then validity checking also applies
for floating-point values, and NaNs and infinities are considered invalid,
as well as out of range values for constrained types. Note that this means
that standard IEEE infinity mode is not allowed. The exact contexts
in which floating-point values are checked depends on the setting of other
options. For example,
@option{-gnatVif} or
@option{-gnatVfi}
(the order does not matter) specifies that floating-point parameters of mode
@code{in} should be validity checked.

@item -gnatVi
@emph{Validity checks for @code{in} mode parameters}
@cindex @option{-gnatVi} (@command{gcc})
Arguments for parameters of mode @code{in} are validity checked in function
and procedure calls at the point of call.

@item -gnatVm
@emph{Validity checks for @code{in out} mode parameters.}
@cindex @option{-gnatVm} (@command{gcc})
Arguments for parameters of mode @code{in out} are validity checked in
procedure calls at the point of call. The @code{'m'} here stands for
modify, since this concerns parameters that can be modified by the call.
Note that there is no specific option to test @code{out} parameters,
but any reference within the subprogram will be tested in the usual
manner, and if an invalid value is copied back, any reference to it
will be subject to validity checking.

@item -gnatVn
@emph{No validity checks.}
@cindex @option{-gnatVn} (@command{gcc})
This switch turns off all validity checking, including the default checking
for case statements and left hand side subscripts. Note that the use of
the switch @option{-gnatp} suppresses all run-time checks, including
validity checks, and thus implies @option{-gnatVn}. When this switch
is used, it cancels any other @option{-gnatV} previously issued.

@item -gnatVo
@emph{Validity checks for operator and attribute operands.}
@cindex @option{-gnatVo} (@command{gcc})
Arguments for predefined operators and attributes are validity checked.
This includes all operators in package @code{Standard},
the shift operators defined as intrinsic in package @code{Interfaces}
and operands for attributes such as @code{Pos}. Checks are also made
on individual component values for composite comparisons, and on the
expressions in type conversions and qualified expressions. Checks are
also made on explicit ranges using @samp{..} (e.g.@: slices, loops etc).

@item -gnatVp
@emph{Validity checks for parameters.}
@cindex @option{-gnatVp} (@command{gcc})
This controls the treatment of parameters within a subprogram (as opposed
to @option{-gnatVi} and @option{-gnatVm} which control validity testing
of parameters on a call. If either of these call options is used, then
normally an assumption is made within a subprogram that the input arguments
have been validity checking at the point of call, and do not need checking
again within a subprogram). If @option{-gnatVp} is set, then this assumption
is not made, and parameters are not assumed to be valid, so their validity
will be checked (or rechecked) within the subprogram.

@item -gnatVr
@emph{Validity checks for function returns.}
@cindex @option{-gnatVr} (@command{gcc})
The expression in @code{return} statements in functions is validity
checked.

@item -gnatVs
@emph{Validity checks for subscripts.}
@cindex @option{-gnatVs} (@command{gcc})
All subscripts expressions are checked for validity, whether they appear
on the right side or left side (in default mode only left side subscripts
are validity checked).

@item -gnatVt
@emph{Validity checks for tests.}
@cindex @option{-gnatVt} (@command{gcc})
Expressions used as conditions in @code{if}, @code{while} or @code{exit}
statements are checked, as well as guard expressions in entry calls.

@end table

@noindent
The @option{-gnatV} switch may be followed by
a string of letters
to turn on a series of validity checking options.
For example,
@option{-gnatVcr}
specifies that in addition to the default validity checking, copies and
function return expressions are to be validity checked.
In order to make it easier
to specify the desired combination of effects,
the upper case letters @code{CDFIMORST} may
be used to turn off the corresponding lower case option.
Thus
@option{-gnatVaM}
turns on all validity checking options except for
checking of @code{@b{in out}} procedure arguments.

The specification of additional validity checking generates extra code (and
in the case of @option{-gnatVa} the code expansion can be substantial).
However, these additional checks can be very useful in detecting
uninitialized variables, incorrect use of unchecked conversion, and other
errors leading to invalid values. The use of pragma @code{Initialize_Scalars}
is useful in conjunction with the extra validity checking, since this
ensures that wherever possible uninitialized variables have invalid values.

See also the pragma @code{Validity_Checks} which allows modification of
the validity checking mode at the program source level, and also allows for
temporary disabling of validity checks.

@node Style Checking
@subsection Style Checking
@findex Style checking

@noindent
The @option{-gnatyx} switch
@cindex @option{-gnaty} (@command{gcc})
causes the compiler to
enforce specified style rules. A limited set of style rules has been used
in writing the GNAT sources themselves. This switch allows user programs
to activate all or some of these checks. If the source program fails a
specified style check, an appropriate message is given, preceded by
the character sequence ``(style)''. This message does not prevent
successful compilation (unless the @option{-gnatwe} switch is used).

Note that this is by no means intended to be a general facility for
checking arbitrary coding standards. It is simply an embedding of the
style rules we have chosen for the GNAT sources. If you are starting
a project which does not have established style standards, you may
find it useful to adopt the entire set of GNAT coding standards, or
some subset of them.
@ifclear FSFEDITION
If you already have an established set of coding
standards, then the selected style checking options may
indeed correspond to choices you have made, but for general checking
of an existing set of coding rules, you should look to the gnatcheck
tool, which is designed for that purpose.
@end ifclear

The string @var{x} is a sequence of letters or digits
indicating the particular style
checks to be performed. The following checks are defined:

@table @option
@c !sort!
@item 0-9
@emph{Specify indentation level.}
If a digit from 1-9 appears
in the string after @option{-gnaty}
then proper indentation is checked, with the digit indicating the
indentation level required. A value of zero turns off this style check.
The general style of required indentation is as specified by
the examples in the Ada Reference Manual. Full line comments must be
aligned with the @code{--} starting on a column that is a multiple of
the alignment level, or they may be aligned the same way as the following
non-blank line (this is useful when full line comments appear in the middle
of a statement, or they may be aligned with the source line on the previous
non-blank line.

@item a
@emph{Check attribute casing.}
Attribute names, including the case of keywords such as @code{digits}
used as attributes names, must be written in mixed case, that is, the
initial letter and any letter following an underscore must be uppercase.
All other letters must be lowercase.

@item A
@emph{Use of array index numbers in array attributes.}
When using the array attributes First, Last, Range,
or Length, the index number must be omitted for one-dimensional arrays
and is required for multi-dimensional arrays.

@item b
@emph{Blanks not allowed at statement end.}
Trailing blanks are not allowed at the end of statements. The purpose of this
rule, together with h (no horizontal tabs), is to enforce a canonical format
for the use of blanks to separate source tokens.

@item B
@emph{Check Boolean operators.}
The use of AND/OR operators is not permitted except in the cases of modular
operands, array operands, and simple stand-alone boolean variables or
boolean constants. In all other cases @code{and then}/@code{or else} are
required.

@item c
@emph{Check comments, double space.}
Comments must meet the following set of rules:

@itemize @bullet

@item
The ``@code{--}'' that starts the column must either start in column one,
or else at least one blank must precede this sequence.

@item
Comments that follow other tokens on a line must have at least one blank
following the ``@code{--}'' at the start of the comment.

@item
Full line comments must have at least two blanks following the
``@code{--}'' that starts the comment, with the following exceptions.

@item
A line consisting only of the ``@code{--}'' characters, possibly preceded
by blanks is permitted.

@item
A comment starting with ``@code{--x}'' where @code{x} is a special character
is permitted.
This allows proper processing of the output generated by specialized tools
including @command{gnatprep} (where ``@code{--!}'' is used) and the SPARK
annotation
language (where ``@code{--#}'' is used). For the purposes of this rule, a
special character is defined as being in one of the ASCII ranges
@code{16#21#@dots{}16#2F#} or @code{16#3A#@dots{}16#3F#}.
Note that this usage is not permitted
in GNAT implementation units (i.e., when @option{-gnatg} is used).

@item
A line consisting entirely of minus signs, possibly preceded by blanks, is
permitted. This allows the construction of box comments where lines of minus
signs are used to form the top and bottom of the box.

@item
A comment that starts and ends with ``@code{--}'' is permitted as long as at
least one blank follows the initial ``@code{--}''. Together with the preceding
rule, this allows the construction of box comments, as shown in the following
example:
@smallexample
---------------------------
-- This is a box comment --
-- with two text lines.  --
---------------------------
@end smallexample
@end itemize

@item C
@emph{Check comments, single space.}
This is identical to @code{c} except that only one space
is required following the @code{--} of a comment instead of two.

@item d
@emph{Check no DOS line terminators present.}
All lines must be terminated by a single ASCII.LF
character (in particular the DOS line terminator sequence CR/LF is not
allowed).

@item e
@emph{Check end/exit labels.}
Optional labels on @code{end} statements ending subprograms and on
@code{exit} statements exiting named loops, are required to be present.

@item f
@emph{No form feeds or vertical tabs.}
Neither form feeds nor vertical tab characters are permitted
in the source text.

@item g
@emph{GNAT style mode.}
The set of style check switches is set to match that used by the GNAT sources.
This may be useful when developing code that is eventually intended to be
incorporated into GNAT. Currently this is equivalent to @option{-gnatwydISux})
but additional style switches may be added to this set in the future without
advance notice.

@item h
@emph{No horizontal tabs.}
Horizontal tab characters are not permitted in the source text.
Together with the b (no blanks at end of line) check, this
enforces a canonical form for the use of blanks to separate
source tokens.

@item i
@emph{Check if-then layout.}
The keyword @code{then} must appear either on the same
line as corresponding @code{if}, or on a line on its own, lined
up under the @code{if}.

@item I
@emph{check mode IN keywords.}
Mode @code{in} (the default mode) is not
allowed to be given explicitly. @code{in out} is fine,
but not @code{in} on its own.

@item k
@emph{Check keyword casing.}
All keywords must be in lower case (with the exception of keywords
such as @code{digits} used as attribute names to which this check
does not apply).

@item l
@emph{Check layout.}
Layout of statement and declaration constructs must follow the
recommendations in the Ada Reference Manual, as indicated by the
form of the syntax rules. For example an @code{else} keyword must
be lined up with the corresponding @code{if} keyword.

There are two respects in which the style rule enforced by this check
option are more liberal than those in the Ada Reference Manual. First
in the case of record declarations, it is permissible to put the
@code{record} keyword on the same line as the @code{type} keyword, and
then the @code{end} in @code{end record} must line up under @code{type}.
This is also permitted when the type declaration is split on two lines.
For example, any of the following three layouts is acceptable:

@smallexample @c ada
@cartouche
@b{type} q @b{is} @b{record}
   a : integer;
   b : integer;
@b{end} @b{record};

@b{type} q @b{is}
   @b{record}
      a : integer;
      b : integer;
   @b{end} @b{record};

@b{type} q @b{is}
   @b{record}
      a : integer;
      b : integer;
@b{end} @b{record};

@end cartouche
@end smallexample

@noindent
Second, in the case of a block statement, a permitted alternative
is to put the block label on the same line as the @code{declare} or
@code{begin} keyword, and then line the @code{end} keyword up under
the block label. For example both the following are permitted:

@smallexample @c ada
@cartouche
Block : @b{declare}
   A : Integer := 3;
@b{begin}
   Proc (A, A);
@b{end} Block;

Block :
   @b{declare}
      A : Integer := 3;
   @b{begin}
      Proc (A, A);
   @b{end} Block;
@end cartouche
@end smallexample

@noindent
The same alternative format is allowed for loops. For example, both of
the following are permitted:

@smallexample @c ada
@cartouche
Clear : @b{while} J < 10 @b{loop}
   A (J) := 0;
@b{end} @b{loop} Clear;

Clear :
   @b{while} J < 10 @b{loop}
      A (J) := 0;
   @b{end} @b{loop} Clear;
@end cartouche
@end smallexample

@item Lnnn
@emph{Set maximum nesting level.}
The maximum level of nesting of constructs (including subprograms, loops,
blocks, packages, and conditionals) may not exceed the given value
@option{nnn}. A value of zero disconnects this style check.

@item m
@emph{Check maximum line length.}
The length of source lines must not exceed 79 characters, including
any trailing blanks. The value of 79 allows convenient display on an
80 character wide device or window, allowing for possible special
treatment of 80 character lines. Note that this count is of
characters in the source text. This means that a tab character counts
as one character in this count and a wide character sequence counts as
a single character (however many bytes are needed in the encoding).

@item Mnnn
@emph{Set maximum line length.}
The length of lines must not exceed the
given value @option{nnn}. The maximum value that can be specified is 32767.
If neither style option for setting the line length is used, then the
default is 255. This also controls the maximum length of lexical elements,
where the only restriction is that they must fit on a single line.

@item n
@emph{Check casing of entities in Standard.}
Any identifier from Standard must be cased
to match the presentation in the Ada Reference Manual (for example,
@code{Integer} and @code{ASCII.NUL}).

@item N
@emph{Turn off all style checks.}
All style check options are turned off.

@item o
@emph{Check order of subprogram bodies.}
All subprogram bodies in a given scope
(e.g.@: a package body) must be in alphabetical order. The ordering
rule uses normal Ada rules for comparing strings, ignoring casing
of letters, except that if there is a trailing numeric suffix, then
the value of this suffix is used in the ordering (e.g.@: Junk2 comes
before Junk10).

@item O
@emph{Check that overriding subprograms are explicitly marked as such.}
The declaration of a primitive operation of a type extension that overrides
an inherited operation must carry an overriding indicator.

@item p
@emph{Check pragma casing.}
Pragma names must be written in mixed case, that is, the
initial letter and any letter following an underscore must be uppercase.
All other letters must be lowercase. An exception is that SPARK_Mode is
allowed as an alternative for Spark_Mode.

@item r
@emph{Check references.}
All identifier references must be cased in the same way as the
corresponding declaration. No specific casing style is imposed on
identifiers. The only requirement is for consistency of references
with declarations.

@item s
@emph{Check separate specs.}
Separate declarations (``specs'') are required for subprograms (a
body is not allowed to serve as its own declaration). The only
exception is that parameterless library level procedures are
not required to have a separate declaration. This exception covers
the most frequent form of main program procedures.

@item S
@emph{Check no statements after @code{then}/@code{else}.}
No statements are allowed
on the same line as a @code{then} or @code{else} keyword following the
keyword in an @code{if} statement. @code{or else} and @code{and then} are not
affected, and a special exception allows a pragma to appear after @code{else}.

@item t
@emph{Check token spacing.}
The following token spacing rules are enforced:

@itemize @bullet

@item
The keywords @code{abs} and @code{not} must be followed by a space.

@item
The token @code{=>} must be surrounded by spaces.

@item
The token @code{<>} must be preceded by a space or a left parenthesis.

@item
Binary operators other than @code{**} must be surrounded by spaces.
There is no restriction on the layout of the @code{**} binary operator.

@item
Colon must be surrounded by spaces.

@item
Colon-equal (assignment, initialization) must be surrounded by spaces.

@item
Comma must be the first non-blank character on the line, or be
immediately preceded by a non-blank character, and must be followed
by a space.

@item
If the token preceding a left parenthesis ends with a letter or digit, then
a space must separate the two tokens.

@item
if the token following a right parenthesis starts with a letter or digit, then
a space must separate the two tokens.

@item
A right parenthesis must either be the first non-blank character on
a line, or it must be preceded by a non-blank character.

@item
A semicolon must not be preceded by a space, and must not be followed by
a non-blank character.

@item
A unary plus or minus may not be followed by a space.

@item
A vertical bar must be surrounded by spaces.
@end itemize

Exactly one blank (and no other white space) must appear between
a @code{not} token and a following @code{in} token.

@item u
@emph{Check unnecessary blank lines.}
Unnecessary blank lines are not allowed. A blank line is considered
unnecessary if it appears at the end of the file, or if more than
one blank line occurs in sequence.

@item x
@emph{Check extra parentheses.}
Unnecessary extra level of parentheses (C-style) are not allowed
around conditions in @code{if} statements, @code{while} statements and
@code{exit} statements.

@item y
@emph{Set all standard style check options}
This is equivalent to @code{gnaty3aAbcefhiklmnprst}, that is all checking
options enabled with the exception of @option{-gnatyB}, @option{-gnatyd},
@option{-gnatyI}, @option{-gnatyLnnn}, @option{-gnatyo}, @option{-gnatyO},
@option{-gnatyS}, @option{-gnatyu}, and @option{-gnatyx}.

@item -
@emph{Remove style check options}
This causes any subsequent options in the string to act as canceling the
corresponding style check option. To cancel maximum nesting level control,
use @option{L} parameter witout any integer value after that, because any
digit following @option{-} in the parameter string of the @option{-gnaty}
option will be threated as canceling indentation check. The same is true
for @option{M} parameter. @option{y} and @option{N} parameters are not
allowed after @option{-}.

@item +
This causes any subsequent options in the string to enable the corresponding
style check option. That is, it cancels the effect of a previous -,
if any.

@end table

@noindent
In the above rules, appearing in column one is always permitted, that is,
counts as meeting either a requirement for a required preceding space,
or as meeting a requirement for no preceding space.

Appearing at the end of a line is also always permitted, that is, counts
as meeting either a requirement for a following space, or as meeting
a requirement for no following space.

@noindent
If any of these style rules is violated, a message is generated giving
details on the violation. The initial characters of such messages are
always ``@code{(style)}''. Note that these messages are treated as warning
messages, so they normally do not prevent the generation of an object
file. The @option{-gnatwe} switch can be used to treat warning messages,
including style messages, as fatal errors.

The switch
@option{-gnaty} on its own (that is not
followed by any letters or digits) is equivalent
to the use of @option{-gnatyy} as described above, that is all
built-in standard style check options are enabled.


The switch
@option{-gnatyN}
clears any previously set style checks.

@node Run-Time Checks
@subsection Run-Time Checks
@cindex Division by zero
@cindex Access before elaboration
@cindex Checks, division by zero
@cindex Checks, access before elaboration
@cindex Checks, stack overflow checking

@noindent
By default, the following checks are suppressed: integer overflow
checks, stack overflow checks, and checks for access before
elaboration on subprogram calls. All other checks, including range
checks and array bounds checks, are turned on by default. The
following @command{gcc} switches refine this default behavior.

@table @option
@c !sort!
@item -gnatp
@cindex @option{-gnatp} (@command{gcc})
@cindex Suppressing checks
@cindex Checks, suppressing
@findex Suppress
This switch causes the unit to be compiled
as though @code{pragma Suppress (All_checks)}
had been present in the source. Validity checks are also eliminated (in
other words @option{-gnatp} also implies @option{-gnatVn}.
Use this switch to improve the performance
of the code at the expense of safety in the presence of invalid data or
program bugs.

Note that when checks are suppressed, the compiler is allowed, but not
required, to omit the checking code. If the run-time cost of the
checking code is zero or near-zero, the compiler will generate it even
if checks are suppressed. In particular, if the compiler can prove
that a certain check will necessarily fail, it will generate code to
do an unconditional ``raise'', even if checks are suppressed. The
compiler warns in this case. Another case in which checks may not be
eliminated is when they are embedded in certain run time routines such
as math library routines.

Of course, run-time checks are omitted whenever the compiler can prove
that they will not fail, whether or not checks are suppressed.

Note that if you suppress a check that would have failed, program
execution is erroneous, which means the behavior is totally
unpredictable. The program might crash, or print wrong answers, or
do anything else. It might even do exactly what you wanted it to do
(and then it might start failing mysteriously next week or next
year). The compiler will generate code based on the assumption that
the condition being checked is true, which can result in erroneous
execution if that assumption is wrong.

The checks subject to suppression include all the checks defined by
the Ada standard, the additional implementation defined checks
@code{Alignment_Check},
@code{Duplicated_Tag_Check}, @code{Predicate_Check}, and
@code{Validity_Check}, as well as any checks introduced using
@code{pragma Check_Name}. Note that @code{Atomic_Synchronization}
is not automatically suppressed by use of this option.

If the code depends on certain checks being active, you can use
pragma @code{Unsuppress} either as a configuration pragma or as
a local pragma to make sure that a specified check is performed
even if @option{gnatp} is specified.

The @option{-gnatp} switch has no effect if a subsequent
@option{-gnat-p} switch appears.

@item -gnat-p
@cindex @option{-gnat-p} (@command{gcc})
@cindex Suppressing checks
@cindex Checks, suppressing
@findex Suppress
This switch cancels the effect of a previous @option{gnatp} switch.

@item -gnato??
@cindex @option{-gnato??} (@command{gcc})
@cindex Overflow checks
@cindex Overflow mode
@cindex Check, overflow
This switch controls the mode used for computing intermediate
arithmetic integer operations, and also enables overflow checking.
For a full description of overflow mode and checking control, see
the ``Overflow Check Handling in GNAT'' appendix in this
User's Guide.

Overflow checks are always enabled by this switch. The argument
controls the mode, using the codes

@itemize
@item 1 = STRICT
In STRICT mode, intermediate operations are always done using the
base type, and overflow checking ensures that the result is within
the base type range.

@item 2 = MINIMIZED
In MINIMIZED mode, overflows in intermediate operations are avoided
where possible by using a larger integer type for the computation
(typically @code{Long_Long_Integer}). Overflow checking ensures that
the result fits in this larger integer type.

@item 3 = ELIMINATED
In ELIMINATED mode, overflows in intermediate operations are avoided
by using multi-precision arithmetic. In this case, overflow checking
has no effect on intermediate operations (since overflow is impossible).
@end itemize

If two digits are present after @option{-gnato} then the first digit
sets the mode for expressions outside assertions, and the second digit
sets the mode for expressions within assertions. Here assertions is used
in the technical sense (which includes for example precondition and
postcondition expressions).

If one digit is present, the corresponding mode is applicable to both
expressions within and outside assertion expressions.

If no digits are present, the default is to enable overflow checks
and set STRICT mode for both kinds of expressions. This is compatible
with the use of @option{-gnato} in previous versions of GNAT.

@findex Machine_Overflows
Note that the @option{-gnato??} switch does not affect the code generated
for any floating-point operations; it applies only to integer semantics.
For floating-point, @value{EDITION} has the @code{Machine_Overflows}
attribute set to @code{False} and the normal mode of operation is to
generate IEEE NaN and infinite values on overflow or invalid operations
(such as dividing 0.0 by 0.0).

The reason that we distinguish overflow checking from other kinds of
range constraint checking is that a failure of an overflow check, unlike
for example the failure of a range check, can result in an incorrect
value, but cannot cause random memory destruction (like an out of range
subscript), or a wild jump (from an out of range case value). Overflow
checking is also quite expensive in time and space, since in general it
requires the use of double length arithmetic.

Note again that the default is @option{-gnato00},
so overflow checking is not performed in default mode. This means that out of
the box, with the default settings, @value{EDITION} does not do all the checks
expected from the language description in the Ada Reference Manual.
If you want all constraint checks to be performed, as described in this Manual,
then you must explicitly use the @option{-gnato??}
switch either on the @command{gnatmake} or @command{gcc} command.

@item -gnatE
@cindex @option{-gnatE} (@command{gcc})
@cindex Elaboration checks
@cindex Check, elaboration
Enables dynamic checks for access-before-elaboration
on subprogram calls and generic instantiations.
Note that @option{-gnatE} is not necessary for safety, because in the
default mode, GNAT ensures statically that the checks would not fail.
For full details of the effect and use of this switch,
@xref{Compiling with gcc}.

@item -fstack-check
@cindex @option{-fstack-check} (@command{gcc})
@cindex Stack Overflow Checking
@cindex Checks, stack overflow checking
Activates stack overflow checking. For full details of the effect and use of
this switch see @ref{Stack Overflow Checking}.
@end table

@findex Unsuppress
@noindent
The setting of these switches only controls the default setting of the
checks. You may modify them using either @code{Suppress} (to remove
checks) or @code{Unsuppress} (to add back suppressed checks) pragmas in
the program source.

@node Using gcc for Syntax Checking
@subsection Using @command{gcc} for Syntax Checking
@table @option
@item -gnats
@cindex @option{-gnats} (@command{gcc})

@noindent
The @code{s} stands for ``syntax''.

Run GNAT in syntax checking only mode. For
example, the command

@smallexample
$ gcc -c -gnats x.adb
@end smallexample

@noindent
compiles file @file{x.adb} in syntax-check-only mode. You can check a
series of files in a single command
, and can use wild cards to specify such a group of files.
Note that you must specify the @option{-c} (compile
only) flag in addition to the @option{-gnats} flag.
.
You may use other switches in conjunction with @option{-gnats}. In
particular, @option{-gnatl} and @option{-gnatv} are useful to control the
format of any generated error messages.

When the source file is empty or contains only empty lines and/or comments,
the output is a warning:

@smallexample
$ gcc -c -gnats -x ada toto.txt
toto.txt:1:01: warning: empty file, contains no compilation units
$
@end smallexample

Otherwise, the output is simply the error messages, if any. No object file or
ALI file is generated by a syntax-only compilation. Also, no units other
than the one specified are accessed. For example, if a unit @code{X}
@code{with}'s a unit @code{Y}, compiling unit @code{X} in syntax
check only mode does not access the source file containing unit
@code{Y}.

@cindex Multiple units, syntax checking
Normally, GNAT allows only a single unit in a source file. However, this
restriction does not apply in syntax-check-only mode, and it is possible
to check a file containing multiple compilation units concatenated
together. This is primarily used by the @code{gnatchop} utility
(@pxref{Renaming Files with gnatchop}).
@end table

@node Using gcc for Semantic Checking
@subsection Using @command{gcc} for Semantic Checking
@table @option
@item -gnatc
@cindex @option{-gnatc} (@command{gcc})

@noindent
The @code{c} stands for ``check''.
Causes the compiler to operate in semantic check mode,
with full checking for all illegalities specified in the
Ada Reference Manual, but without generation of any object code
(no object file is generated).

Because dependent files must be accessed, you must follow the GNAT
semantic restrictions on file structuring to operate in this mode:

@itemize @bullet
@item
The needed source files must be accessible
(@pxref{Search Paths and the Run-Time Library (RTL)}).

@item
Each file must contain only one compilation unit.

@item
The file name and unit name must match (@pxref{File Naming Rules}).
@end itemize

The output consists of error messages as appropriate. No object file is
generated. An @file{ALI} file is generated for use in the context of
cross-reference tools, but this file is marked as not being suitable
for binding (since no object file is generated).
The checking corresponds exactly to the notion of
legality in the Ada Reference Manual.

Any unit can be compiled in semantics-checking-only mode, including
units that would not normally be compiled (subunits,
and specifications where a separate body is present).
@end table

@node Compiling Different Versions of Ada
@subsection Compiling Different Versions of Ada

@noindent
The switches described in this section allow you to explicitly specify
the version of the Ada language that your programs are written in.
The default mode is Ada 2012,
but you can also specify Ada 95, Ada 2005 mode, or
indicate Ada 83 compatibility mode.

@table @option
@cindex Compatibility with Ada 83

@item -gnat83 (Ada 83 Compatibility Mode)
@cindex @option{-gnat83} (@command{gcc})
@cindex ACVC, Ada 83 tests
@cindex Ada 83 mode

@noindent
Although GNAT is primarily an Ada 95 / Ada 2005 compiler, this switch
specifies that the program is to be compiled in Ada 83 mode. With
@option{-gnat83}, GNAT rejects most post-Ada 83 extensions and applies Ada 83
semantics where this can be done easily.
It is not possible to guarantee this switch does a perfect
job; some subtle tests, such as are
found in earlier ACVC tests (and that have been removed from the ACATS suite
for Ada 95), might not compile correctly.
Nevertheless, this switch may be useful in some circumstances, for example
where, due to contractual reasons, existing code needs to be maintained
using only Ada 83 features.

With few exceptions (most notably the need to use @code{<>} on
@cindex Generic formal parameters
unconstrained generic formal parameters, the use of the new Ada 95 / Ada 2005
reserved words, and the use of packages
with optional bodies), it is not necessary to specify the
@option{-gnat83} switch when compiling Ada 83 programs, because, with rare
exceptions, Ada 95 and Ada 2005 are upwardly compatible with Ada 83. Thus
a correct Ada 83 program is usually also a correct program
in these later versions of the language standard.
For further information, please refer to @ref{Compatibility and Porting Guide}.

@item -gnat95 (Ada 95 mode)
@cindex @option{-gnat95} (@command{gcc})
@cindex Ada 95 mode

@noindent
This switch directs the compiler to implement the Ada 95 version of the
language.
Since Ada 95 is almost completely upwards
compatible with Ada 83, Ada 83 programs may generally be compiled using
this switch (see the description of the @option{-gnat83} switch for further
information about Ada 83 mode).
If an Ada 2005 program is compiled in Ada 95 mode,
uses of the new Ada 2005 features will cause error
messages or warnings.

This switch also can be used to cancel the effect of a previous
@option{-gnat83}, @option{-gnat05/2005}, or @option{-gnat12/2012}
switch earlier in the command line.

@item -gnat05 or -gnat2005 (Ada 2005 mode)
@cindex @option{-gnat05} (@command{gcc})
@cindex @option{-gnat2005} (@command{gcc})
@cindex Ada 2005 mode

@noindent
This switch directs the compiler to implement the Ada 2005 version of the
language, as documented in the official Ada standards document.
Since Ada 2005 is almost completely upwards
compatible with Ada 95 (and thus also with Ada 83), Ada 83 and Ada 95 programs
may generally be compiled using this switch (see the description of the
@option{-gnat83} and @option{-gnat95} switches for further
information).

@item -gnat12 or -gnat2012 (Ada 2012 mode)
@cindex @option{-gnat12} (@command{gcc})
@cindex @option{-gnat2012} (@command{gcc})
@cindex Ada 2012 mode

@noindent
This switch directs the compiler to implement the Ada 2012 version of the
language (also the default).
Since Ada 2012 is almost completely upwards
compatible with Ada 2005 (and thus also with Ada 83, and Ada 95),
Ada 83 and Ada 95 programs
may generally be compiled using this switch (see the description of the
@option{-gnat83}, @option{-gnat95}, and @option{-gnat05/2005} switches
for further information).

@item -gnatX (Enable GNAT Extensions)
@cindex @option{-gnatX} (@command{gcc})
@cindex Ada language extensions
@cindex GNAT extensions

@noindent
This switch directs the compiler to implement the latest version of the
language (currently Ada 2012) and also to enable certain GNAT implementation
extensions that are not part of any Ada standard. For a full list of these
extensions, see the GNAT reference manual.

@end table

@node Character Set Control
@subsection Character Set Control
@table @option
@item -gnati@var{c}
@cindex @option{-gnati} (@command{gcc})

@noindent
Normally GNAT recognizes the Latin-1 character set in source program
identifiers, as described in the Ada Reference Manual.
This switch causes
GNAT to recognize alternate character sets in identifiers. @var{c} is a
single character  indicating the character set, as follows:

@table @code
@item 1
ISO 8859-1 (Latin-1) identifiers

@item 2
ISO 8859-2 (Latin-2) letters allowed in identifiers

@item 3
ISO 8859-3 (Latin-3) letters allowed in identifiers

@item 4
ISO 8859-4 (Latin-4) letters allowed in identifiers

@item 5
ISO 8859-5 (Cyrillic) letters allowed in identifiers

@item 9
ISO 8859-15 (Latin-9) letters allowed in identifiers

@item p
IBM PC letters (code page 437) allowed in identifiers

@item 8
IBM PC letters (code page 850) allowed in identifiers

@item f
Full upper-half codes allowed in identifiers

@item n
No upper-half codes allowed in identifiers

@item w
Wide-character codes (that is, codes greater than 255)
allowed in identifiers
@end table

@xref{Foreign Language Representation}, for full details on the
implementation of these character sets.

@item -gnatW@var{e}
@cindex @option{-gnatW} (@command{gcc})
Specify the method of encoding for wide characters.
@var{e} is one of the following:

@table @code

@item h
Hex encoding (brackets coding also recognized)

@item u
Upper half encoding (brackets encoding also recognized)

@item s
Shift/JIS encoding (brackets encoding also recognized)

@item e
EUC encoding (brackets encoding also recognized)

@item 8
UTF-8 encoding (brackets encoding also recognized)

@item b
Brackets encoding only (default value)
@end table
For full details on these encoding
methods see @ref{Wide_Character Encodings}.
Note that brackets coding is always accepted, even if one of the other
options is specified, so for example @option{-gnatW8} specifies that both
brackets and UTF-8 encodings will be recognized. The units that are
with'ed directly or indirectly will be scanned using the specified
representation scheme, and so if one of the non-brackets scheme is
used, it must be used consistently throughout the program. However,
since brackets encoding is always recognized, it may be conveniently
used in standard libraries, allowing these libraries to be used with
any of the available coding schemes.

Note that brackets encoding only applies to program text. Within comments,
brackets are considered to be normal graphic characters, and bracket sequences
are never recognized as wide characters.

If no @option{-gnatW?} parameter is present, then the default
representation is normally Brackets encoding only. However, if the
first three characters of the file are 16#EF# 16#BB# 16#BF# (the standard
byte order mark or BOM for UTF-8), then these three characters are
skipped and the default representation for the file is set to UTF-8.

Note that the wide character representation that is specified (explicitly
or by default) for the main program also acts as the default encoding used
for Wide_Text_IO files if not specifically overridden by a WCEM form
parameter.

@end table

When no @option{-gnatW?} is specified, then characters (other than wide
characters represented using brackets notation) are treated as 8-bit
Latin-1 codes. The codes recognized are the Latin-1 graphic characters,
and ASCII format effectors (CR, LF, HT, VT). Other lower half control
characters in the range 16#00#..16#1F# are not accepted in program text
or in comments. Upper half control characters (16#80#..16#9F#) are rejected
in program text, but allowed and ignored in comments. Note in particular
that the Next Line (NEL) character whose encoding is 16#85# is not recognized
as an end of line in this default mode. If your source program contains
instances of the NEL character used as a line terminator,
you must use UTF-8 encoding for the whole
source program. In default mode, all lines must be ended by a standard
end of line sequence (CR, CR/LF, or LF).

Note that the convention of simply accepting all upper half characters in
comments means that programs that use standard ASCII for program text, but
UTF-8 encoding for comments are accepted in default mode, providing that the
comments are ended by an appropriate (CR, or CR/LF, or LF) line terminator.
This is a common mode for many programs with foreign language comments.

@node File Naming Control
@subsection File Naming Control

@table @option
@item -gnatk@var{n}
@cindex @option{-gnatk} (@command{gcc})
Activates file name ``krunching''. @var{n}, a decimal integer in the range
1-999, indicates the maximum allowable length of a file name (not
including the @file{.ads} or @file{.adb} extension). The default is not
to enable file name krunching.

For the source file naming rules, @xref{File Naming Rules}.
@end table

@node Subprogram Inlining Control
@subsection Subprogram Inlining Control

@table @option
@c !sort!
@item -gnatn[12]
@cindex @option{-gnatn} (@command{gcc})
The @code{n} here is intended to suggest the first syllable of the
word ``inline''.
GNAT recognizes and processes @code{Inline} pragmas. However, for the
inlining to actually occur, optimization must be enabled and, in order
to enable inlining of subprograms specified by pragma @code{Inline},
you must also specify this switch.
In the absence of this switch, GNAT does not attempt
inlining and does not need to access the bodies of
subprograms for which @code{pragma Inline} is specified if they are not
in the current unit.

You can optionally specify the inlining level: 1 for moderate inlining across
modules, which is a good compromise between compilation times and performances
at run time, or 2 for full inlining across modules, which may bring about
longer compilation times. If no inlining level is specified, the compiler will
pick it based on the optimization level: 1 for @option{-O1}, @option{-O2} or
@option{-Os} and 2 for @option{-O3}.

If you specify this switch the compiler will access these bodies,
creating an extra source dependency for the resulting object file, and
where possible, the call will be inlined.
For further details on when inlining is possible
see @ref{Inlining of Subprograms}.

@item -gnatN
@cindex @option{-gnatN} (@command{gcc})
This switch activates front-end inlining which also
generates additional dependencies.

When using a gcc-based back end (in practice this means using any version
of GNAT other than the JGNAT, .NET or GNAAMP versions), then the use of
@option{-gnatN} is deprecated, and the use of @option{-gnatn} is preferred.
Historically front end inlining was more extensive than the gcc back end
inlining, but that is no longer the case.
@end table

@node Auxiliary Output Control
@subsection Auxiliary Output Control

@table @option
@item -gnatt
@cindex @option{-gnatt} (@command{gcc})
@cindex Writing internal trees
@cindex Internal trees, writing to file
Causes GNAT to write the internal tree for a unit to a file (with the
extension @file{.adt}.
This not normally required, but is used by separate analysis tools.
Typically
these tools do the necessary compilations automatically, so you should
not have to specify this switch in normal operation.
Note that the combination of switches @option{-gnatct}
generates a tree in the form required by ASIS applications.

@item -gnatu
@cindex @option{-gnatu} (@command{gcc})
Print a list of units required by this compilation on @file{stdout}.
The listing includes all units on which the unit being compiled depends
either directly or indirectly.

@item -pass-exit-codes
@cindex @option{-pass-exit-codes} (@command{gcc})
If this switch is not used, the exit code returned by @command{gcc} when
compiling multiple files indicates whether all source files have
been successfully used to generate object files or not.

When @option{-pass-exit-codes} is used, @command{gcc} exits with an extended
exit status and allows an integrated development environment to better
react to a compilation failure. Those exit status are:

@table @asis
@item 5
There was an error in at least one source file.
@item 3
At least one source file did not generate an object file.
@item 2
The compiler died unexpectedly (internal error for example).
@item 0
An object file has been generated for every source file.
@end table
@end table

@node Debugging Control
@subsection Debugging Control

@table @option
@c !sort!
@cindex Debugging options
@item -gnatd@var{x}
@cindex @option{-gnatd} (@command{gcc})
Activate internal debugging switches. @var{x} is a letter or digit, or
string of letters or digits, which specifies the type of debugging
outputs desired. Normally these are used only for internal development
or system debugging purposes. You can find full documentation for these
switches in the body of the @code{Debug} unit in the compiler source
file @file{debug.adb}.

@item -gnatG[=nn]
@cindex @option{-gnatG} (@command{gcc})
This switch causes the compiler to generate auxiliary output containing
a pseudo-source listing of the generated expanded code. Like most Ada
compilers, GNAT works by first transforming the high level Ada code into
lower level constructs. For example, tasking operations are transformed
into calls to the tasking run-time routines. A unique capability of GNAT
is to list this expanded code in a form very close to normal Ada source.
This is very useful in understanding the implications of various Ada
usage on the efficiency of the generated code. There are many cases in
Ada (e.g.@: the use of controlled types), where simple Ada statements can
generate a lot of run-time code. By using @option{-gnatG} you can identify
these cases, and consider whether it may be desirable to modify the coding
approach to improve efficiency.

The optional parameter @code{nn} if present after -gnatG specifies an
alternative maximum line length that overrides the normal default of 72.
This value is in the range 40-999999, values less than 40 being silently
reset to 40. The equal sign is optional.

The format of the output is very similar to standard Ada source, and is
easily understood by an Ada programmer. The following special syntactic
additions correspond to low level features used in the generated code that
do not have any exact analogies in pure Ada source form. The following
is a partial list of these special constructions. See the spec
of package @code{Sprint} in file @file{sprint.ads} for a full list.

If the switch @option{-gnatL} is used in conjunction with
@cindex @option{-gnatL} (@command{gcc})
@option{-gnatG}, then the original source lines are interspersed
in the expanded source (as comment lines with the original line number).

@table @code
@item new @var{xxx} @r{[}storage_pool = @var{yyy}@r{]}
Shows the storage pool being used for an allocator.

@item at end @var{procedure-name};
Shows the finalization (cleanup) procedure for a scope.

@item (if @var{expr} then @var{expr} else @var{expr})
Conditional expression equivalent to the @code{x?y:z} construction in C.

@item @var{target}^(@var{source})
A conversion with floating-point truncation instead of rounding.

@item @var{target}?(@var{source})
A conversion that bypasses normal Ada semantic checking. In particular
enumeration types and fixed-point types are treated simply as integers.

@item @var{target}?^(@var{source})
Combines the above two cases.

@item @var{x} #/ @var{y}
@itemx @var{x} #mod @var{y}
@itemx @var{x} #* @var{y}
@itemx @var{x} #rem @var{y}
A division or multiplication of fixed-point values which are treated as
integers without any kind of scaling.

@item free @var{expr} @r{[}storage_pool = @var{xxx}@r{]}
Shows the storage pool associated with a @code{free} statement.

@item [subtype or type declaration]
Used to list an equivalent declaration for an internally generated
type that is referenced elsewhere in the listing.

@c @item freeze @var{type-name} @ovar{actions}
@c Expanding @ovar macro inline (explanation in macro def comments)
@item freeze @var{type-name} @r{[}@var{actions}@r{]}
Shows the point at which @var{type-name} is frozen, with possible
associated actions to be performed at the freeze point.

@item reference @var{itype}
Reference (and hence definition) to internal type @var{itype}.

@item @var{function-name}! (@var{arg}, @var{arg}, @var{arg})
Intrinsic function call.

@item @var{label-name} : label
Declaration of label @var{labelname}.

@item #$ @var{subprogram-name}
An implicit call to a run-time support routine
(to meet the requirement of H.3.1(9) in a
convenient manner).

@item @var{expr} && @var{expr} && @var{expr} @dots{} && @var{expr}
A multiple concatenation (same effect as @var{expr} & @var{expr} &
@var{expr}, but handled more efficiently).

@item [constraint_error]
Raise the @code{Constraint_Error} exception.

@item @var{expression}'reference
A pointer to the result of evaluating @var{expression}.

@item @var{target-type}!(@var{source-expression})
An unchecked conversion of @var{source-expression} to @var{target-type}.

@item [@var{numerator}/@var{denominator}]
Used to represent internal real literals (that) have no exact
representation in base 2-16 (for example, the result of compile time
evaluation of the expression 1.0/27.0).
@end table

@item -gnatD[=nn]
@cindex @option{-gnatD} (@command{gcc})
When used in conjunction with @option{-gnatG}, this switch causes
the expanded source, as described above for
@option{-gnatG} to be written to files with names
@file{xxx.dg}, where @file{xxx} is the normal file name,
instead of to the standard output file. For
example, if the source file name is @file{hello.adb}, then a file
@file{hello.adb.dg} will be written.  The debugging
information generated by the @command{gcc} @option{-g} switch
will refer to the generated @file{xxx.dg} file. This allows
you to do source level debugging using the generated code which is
sometimes useful for complex code, for example to find out exactly
which part of a complex construction raised an exception. This switch
also suppress generation of cross-reference information (see
@option{-gnatx}) since otherwise the cross-reference information
would refer to the @file{.dg} file, which would cause
confusion since this is not the original source file.

Note that @option{-gnatD} actually implies @option{-gnatG}
automatically, so it is not necessary to give both options.
In other words @option{-gnatD} is equivalent to @option{-gnatDG}).

If the switch @option{-gnatL} is used in conjunction with
@cindex @option{-gnatL} (@command{gcc})
@option{-gnatDG}, then the original source lines are interspersed
in the expanded source (as comment lines with the original line number).

The optional parameter @code{nn} if present after -gnatD specifies an
alternative maximum line length that overrides the normal default of 72.
This value is in the range 40-999999, values less than 40 being silently
reset to 40. The equal sign is optional.

@item -gnatr
@cindex @option{-gnatr} (@command{gcc})
@cindex pragma Restrictions
This switch causes pragma Restrictions to be treated as Restriction_Warnings
so that violation of restrictions causes warnings rather than illegalities.
This is useful during the development process when new restrictions are added
or investigated. The switch also causes pragma Profile to be treated as
Profile_Warnings, and pragma Restricted_Run_Time and pragma Ravenscar set
restriction warnings rather than restrictions.

@item -gnatR@r{[}0@r{|}1@r{|}2@r{|}3@r{[}s@r{]]}
@cindex @option{-gnatR} (@command{gcc})
This switch controls output from the compiler of a listing showing
representation information for declared types and objects. For
@option{-gnatR0}, no information is output (equivalent to omitting
the @option{-gnatR} switch). For @option{-gnatR1} (which is the default,
so @option{-gnatR} with no parameter has the same effect), size and alignment
information is listed for declared array and record types. For
@option{-gnatR2}, size and alignment information is listed for all
declared types and objects. The @code{Linker_Section} is also listed for any
entity for which the @code{Linker_Section} is set explicitly or implicitly (the
latter case occurs for objects of a type for which a @code{Linker_Section}
is set).

Finally @option{-gnatR3} includes symbolic
expressions for values that are computed at run time for
variant records. These symbolic expressions have a mostly obvious
format with #n being used to represent the value of the n'th
discriminant. See source files @file{repinfo.ads/adb} in the
@code{GNAT} sources for full details on the format of @option{-gnatR3}
output. If the switch is followed by an s (e.g.@: @option{-gnatR2s}), then
the output is to a file with the name @file{file.rep} where
file is the name of the corresponding source file.

@item -gnatRm[s]
This form of the switch controls output of subprogram conventions
and parameter passing mechanisms for all subprograms. A following
@code{s} means output to a file as described above.

Note that it is possible for record components to have zero size. In
this case, the component clause uses an obvious extension of permitted
Ada syntax, for example @code{at 0 range 0 .. -1}.

Representation information requires that code be generated (since it is the
code generator that lays out complex data structures). If an attempt is made
to output representation information when no code is generated, for example
when a subunit is compiled on its own, then no information can be generated
and the compiler outputs a message to this effect.

@item -gnatS
@cindex @option{-gnatS} (@command{gcc})
The use of the switch @option{-gnatS} for an
Ada compilation will cause the compiler to output a
representation of package Standard in a form very
close to standard Ada. It is not quite possible to
do this entirely in standard Ada (since new
numeric base types cannot be created in standard
Ada), but the output is easily
readable to any Ada programmer, and is useful to
determine the characteristics of target dependent
types in package Standard.

@item -gnatx
@cindex @option{-gnatx} (@command{gcc})
Normally the compiler generates full cross-referencing information in
the @file{ALI} file. This information is used by a number of tools,
including @code{gnatfind} and @code{gnatxref}. The @option{-gnatx} switch
suppresses this information. This saves some space and may slightly
speed up compilation, but means that these tools cannot be used.
@end table

@node Exception Handling Control
@subsection Exception Handling Control

@noindent
GNAT uses two methods for handling exceptions at run-time. The
@code{setjmp/longjmp} method saves the context when entering
a frame with an exception handler. Then when an exception is
raised, the context can be restored immediately, without the
need for tracing stack frames. This method provides very fast
exception propagation, but introduces significant overhead for
the use of exception handlers, even if no exception is raised.

The other approach is called ``zero cost'' exception handling.
With this method, the compiler builds static tables to describe
the exception ranges. No dynamic code is required when entering
a frame containing an exception handler. When an exception is
raised, the tables are used to control a back trace of the
subprogram invocation stack to locate the required exception
handler. This method has considerably poorer performance for
the propagation of exceptions, but there is no overhead for
exception handlers if no exception is raised. Note that in this
mode and in the context of mixed Ada and C/C++ programming,
to propagate an exception through a C/C++ code, the C/C++ code
must be compiled with the @option{-funwind-tables} GCC's
option.

The following switches may be used to control which of the
two exception handling methods is used.

@table @option
@c !sort!

@item --RTS=sjlj
@cindex @option{--RTS=sjlj} (@command{gnatmake})
This switch causes the setjmp/longjmp run-time (when available) to be used
for exception handling. If the default
mechanism for the target is zero cost exceptions, then
this switch can be used to modify this default, and must be
used for all units in the partition.
This option is rarely used. One case in which it may be
advantageous is if you have an application where exception
raising is common and the overall performance of the
application is improved by favoring exception propagation.

@item --RTS=zcx
@cindex @option{--RTS=zcx} (@command{gnatmake})
@cindex Zero Cost Exceptions
This switch causes the zero cost approach to be used
for exception handling. If this is the default mechanism for the
target (see below), then this switch is unneeded. If the default
mechanism for the target is setjmp/longjmp exceptions, then
this switch can be used to modify this default, and must be
used for all units in the partition.
This option can only be used if the zero cost approach
is available for the target in use, otherwise it will generate an error.
@end table

@noindent
The same option @option{--RTS} must be used both for @command{gcc}
and @command{gnatbind}. Passing this option to @command{gnatmake}
(@pxref{Switches for gnatmake}) will ensure the required consistency
through the compilation and binding steps.

@node Units to Sources Mapping Files
@subsection Units to Sources Mapping Files

@table @option

@item -gnatem=@var{path}
@cindex @option{-gnatem} (@command{gcc})
A mapping file is a way to communicate to the compiler two mappings:
from unit names to file names (without any directory information) and from
file names to path names (with full directory information). These mappings
are used by the compiler to short-circuit the path search.

The use of mapping files is not required for correct operation of the
compiler, but mapping files can improve efficiency, particularly when
sources are read over a slow network connection. In normal operation,
you need not be concerned with the format or use of mapping files,
and the @option{-gnatem} switch is not a switch that you would use
explicitly. It is intended primarily for use by automatic tools such as
@command{gnatmake} running under the project file facility. The
description here of the format of mapping files is provided
for completeness and for possible use by other tools.

A mapping file is a sequence of sets of three lines. In each set, the
first line is the unit name, in lower case, with @code{%s} appended
for specs and @code{%b} appended for bodies; the second line is the
file name; and the third line is the path name.

Example:
@smallexample
   main%b
   main.2.ada
   /gnat/project1/sources/main.2.ada
@end smallexample

When the switch @option{-gnatem} is specified, the compiler will
create in memory the two mappings from the specified file. If there is
any problem (nonexistent file, truncated file or duplicate entries),
no mapping will be created.

Several @option{-gnatem} switches may be specified; however, only the
last one on the command line will be taken into account.

When using a project file, @command{gnatmake} creates a temporary
mapping file and communicates it to the compiler using this switch.

@end table

@node Integrated Preprocessing
@subsection Integrated Preprocessing

@noindent
GNAT sources may be preprocessed immediately before compilation.
In this case, the actual
text of the source is not the text of the source file, but is derived from it
through a process called preprocessing. Integrated preprocessing is specified
through switches @option{-gnatep} and/or @option{-gnateD}. @option{-gnatep}
indicates, through a text file, the preprocessing data to be used.
@option{-gnateD} specifies or modifies the values of preprocessing symbol.
Note that integrated preprocessing applies only to Ada source files, it is
not available for configuration pragma files.

@noindent
Note that when integrated preprocessing is used, the output from the
preprocessor is not written to any external file. Instead it is passed
internally to the compiler. If you need to preserve the result of
preprocessing in a file, then you should use @command{gnatprep}
to perform the desired preprocessing in stand-alone mode.

@noindent
It is recommended that @command{gnatmake} switch -s should be
used when Integrated Preprocessing is used. The reason is that preprocessing
with another Preprocessing Data file without changing the sources will
not trigger recompilation without this switch.

@noindent
Note that @command{gnatmake} switch -m will almost
always trigger recompilation for sources that are preprocessed,
because @command{gnatmake} cannot compute the checksum of the source after
preprocessing.

@noindent
The actual preprocessing function is described in details in section
@ref{Preprocessing with gnatprep}. This section only describes how integrated
preprocessing is triggered and parameterized.

@table @code

@item -gnatep=@var{file}
@cindex @option{-gnatep} (@command{gcc})
This switch indicates to the compiler the file name (without directory
information) of the preprocessor data file to use. The preprocessor data file
should be found in the source directories. Note that when the compiler is
called by a builder such as (@command{gnatmake} with a project
file, if the object directory is not also a source directory, the builder needs
to be called with @option{-x}.

@noindent
A preprocessing data file is a text file with significant lines indicating
how should be preprocessed either a specific source or all sources not
mentioned in other lines. A significant line is a nonempty, non-comment line.
Comments are similar to Ada comments.

@noindent
Each significant line starts with either a literal string or the character '*'.
A literal string is the file name (without directory information) of the source
to preprocess. A character '*' indicates the preprocessing for all the sources
that are not specified explicitly on other lines (order of the lines is not
significant). It is an error to have two lines with the same file name or two
lines starting with the character '*'.

@noindent
After the file name or the character '*', another optional literal string
indicating the file name of the definition file to be used for preprocessing
(@pxref{Form of Definitions File}). The definition files are found by the
compiler in one of the source directories. In some cases, when compiling
a source in a directory other than the current directory, if the definition
file is in the current directory, it may be necessary to add the current
directory as a source directory through switch -I., otherwise
the compiler would not find the definition file.

@noindent
Then, optionally, switches similar to those of @code{gnatprep} may
be found. Those switches are:

@table @code

@item -b
Causes both preprocessor lines and the lines deleted by
preprocessing to be replaced by blank lines, preserving the line number.
This switch is always implied; however, if specified after @option{-c}
it cancels the effect of @option{-c}.

@item -c
Causes both preprocessor lines and the lines deleted
by preprocessing to be retained as comments marked
with the special string ``@code{--! }''.

@item -Dsymbol=value
Define or redefine a symbol, associated with value. A symbol is an Ada
identifier, or an Ada reserved word, with the exception of @code{if},
@code{else}, @code{elsif}, @code{end}, @code{and}, @code{or} and @code{then}.
@code{value} is either a literal string, an Ada identifier or any Ada reserved
word. A symbol declared with this switch replaces a symbol with the
same name defined in a definition file.

@item -s
Causes a sorted list of symbol names and values to be
listed on the standard output file.

@item -u
Causes undefined symbols to be treated as having the value @code{FALSE}
in the context
of a preprocessor test. In the absence of this option, an undefined symbol in
a @code{#if} or @code{#elsif} test will be treated as an error.

@end table

@noindent
Examples of valid lines in a preprocessor data file:

@smallexample
  "toto.adb"  "prep.def" -u
  --  preprocess "toto.adb", using definition file "prep.def",
  --  undefined symbol are False.

  * -c -DVERSION=V101
  --  preprocess all other sources without a definition file;
  --  suppressed lined are commented; symbol VERSION has the value V101.

  "titi.adb" "prep2.def" -s
  --  preprocess "titi.adb", using definition file "prep2.def";
  --  list all symbols with their values.
@end smallexample

@item -gnateDsymbol@r{[}=value@r{]}
@cindex @option{-gnateD} (@command{gcc})
Define or redefine a preprocessing symbol, associated with value. If no value
is given on the command line, then the value of the symbol is @code{True}.
A symbol is an identifier, following normal Ada (case-insensitive)
rules for its syntax, and value is either an arbitrary string between double
quotes or any sequence (including an empty sequence) of characters from the
set (letters, digits, period, underline).
Ada reserved words may be used as symbols, with the exceptions of @code{if},
@code{else}, @code{elsif}, @code{end}, @code{and}, @code{or} and @code{then}.

@noindent
Examples:

@smallexample
   -gnateDToto=Titi
   -gnateDFoo
   -gnateDFoo=\"Foo-Bar\"
@end smallexample

@noindent
A symbol declared with this switch on the command line replaces a
symbol with the same name either in a definition file or specified with a
switch -D in the preprocessor data file.

@noindent
This switch is similar to switch @option{-D} of @code{gnatprep}.

@item -gnateG
When integrated preprocessing is performed and the preprocessor modifies
the source text, write the result of this preprocessing into a file
<source>.prep.

@end table

@node Code Generation Control
@subsection Code Generation Control

@noindent

The GCC technology provides a wide range of target dependent
@option{-m} switches for controlling
details of code generation with respect to different versions of
architectures. This includes variations in instruction sets (e.g.@:
different members of the power pc family), and different requirements
for optimal arrangement of instructions (e.g.@: different members of
the x86 family). The list of available @option{-m} switches may be
found in the GCC documentation.

Use of these @option{-m} switches may in some cases result in improved
code performance.

The @value{EDITION} technology is tested and qualified without any
@option{-m} switches,
so generally the most reliable approach is to avoid the use of these
switches. However, we generally expect most of these switches to work
successfully with @value{EDITION}, and many customers have reported successful
use of these options.

Our general advice is to avoid the use of @option{-m} switches unless
special needs lead to requirements in this area. In particular,
there is no point in using @option{-m} switches to improve performance
unless you actually see a performance improvement.


@node Search Paths and the Run-Time Library (RTL)
@section Search Paths and the Run-Time Library (RTL)

@noindent
With the GNAT source-based library system, the compiler must be able to
find source files for units that are needed by the unit being compiled.
Search paths are used to guide this process.

The compiler compiles one source file whose name must be given
explicitly on the command line. In other words, no searching is done
for this file. To find all other source files that are needed (the most
common being the specs of units), the compiler examines the following
directories, in the following order:

@enumerate
@item
The directory containing the source file of the main unit being compiled
(the file name on the command line).

@item
Each directory named by an @option{-I} switch given on the
@command{gcc} command line, in the order given.

@item
@findex ADA_PRJ_INCLUDE_FILE
Each of the directories listed in the text file whose name is given
by the @env{ADA_PRJ_INCLUDE_FILE} environment variable.

@noindent
@env{ADA_PRJ_INCLUDE_FILE} is normally set by gnatmake or by the gnat
driver when project files are used. It should not normally be set
by other means.

@item
@findex ADA_INCLUDE_PATH
Each of the directories listed in the value of the
@env{ADA_INCLUDE_PATH} environment variable.
Construct this value
exactly as the @env{PATH} environment variable: a list of directory
names separated by colons (semicolons when working with the NT version).

@item
The content of the @file{ada_source_path} file which is part of the GNAT
installation tree and is used to store standard libraries such as the
GNAT Run Time Library (RTL) source files.
@ref{Installing a library}
@end enumerate

@noindent
Specifying the switch @option{-I-}
inhibits the use of the directory
containing the source file named in the command line. You can still
have this directory on your search path, but in this case it must be
explicitly requested with a @option{-I} switch.

Specifying the switch @option{-nostdinc}
inhibits the search of the default location for the GNAT Run Time
Library (RTL) source files.

The compiler outputs its object files and ALI files in the current
working directory.
Caution: The object file can be redirected with the @option{-o} switch;
however, @command{gcc} and @code{gnat1} have not been coordinated on this
so the @file{ALI} file will not go to the right place. Therefore, you should
avoid using the @option{-o} switch.

@findex System.IO
The packages @code{Ada}, @code{System}, and @code{Interfaces} and their
children make up the GNAT RTL, together with the simple @code{System.IO}
package used in the @code{"Hello World"} example. The sources for these units
are needed by the compiler and are kept together in one directory. Not
all of the bodies are needed, but all of the sources are kept together
anyway. In a normal installation, you need not specify these directory
names when compiling or binding. Either the environment variables or
the built-in defaults cause these files to be found.

In addition to the language-defined hierarchies (@code{System}, @code{Ada} and
@code{Interfaces}), the GNAT distribution provides a fourth hierarchy,
consisting of child units of @code{GNAT}. This is a collection of generally
useful types, subprograms, etc. @xref{Top, GNAT Reference Manual, About
This Guid, gnat_rm, GNAT Reference Manual}, for further details.

Besides simplifying access to the RTL, a major use of search paths is
in compiling sources from multiple directories. This can make
development environments much more flexible.

@node Order of Compilation Issues
@section Order of Compilation Issues

@noindent
If, in our earlier example, there was a spec for the @code{hello}
procedure, it would be contained in the file @file{hello.ads}; yet this
file would not have to be explicitly compiled. This is the result of the
model we chose to implement library management. Some of the consequences
of this model are as follows:

@itemize @bullet
@item
There is no point in compiling specs (except for package
specs with no bodies) because these are compiled as needed by clients. If
you attempt a useless compilation, you will receive an error message.
It is also useless to compile subunits because they are compiled as needed
by the parent.

@item
There are no order of compilation requirements: performing a
compilation never obsoletes anything. The only way you can obsolete
something and require recompilations is to modify one of the
source files on which it depends.

@item
There is no library as such, apart from the ALI files
(@pxref{The Ada Library Information Files}, for information on the format
of these files). For now we find it convenient to create separate ALI files,
but eventually the information therein may be incorporated into the object
file directly.

@item
When you compile a unit, the source files for the specs of all units
that it @code{with}'s, all its subunits, and the bodies of any generics it
instantiates must be available (reachable by the search-paths mechanism
described above), or you will receive a fatal error message.
@end itemize

@node Examples
@section Examples

@noindent
The following are some typical Ada compilation command line examples:

@table @code
@item $ gcc -c xyz.adb
Compile body in file @file{xyz.adb} with all default options.

@item $ gcc -c -O2 -gnata xyz-def.adb

Compile the child unit package in file @file{xyz-def.adb} with extensive
optimizations, and pragma @code{Assert}/@code{Debug} statements
enabled.

@item $ gcc -c -gnatc abc-def.adb
Compile the subunit in file @file{abc-def.adb} in semantic-checking-only
mode.
@end table

@node Binding with gnatbind
@chapter Binding with @code{gnatbind}
@findex gnatbind

@menu
* Running gnatbind::
* Switches for gnatbind::
* Command-Line Access::
* Search Paths for gnatbind::
* Examples of gnatbind Usage::
@end menu

@noindent
This chapter describes the GNAT binder, @code{gnatbind}, which is used
to bind compiled GNAT objects.

Note: to invoke @code{gnatbind} with a project file, use the @code{gnat}
driver (see @ref{The GNAT Driver and Project Files}).

The @code{gnatbind} program performs four separate functions:

@enumerate
@item
Checks that a program is consistent, in accordance with the rules in
Chapter 10 of the Ada Reference Manual. In particular, error
messages are generated if a program uses inconsistent versions of a
given unit.

@item
Checks that an acceptable order of elaboration exists for the program
and issues an error message if it cannot find an order of elaboration
that satisfies the rules in Chapter 10 of the Ada Language Manual.

@item
Generates a main program incorporating the given elaboration order.
This program is a small Ada package (body and spec) that
must be subsequently compiled
using the GNAT compiler. The necessary compilation step is usually
performed automatically by @command{gnatlink}. The two most important
functions of this program
are to call the elaboration routines of units in an appropriate order
and to call the main program.

@item
Determines the set of object files required by the given main program.
This information is output in the forms of comments in the generated program,
to be read by the @command{gnatlink} utility used to link the Ada application.
@end enumerate

@node Running gnatbind
@section Running @code{gnatbind}

@noindent
The form of the @code{gnatbind} command is

@smallexample
@c $ gnatbind @ovar{switches} @var{mainprog}@r{[}.ali@r{]} @ovar{switches}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatbind @r{[}@var{switches}@r{]} @var{mainprog}@r{[}.ali@r{]} @r{[}@var{switches}@r{]}
@end smallexample

@noindent
where @file{@var{mainprog}.adb} is the Ada file containing the main program
unit body. @code{gnatbind} constructs an Ada
package in two files whose names are
@file{b~@var{mainprog}.ads}, and @file{b~@var{mainprog}.adb}.
For example, if given the
parameter @file{hello.ali}, for a main program contained in file
@file{hello.adb}, the binder output files would be @file{b~hello.ads}
and @file{b~hello.adb}.

When doing consistency checking, the binder takes into consideration
any source files it can locate. For example, if the binder determines
that the given main program requires the package @code{Pack}, whose
@file{.ALI}
file is @file{pack.ali} and whose corresponding source spec file is
@file{pack.ads}, it attempts to locate the source file @file{pack.ads}
(using the same search path conventions as previously described for the
@command{gcc} command). If it can locate this source file, it checks that
the time stamps
or source checksums of the source and its references to in @file{ALI} files
match. In other words, any @file{ALI} files that mentions this spec must have
resulted from compiling this version of the source file (or in the case
where the source checksums match, a version close enough that the
difference does not matter).

@cindex Source files, use by binder
The effect of this consistency checking, which includes source files, is
that the binder ensures that the program is consistent with the latest
version of the source files that can be located at bind time. Editing a
source file without compiling files that depend on the source file cause
error messages to be generated by the binder.

For example, suppose you have a main program @file{hello.adb} and a
package @code{P}, from file @file{p.ads} and you perform the following
steps:

@enumerate
@item
Enter @code{gcc -c hello.adb} to compile the main program.

@item
Enter @code{gcc -c p.ads} to compile package @code{P}.

@item
Edit file @file{p.ads}.

@item
Enter @code{gnatbind hello}.
@end enumerate

@noindent
At this point, the file @file{p.ali} contains an out-of-date time stamp
because the file @file{p.ads} has been edited. The attempt at binding
fails, and the binder generates the following error messages:

@smallexample
error: "hello.adb" must be recompiled ("p.ads" has been modified)
error: "p.ads" has been modified and must be recompiled
@end smallexample

@noindent
Now both files must be recompiled as indicated, and then the bind can
succeed, generating a main program. You need not normally be concerned
with the contents of this file, but for reference purposes a sample
binder output file is given in @ref{Example of Binder Output File}.

In most normal usage, the default mode of @command{gnatbind} which is to
generate the main package in Ada, as described in the previous section.
In particular, this means that any Ada programmer can read and understand
the generated main program. It can also be debugged just like any other
Ada code provided the @option{-g} switch is used for
@command{gnatbind} and @command{gnatlink}.

@node Switches for gnatbind
@section Switches for @command{gnatbind}

@noindent
The following switches are available with @code{gnatbind}; details will
be presented in subsequent sections.

@menu
* Consistency-Checking Modes::
* Binder Error Message Control::
* Elaboration Control::
* Output Control::
* Dynamic Allocation Control::
* Binding with Non-Ada Main Programs::
* Binding Programs with No Main Subprogram::
@end menu

@table @option
@c !sort!

@item --version
@cindex @option{--version} @command{gnatbind}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatbind}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item -a
@cindex @option{-a} @command{gnatbind}
Indicates that, if supported by the platform, the adainit procedure should
be treated as an initialisation routine by the linker (a constructor). This
is intended to be used by the Project Manager to automatically initialize
shared Stand-Alone Libraries.

@item -aO
@cindex @option{-aO} (@command{gnatbind})
Specify directory to be searched for ALI files.

@item -aI
@cindex @option{-aI} (@command{gnatbind})
Specify directory to be searched for source file.

@item -A@r{[=}@var{filename}@r{]}
@cindex @option{-A} (@command{gnatbind})
Output ALI list (to standard output or to the named file).

@item -b
@cindex @option{-b} (@command{gnatbind})
Generate brief messages to @file{stderr} even if verbose mode set.

@item -c
@cindex @option{-c} (@command{gnatbind})
Check only, no generation of binder output file.

@item -d@var{nn}@r{[}k@r{|}m@r{]}
@cindex @option{-d@var{nn}@r{[}k@r{|}m@r{]}} (@command{gnatbind})
This switch can be used to change the default task stack size value
to a specified size @var{nn}, which is expressed in bytes by default, or
in kilobytes when suffixed with @var{k} or in megabytes when suffixed
with @var{m}.
In the absence of a @samp{@r{[}k@r{|}m@r{]}} suffix, this switch is equivalent,
in effect, to completing all task specs with
@smallexample @c ada
   @b{pragma} Storage_Size (nn);
@end smallexample
When they do not already have such a pragma.

@item -D@var{nn}@r{[}k@r{|}m@r{]}
@cindex @option{-D} (@command{gnatbind})
This switch can be used to change the default secondary stack size value
to a specified size @var{nn}, which is expressed in bytes by default, or
in kilobytes when suffixed with @var{k} or in megabytes when suffixed
with @var{m}.

The secondary stack is used to deal with functions that return a variable
sized result, for example a function returning an unconstrained
String. There are two ways in which this secondary stack is allocated.

For most targets, the secondary stack is growing on demand and is allocated
as a chain of blocks in the heap. The -D option is not very
relevant. It only give some control over the size of the allocated
blocks (whose size is the minimum of the default secondary stack size value,
and the actual size needed for the current allocation request).

For certain targets, notably VxWorks 653,
the secondary stack is allocated by carving off a fixed ratio chunk of the
primary task stack. The -D option is used to define the
size of the environment task's secondary stack.

@item -e
@cindex @option{-e} (@command{gnatbind})
Output complete list of elaboration-order dependencies.

@item -E
@cindex @option{-E} (@command{gnatbind})
Store tracebacks in exception occurrences when the target supports it.
@ignore
@c The following may get moved to an appendix
This option is currently supported on the following targets:
all x86 ports, Solaris, Windows, HP-UX, AIX, PowerPC VxWorks and Alpha VxWorks.
@end ignore
See also the packages @code{GNAT.Traceback} and
@code{GNAT.Traceback.Symbolic} for more information.
Note that on x86 ports, you must not use @option{-fomit-frame-pointer}
@command{gcc} option.

@item -F
@cindex @option{-F} (@command{gnatbind})
Force the checks of elaboration flags. @command{gnatbind} does not normally
generate checks of elaboration flags for the main executable, except when
a Stand-Alone Library is used. However, there are cases when this cannot be
detected by gnatbind. An example is importing an interface of a Stand-Alone
Library through a pragma Import and only specifying through a linker switch
this Stand-Alone Library. This switch is used to guarantee that elaboration
flag checks are generated.

@item -h
@cindex @option{-h} (@command{gnatbind})
Output usage (help) information

@item -H32
@cindex @option{-H32} (@command{gnatbind})
Use 32-bit allocations for @code{__gnat_malloc} (and thus for access types).
For further details see @ref{Dynamic Allocation Control}.

@item -H64
@cindex @option{-H64} (@command{gnatbind})
Use 64-bit allocations for @code{__gnat_malloc} (and thus for access types).
@cindex @code{__gnat_malloc}
For further details see @ref{Dynamic Allocation Control}.

@item -I
@cindex @option{-I} (@command{gnatbind})
Specify directory to be searched for source and ALI files.

@item -I-
@cindex @option{-I-} (@command{gnatbind})
Do not look for sources in the current directory where @code{gnatbind} was
invoked, and do not look for ALI files in the directory containing the
ALI file named in the @code{gnatbind} command line.

@item -l
@cindex @option{-l} (@command{gnatbind})
Output chosen elaboration order.

@item -L@var{xxx}
@cindex @option{-L} (@command{gnatbind})
Bind the units for library building. In this case the adainit and
adafinal procedures (@pxref{Binding with Non-Ada Main Programs})
are renamed to @var{xxx}init and
@var{xxx}final.
Implies -n.
(@xref{GNAT and Libraries}, for more details.)

@item -Mxyz
@cindex @option{-M} (@command{gnatbind})
Rename generated main program from main to xyz. This option is
supported on cross environments only.

@item -m@var{n}
@cindex @option{-m} (@command{gnatbind})
Limit number of detected errors or warnings to @var{n}, where @var{n} is
in the range 1..999999. The default value if no switch is
given is 9999. If the number of warnings reaches this limit, then a
message is output and further warnings are suppressed, the bind
continues in this case. If the number of errors reaches this
limit, then a message is output and the bind is abandoned.
A value of zero means that no limit is enforced. The equal
sign is optional.

@item -n
@cindex @option{-n} (@command{gnatbind})
No main program.

@item -nostdinc
@cindex @option{-nostdinc} (@command{gnatbind})
Do not look for sources in the system default directory.

@item -nostdlib
@cindex @option{-nostdlib} (@command{gnatbind})
Do not look for library files in the system default directory.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@code{gnatbind})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item -o @var{file}
@cindex @option{-o } (@command{gnatbind})
Name the output file @var{file} (default is @file{b~@var{xxx}.adb}).
Note that if this option is used, then linking must be done manually,
gnatlink cannot be used.

@item -O@r{[=}@var{filename}@r{]}
@cindex @option{-O} (@command{gnatbind})
Output object list (to standard output or to the named file).

@item -p
@cindex @option{-p} (@command{gnatbind})
Pessimistic (worst-case) elaboration order

@item -P
@cindex @option{-P} (@command{gnatbind})
Generate binder file suitable for CodePeer.

@item -R
@cindex @option{-R} (@command{gnatbind})
Output closure source list, which includes all non-run-time units that are
included in the bind.

@item -Ra
@cindex @option{-Ra} (@command{gnatbind})
Like @option{-R} but the list includes run-time units.

@item -s
@cindex @option{-s} (@command{gnatbind})
Require all source files to be present.

@item -S@var{xxx}
@cindex @option{-S} (@command{gnatbind})
Specifies the value to be used when detecting uninitialized scalar
objects with pragma Initialize_Scalars.
The @var{xxx} string specified with the switch is one of:
@itemize @bullet

@item ``@option{in}'' for an invalid value
If zero is invalid for the discrete type in question,
then the scalar value is set to all zero bits.
For signed discrete types, the largest possible negative value of
the underlying scalar is set (i.e. a one bit followed by all zero bits).
For unsigned discrete types, the underlying scalar value is set to all
one bits. For floating-point types, a NaN value is set
(see body of package System.Scalar_Values for exact values).

@item ``@option{lo}'' for low value
If zero is invalid for the discrete type in question,
then the scalar value is set to all zero bits.
For signed discrete types, the largest possible negative value of
the underlying scalar is set (i.e. a one bit followed by all zero bits).
For unsigned discrete types, the underlying scalar value is set to all
zero bits. For floating-point, a small value is set
(see body of package System.Scalar_Values for exact values).

@item ``@option{hi}'' for high value
If zero is invalid for the discrete type in question,
then the scalar value is set to all one bits.
For signed discrete types, the largest possible positive value of
the underlying scalar is set (i.e. a zero bit followed by all one bits).
For unsigned discrete types, the underlying scalar value is set to all
one bits. For floating-point, a large value is set
(see body of package System.Scalar_Values for exact values).

@item ``@option{@var{xx}}'' for hex value (two hex digits)
The underlying scalar is set to a value consisting of repeated bytes, whose
value corresponds to the given value. For example if @option{BF} is given,
then a 32-bit scalar value will be set to the bit patterm 16#BFBFBFBF#.
@end itemize

In addition, you can specify @option{-Sev} to indicate that the value is
to be set at run time. In this case, the program will look for an environment
@cindex GNAT_INIT_SCALARS
variable of the form @env{GNAT_INIT_SCALARS=@var{xx}}, where @var{xx} is one
of @option{in/lo/hi/@var{xx}} with the same meanings as above.
If no environment variable is found, or if it does not have a valid value,
then the default is @option{in} (invalid values).

@item -static
@cindex @option{-static} (@code{gnatbind})
Link against a static GNAT run time.

@item -shared
@cindex @option{-shared} (@code{gnatbind})
Link against a shared GNAT run time when available.

@item -t
@cindex @option{-t} (@code{gnatbind})
Tolerate time stamp and other consistency errors

@item -T@var{n}
@cindex @option{-T} (@code{gnatbind})
Set the time slice value to @var{n} milliseconds. If the system supports
the specification of a specific time slice value, then the indicated value
is used. If the system does not support specific time slice values, but
does support some general notion of round-robin scheduling, then any
nonzero value will activate round-robin scheduling.

A value of zero is treated specially. It turns off time
slicing, and in addition, indicates to the tasking run time that the
semantics should match as closely as possible the Annex D
requirements of the Ada RM, and in particular sets the default
scheduling policy to @code{FIFO_Within_Priorities}.

@item -u@var{n}
@cindex @option{-u} (@code{gnatbind})
Enable dynamic stack usage, with @var{n} results stored and displayed
at program termination. A result is generated when a task
terminates. Results that can't be stored are displayed on the fly, at
task termination. This option is currently not supported on Itanium
platforms. (See @ref{Dynamic Stack Usage Analysis} for details.)

@item -v
@cindex @option{-v} (@code{gnatbind})
Verbose mode. Write error messages, header, summary output to
@file{stdout}.

@item -w@var{x}
@cindex @option{-w} (@code{gnatbind})
Warning mode (@var{x}=s/e for suppress/treat as error)


@item -Wx@var{e}
@cindex @option{-Wx} (@code{gnatbind})
Override default wide character encoding for standard Text_IO files.

@item -x
@cindex @option{-x} (@code{gnatbind})
Exclude source files (check object consistency only).


@item -X@var{nnn}
@cindex @option{-X@var{nnn}} (@code{gnatbind})
Set default exit status value, normally 0 for POSIX compliance.


@item -y
@cindex @option{-y} (@code{gnatbind})
Enable leap seconds support in @code{Ada.Calendar} and its children.

@item -z
@cindex @option{-z} (@code{gnatbind})
No main subprogram.
@end table

@noindent
You may obtain this listing of switches by running @code{gnatbind} with
no arguments.

@node Consistency-Checking Modes
@subsection Consistency-Checking Modes

@noindent
As described earlier, by default @code{gnatbind} checks
that object files are consistent with one another and are consistent
with any source files it can locate. The following switches control binder
access to sources.

@table @option
@c !sort!
@item -s
@cindex @option{-s} (@code{gnatbind})
Require source files to be present. In this mode, the binder must be
able to locate all source files that are referenced, in order to check
their consistency. In normal mode, if a source file cannot be located it
is simply ignored. If you specify this switch, a missing source
file is an error.

@item -Wx@var{e}
@cindex @option{-Wx} (@code{gnatbind})
Override default wide character encoding for standard Text_IO files.
Normally the default wide character encoding method used for standard
[Wide_[Wide_]]Text_IO files is taken from the encoding specified for
the main source input (see description of switch
@option{-gnatWx} for the compiler). The
use of this switch for the binder (which has the same set of
possible arguments) overrides this default as specified.

@item -x
@cindex @option{-x} (@code{gnatbind})
Exclude source files. In this mode, the binder only checks that ALI
files are consistent with one another. Source files are not accessed.
The binder runs faster in this mode, and there is still a guarantee that
the resulting program is self-consistent.
If a source file has been edited since it was last compiled, and you
specify this switch, the binder will not detect that the object
file is out of date with respect to the source file. Note that this is the
mode that is automatically used by @command{gnatmake} because in this
case the checking against sources has already been performed by
@command{gnatmake} in the course of compilation (i.e.@: before binding).

@end table

@node Binder Error Message Control
@subsection Binder Error Message Control

@noindent
The following switches provide control over the generation of error
messages from the binder:

@table @option
@c !sort!
@item -v
@cindex @option{-v} (@code{gnatbind})
Verbose mode. In the normal mode, brief error messages are generated to
@file{stderr}. If this switch is present, a header is written
to @file{stdout} and any error messages are directed to @file{stdout}.
All that is written to @file{stderr} is a brief summary message.

@item -b
@cindex @option{-b} (@code{gnatbind})
Generate brief error messages to @file{stderr} even if verbose mode is
specified. This is relevant only when used with the
@option{-v} switch.

@item -m@var{n}
@cindex @option{-m} (@code{gnatbind})
Limits the number of error messages to @var{n}, a decimal integer in the
range 1-999. The binder terminates immediately if this limit is reached.

@item -M@var{xxx}
@cindex @option{-M} (@code{gnatbind})
Renames the generated main program from @code{main} to @code{xxx}.
This is useful in the case of some cross-building environments, where
the actual main program is separate from the one generated
by @code{gnatbind}.

@item -ws
@cindex @option{-ws} (@code{gnatbind})
@cindex Warnings
Suppress all warning messages.

@item -we
@cindex @option{-we} (@code{gnatbind})
Treat any warning messages as fatal errors.


@item -t
@cindex @option{-t} (@code{gnatbind})
@cindex Time stamp checks, in binder
@cindex Binder consistency checks
@cindex Consistency checks, in binder
The binder performs a number of consistency checks including:

@itemize @bullet
@item
Check that time stamps of a given source unit are consistent
@item
Check that checksums of a given source unit are consistent
@item
Check that consistent versions of @code{GNAT} were used for compilation
@item
Check consistency of configuration pragmas as required
@end itemize

@noindent
Normally failure of such checks, in accordance with the consistency
requirements of the Ada Reference Manual, causes error messages to be
generated which abort the binder and prevent the output of a binder
file and subsequent link to obtain an executable.

The @option{-t} switch converts these error messages
into warnings, so that
binding and linking can continue to completion even in the presence of such
errors. The result may be a failed link (due to missing symbols), or a
non-functional executable which has undefined semantics.
@emph{This means that
@option{-t} should be used only in unusual situations,
with extreme care.}
@end table

@node Elaboration Control
@subsection Elaboration Control

@noindent
The following switches provide additional control over the elaboration
order. For full details see @ref{Elaboration Order Handling in GNAT}.

@table @option
@item -p
@cindex @option{-p} (@code{gnatbind})
Normally the binder attempts to choose an elaboration order that is
likely to minimize the likelihood of an elaboration order error resulting
in raising a @code{Program_Error} exception. This switch reverses the
action of the binder, and requests that it deliberately choose an order
that is likely to maximize the likelihood of an elaboration error.
This is useful in ensuring portability and avoiding dependence on
accidental fortuitous elaboration ordering.

Normally it only makes sense to use the @option{-p}
switch if dynamic
elaboration checking is used (@option{-gnatE} switch used for compilation).
This is because in the default static elaboration mode, all necessary
@code{Elaborate} and @code{Elaborate_All} pragmas are implicitly inserted.
These implicit pragmas are still respected by the binder in
@option{-p} mode, so a
safe elaboration order is assured.

Note that @option{-p} is not intended for
production use; it is more for debugging/experimental use.
@end table

@node Output Control
@subsection Output Control

@noindent
The following switches allow additional control over the output
generated by the binder.

@table @option
@c !sort!

@item -c
@cindex @option{-c} (@code{gnatbind})
Check only. Do not generate the binder output file. In this mode the
binder performs all error checks but does not generate an output file.

@item -e
@cindex @option{-e} (@code{gnatbind})
Output complete list of elaboration-order dependencies, showing the
reason for each dependency. This output can be rather extensive but may
be useful in diagnosing problems with elaboration order. The output is
written to @file{stdout}.

@item -h
@cindex @option{-h} (@code{gnatbind})
Output usage information. The output is written to @file{stdout}.

@item -K
@cindex @option{-K} (@code{gnatbind})
Output linker options to @file{stdout}. Includes library search paths,
contents of pragmas Ident and Linker_Options, and libraries added
by @code{gnatbind}.

@item -l
@cindex @option{-l} (@code{gnatbind})
Output chosen elaboration order. The output is written to @file{stdout}.

@item -O
@cindex @option{-O} (@code{gnatbind})
Output full names of all the object files that must be linked to provide
the Ada component of the program. The output is written to @file{stdout}.
This list includes the files explicitly supplied and referenced by the user
as well as implicitly referenced run-time unit files. The latter are
omitted if the corresponding units reside in shared libraries. The
directory names for the run-time units depend on the system configuration.

@item -o @var{file}
@cindex @option{-o} (@code{gnatbind})
Set name of output file to @var{file} instead of the normal
@file{b~@var{mainprog}.adb} default. Note that @var{file} denote the Ada
binder generated body filename.
Note that if this option is used, then linking must be done manually.
It is not possible to use gnatlink in this case, since it cannot locate
the binder file.

@item -r
@cindex @option{-r} (@code{gnatbind})
Generate list of @code{pragma Restrictions} that could be applied to
the current unit. This is useful for code audit purposes, and also may
be used to improve code generation in some cases.

@end table

@node Dynamic Allocation Control
@subsection Dynamic Allocation Control

@noindent
The heap control switches -- @option{-H32} and @option{-H64} --
determine whether dynamic allocation uses 32-bit or 64-bit memory.
They only affect compiler-generated allocations via @code{__gnat_malloc};
explicit calls to @code{malloc} and related functions from the C
run-time library are unaffected.

@table @option
@item -H32
Allocate memory on 32-bit heap

@item -H64
Allocate memory on 64-bit heap.  This is the default
unless explicitly overridden by a @code{'Size} clause on the access type.
@end table

@noindent
These switches are only effective on VMS platforms.


@node Binding with Non-Ada Main Programs
@subsection Binding with Non-Ada Main Programs

@noindent
In our description so far we have assumed that the main
program is in Ada, and that the task of the binder is to generate a
corresponding function @code{main} that invokes this Ada main
program. GNAT also supports the building of executable programs where
the main program is not in Ada, but some of the called routines are
written in Ada and compiled using GNAT (@pxref{Mixed Language Programming}).
The following switch is used in this situation:

@table @option
@item -n
@cindex @option{-n} (@code{gnatbind})
No main program. The main program is not in Ada.
@end table

@noindent
In this case, most of the functions of the binder are still required,
but instead of generating a main program, the binder generates a file
containing the following callable routines:

@table @code
@item adainit
@findex adainit
You must call this routine to initialize the Ada part of the program by
calling the necessary elaboration routines. A call to @code{adainit} is
required before the first call to an Ada subprogram.

Note that it is assumed that the basic execution environment must be setup
to be appropriate for Ada execution at the point where the first Ada
subprogram is called. In particular, if the Ada code will do any
floating-point operations, then the FPU must be setup in an appropriate
manner. For the case of the x86, for example, full precision mode is
required. The procedure GNAT.Float_Control.Reset may be used to ensure
that the FPU is in the right state.

@item adafinal
@findex adafinal
You must call this routine to perform any library-level finalization
required by the Ada subprograms. A call to @code{adafinal} is required
after the last call to an Ada subprogram, and before the program
terminates.
@end table

@noindent
If the @option{-n} switch
@cindex @option{-n} (@command{gnatbind})
@cindex Binder, multiple input files
is given, more than one ALI file may appear on
the command line for @code{gnatbind}. The normal @dfn{closure}
calculation is performed for each of the specified units. Calculating
the closure means finding out the set of units involved by tracing
@code{with} references. The reason it is necessary to be able to
specify more than one ALI file is that a given program may invoke two or
more quite separate groups of Ada units.

The binder takes the name of its output file from the last specified ALI
file, unless overridden by the use of the @option{-o file}.
@cindex @option{-o} (@command{gnatbind})
The output is an Ada unit in source form that can be compiled with GNAT.
This compilation occurs automatically as part of the @command{gnatlink}
processing.

Currently the GNAT run time requires a FPU using 80 bits mode
precision. Under targets where this is not the default it is required to
call GNAT.Float_Control.Reset before using floating point numbers (this
include float computation, float input and output) in the Ada code. A
side effect is that this could be the wrong mode for the foreign code
where floating point computation could be broken after this call.

@node Binding Programs with No Main Subprogram
@subsection Binding Programs with No Main Subprogram

@noindent
It is possible to have an Ada program which does not have a main
subprogram. This program will call the elaboration routines of all the
packages, then the finalization routines.

The following switch is used to bind programs organized in this manner:

@table @option
@item -z
@cindex @option{-z} (@code{gnatbind})
Normally the binder checks that the unit name given on the command line
corresponds to a suitable main subprogram. When this switch is used,
a list of ALI files can be given, and the execution of the program
consists of elaboration of these units in an appropriate order. Note
that the default wide character encoding method for standard Text_IO
files is always set to Brackets if this switch is set (you can use
the binder switch
@option{-Wx} to override this default).
@end table

@node Command-Line Access
@section Command-Line Access

@noindent
The package @code{Ada.Command_Line} provides access to the command-line
arguments and program name. In order for this interface to operate
correctly, the two variables

@smallexample
@group
int gnat_argc;
char **gnat_argv;
@end group
@end smallexample

@noindent
@findex gnat_argv
@findex gnat_argc
are declared in one of the GNAT library routines. These variables must
be set from the actual @code{argc} and @code{argv} values passed to the
main program. With no @option{n} present, @code{gnatbind}
generates the C main program to automatically set these variables.
If the @option{n} switch is used, there is no automatic way to
set these variables. If they are not set, the procedures in
@code{Ada.Command_Line} will not be available, and any attempt to use
them will raise @code{Constraint_Error}. If command line access is
required, your main program must set @code{gnat_argc} and
@code{gnat_argv} from the @code{argc} and @code{argv} values passed to
it.

@node Search Paths for gnatbind
@section Search Paths for @code{gnatbind}

@noindent
The binder takes the name of an ALI file as its argument and needs to
locate source files as well as other ALI files to verify object consistency.

For source files, it follows exactly the same search rules as @command{gcc}
(@pxref{Search Paths and the Run-Time Library (RTL)}). For ALI files the
directories searched are:

@enumerate
@item
The directory containing the ALI file named in the command line, unless
the switch @option{-I-} is specified.

@item
All directories specified by @option{-I}
switches on the @code{gnatbind}
command line, in the order given.

@item
@findex ADA_PRJ_OBJECTS_FILE
Each of the directories listed in the text file whose name is given
by the @env{ADA_PRJ_OBJECTS_FILE} environment variable.

@noindent
@env{ADA_PRJ_OBJECTS_FILE} is normally set by gnatmake or by the gnat
driver when project files are used. It should not normally be set
by other means.

@item
@findex ADA_OBJECTS_PATH
Each of the directories listed in the value of the
@env{ADA_OBJECTS_PATH} environment variable.
Construct this value
exactly as the @env{PATH} environment variable: a list of directory
names separated by colons (semicolons when working with the NT version
of GNAT).

@item
The content of the @file{ada_object_path} file which is part of the GNAT
installation tree and is used to store standard libraries such as the
GNAT Run Time Library (RTL) unless the switch @option{-nostdlib} is
specified.
@ref{Installing a library}
@end enumerate

@noindent
In the binder the switch @option{-I}
@cindex @option{-I} (@command{gnatbind})
is used to specify both source and
library file paths. Use @option{-aI}
@cindex @option{-aI} (@command{gnatbind})
instead if you want to specify
source paths only, and @option{-aO}
@cindex @option{-aO} (@command{gnatbind})
if you want to specify library paths
only. This means that for the binder
@option{-I}@var{dir} is equivalent to
@option{-aI}@var{dir}
@option{-aO}@var{dir}.
The binder generates the bind file (a C language source file) in the
current working directory.

@findex Ada
@findex System
@findex Interfaces
@findex GNAT
The packages @code{Ada}, @code{System}, and @code{Interfaces} and their
children make up the GNAT Run-Time Library, together with the package
GNAT and its children, which contain a set of useful additional
library functions provided by GNAT. The sources for these units are
needed by the compiler and are kept together in one directory. The ALI
files and object files generated by compiling the RTL are needed by the
binder and the linker and are kept together in one directory, typically
different from the directory containing the sources. In a normal
installation, you need not specify these directory names when compiling
or binding. Either the environment variables or the built-in defaults
cause these files to be found.

Besides simplifying access to the RTL, a major use of search paths is
in compiling sources from multiple directories. This can make
development environments much more flexible.

@node Examples of gnatbind Usage
@section Examples of @code{gnatbind} Usage

@noindent
This section contains a number of examples of using the GNAT binding
utility @code{gnatbind}.

@table @code
@item gnatbind hello
The main program @code{Hello} (source program in @file{hello.adb}) is
bound using the standard switch settings. The generated main program is
@file{b~hello.adb}. This is the normal, default use of the binder.

@item gnatbind hello -o mainprog.adb
The main program @code{Hello} (source program in @file{hello.adb}) is
bound using the standard switch settings. The generated main program is
@file{mainprog.adb} with the associated spec in
@file{mainprog.ads}. Note that you must specify the body here not the
spec. Note that if this option is used, then linking must be done manually,
since gnatlink will not be able to find the generated file.
@end table

@c ------------------------------------
@node Linking with gnatlink
@chapter Linking with @command{gnatlink}
@c ------------------------------------
@findex gnatlink

@noindent
This chapter discusses @command{gnatlink}, a tool that links
an Ada program and builds an executable file. This utility
invokes the system linker (via the @command{gcc} command)
with a correct list of object files and library references.
@command{gnatlink} automatically determines the list of files and
references for the Ada part of a program. It uses the binder file
generated by the @command{gnatbind} to determine this list.

Note: to invoke @code{gnatlink} with a project file, use the @code{gnat}
driver (see @ref{The GNAT Driver and Project Files}).

@menu
* Running gnatlink::
* Switches for gnatlink::
@end menu

@node Running gnatlink
@section Running @command{gnatlink}

@noindent
The form of the @command{gnatlink} command is

@smallexample
@c $ gnatlink @ovar{switches} @var{mainprog}@r{[}.ali@r{]}
@c            @ovar{non-Ada objects} @ovar{linker options}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatlink @r{[}@var{switches}@r{]} @var{mainprog}@r{[}.ali@r{]}
           @r{[}@var{non-Ada objects}@r{]} @r{[}@var{linker options}@r{]}

@end smallexample

@noindent
The arguments of @command{gnatlink} (switches, main @file{ALI} file,
non-Ada objects
or linker options) may be in any order, provided that no non-Ada object may
be mistaken for a main @file{ALI} file.
Any file name @file{F} without the @file{.ali}
extension will be taken as the main @file{ALI} file if a file exists
whose name is the concatenation of @file{F} and @file{.ali}.

@noindent
@file{@var{mainprog}.ali} references the ALI file of the main program.
The @file{.ali} extension of this file can be omitted. From this
reference, @command{gnatlink} locates the corresponding binder file
@file{b~@var{mainprog}.adb} and, using the information in this file along
with the list of non-Ada objects and linker options, constructs a
linker command file to create the executable.

The arguments other than the @command{gnatlink} switches and the main
@file{ALI} file are passed to the linker uninterpreted.
They typically include the names of
object files for units written in other languages than Ada and any library
references required to resolve references in any of these foreign language
units, or in @code{Import} pragmas in any Ada units.

@var{linker options} is an optional list of linker specific
switches.
The default linker called by gnatlink is @command{gcc} which in
turn calls the appropriate system linker.

One useful option for the linker is @option{-s}: it reduces the size of the
executable by removing all symbol table and relocation information from the
executable.

Standard options for the linker such as @option{-lmy_lib} or
@option{-Ldir} can be added as is.
For options that are not recognized by
@command{gcc} as linker options, use the @command{gcc} switches
@option{-Xlinker} or @option{-Wl,}.

Refer to the GCC documentation for
details.

Here is an example showing how to generate a linker map:

@smallexample
$ gnatlink my_prog -Wl,-Map,MAPFILE
@end smallexample

Using @var{linker options} it is possible to set the program stack and
heap size.
See @ref{Setting Stack Size from gnatlink} and
@ref{Setting Heap Size from gnatlink}.

@command{gnatlink} determines the list of objects required by the Ada
program and prepends them to the list of objects passed to the linker.
@command{gnatlink} also gathers any arguments set by the use of
@code{pragma Linker_Options} and adds them to the list of arguments
presented to the linker.


@node Switches for gnatlink
@section Switches for @command{gnatlink}

@noindent
The following switches are available with the @command{gnatlink} utility:

@table @option
@c !sort!

@item --version
@cindex @option{--version} @command{gnatlink}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatlink}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item -f
@cindex Command line length
@cindex @option{-f} (@command{gnatlink})
On some targets, the command line length is limited, and @command{gnatlink}
will generate a separate file for the linker if the list of object files
is too long.
The @option{-f} switch forces this file
to be generated even if
the limit is not exceeded. This is useful in some cases to deal with
special situations where the command line length is exceeded.

@item -g
@cindex Debugging information, including
@cindex @option{-g} (@command{gnatlink})
The option to include debugging information causes the Ada bind file (in
other words, @file{b~@var{mainprog}.adb}) to be compiled with
@option{-g}.
In addition, the binder does not delete the @file{b~@var{mainprog}.adb},
@file{b~@var{mainprog}.o} and @file{b~@var{mainprog}.ali} files.
Without @option{-g}, the binder removes these files by
default. The same procedure apply if a C bind file was generated using
@option{-C} @code{gnatbind} option, in this case the filenames
are @file{b_@var{mainprog}.c} and @file{b_@var{mainprog}.o}.

@item -n
@cindex @option{-n} (@command{gnatlink})
Do not compile the file generated by the binder. This may be used when
a link is rerun with different options, but there is no need to recompile
the binder file.

@item -v
@cindex @option{-v} (@command{gnatlink})
Causes additional information to be output, including a full list of the
included object files. This switch option is most useful when you want
to see what set of object files are being used in the link step.

@item -v -v
@cindex @option{-v -v} (@command{gnatlink})
Very verbose mode. Requests that the compiler operate in verbose mode when
it compiles the binder file, and that the system linker run in verbose mode.

@item -o @var{exec-name}
@cindex @option{-o} (@command{gnatlink})
@var{exec-name} specifies an alternate name for the generated
executable program. If this switch is omitted, the executable has the same
name as the main unit. For example, @code{gnatlink try.ali} creates
an executable called @file{try}.

@item -b @var{target}
@cindex @option{-b} (@command{gnatlink})
Compile your program to run on @var{target}, which is the name of a
system configuration. You must have a GNAT cross-compiler built if
@var{target} is not the same as your host system.

@item -B@var{dir}
@cindex @option{-B} (@command{gnatlink})
Load compiler executables (for example, @code{gnat1}, the Ada compiler)
from @var{dir} instead of the default location. Only use this switch
when multiple versions of the GNAT compiler are available.
@xref{Directory Options,,, gcc, The GNU Compiler Collection},
for further details. You would normally use the @option{-b} or
@option{-V} switch instead.

@item -M
When linking an executable, create a map file. The name of the map file
has the same name as the executable with extension ".map".

@item -M=mapfile
When linking an executable, create a map file. The name of the map file is
"mapfile".

@item --GCC=@var{compiler_name}
@cindex @option{--GCC=compiler_name} (@command{gnatlink})
Program used for compiling the binder file. The default is
@command{gcc}. You need to use quotes around @var{compiler_name} if
@code{compiler_name} contains spaces or other separator characters.
As an example @option{--GCC="foo -x -y"} will instruct @command{gnatlink} to
use @code{foo -x -y} as your compiler. Note that switch @option{-c} is always
inserted after your command name. Thus in the above example the compiler
command that will be used by @command{gnatlink} will be @code{foo -c -x -y}.
A limitation of this syntax is that the name and path name of the executable
itself must not include any embedded spaces. If the compiler executable is
different from the default one (gcc or <prefix>-gcc), then the back-end
switches in the ALI file are not used to compile the binder generated source.
For example, this is the case with @option{--GCC="foo -x -y"}. But the back end
switches will be used for @option{--GCC="gcc -gnatv"}. If several
@option{--GCC=compiler_name} are used, only the last @var{compiler_name}
is taken into account. However, all the additional switches are also taken
into account. Thus,
@option{--GCC="foo -x -y" --GCC="bar -z -t"} is equivalent to
@option{--GCC="bar -x -y -z -t"}.

@item --LINK=@var{name}
@cindex @option{--LINK=} (@command{gnatlink})
@var{name} is the name of the linker to be invoked. This is especially
useful in mixed language programs since languages such as C++ require
their own linker to be used. When this switch is omitted, the default
name for the linker is @command{gcc}. When this switch is used, the
specified linker is called instead of @command{gcc} with exactly the same
parameters that would have been passed to @command{gcc} so if the desired
linker requires different parameters it is necessary to use a wrapper
script that massages the parameters before invoking the real linker. It
may be useful to control the exact invocation by using the verbose
switch.



@end table

@node The GNAT Make Program gnatmake
@chapter The GNAT Make Program @command{gnatmake}
@findex gnatmake

@menu
* Running gnatmake::
* Switches for gnatmake::
* Mode Switches for gnatmake::
* Notes on the Command Line::
* How gnatmake Works::
* Examples of gnatmake Usage::
@end menu
@noindent
A typical development cycle when working on an Ada program consists of
the following steps:

@enumerate
@item
Edit some sources to fix bugs.

@item
Add enhancements.

@item
Compile all sources affected.

@item
Rebind and relink.

@item
Test.
@end enumerate

@noindent
The third step can be tricky, because not only do the modified files
@cindex Dependency rules
have to be compiled, but any files depending on these files must also be
recompiled. The dependency rules in Ada can be quite complex, especially
in the presence of overloading, @code{use} clauses, generics and inlined
subprograms.

@command{gnatmake} automatically takes care of the third and fourth steps
of this process. It determines which sources need to be compiled,
compiles them, and binds and links the resulting object files.

Unlike some other Ada make programs, the dependencies are always
accurately recomputed from the new sources. The source based approach of
the GNAT compilation model makes this possible. This means that if
changes to the source program cause corresponding changes in
dependencies, they will always be tracked exactly correctly by
@command{gnatmake}.

@node Running gnatmake
@section Running @command{gnatmake}

@noindent
The usual form of the @command{gnatmake} command is

@smallexample
@c $ gnatmake @ovar{switches} @var{file_name}
@c       @ovar{file_names} @ovar{mode_switches}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatmake @r{[}@var{switches}@r{]} @var{file_name}
      @r{[}@var{file_names}@r{]} @r{[}@var{mode_switches}@r{]}
@end smallexample

@noindent
The only required argument is one @var{file_name}, which specifies
a compilation unit that is a main program. Several @var{file_names} can be
specified: this will result in several executables being built.
If @code{switches} are present, they can be placed before the first
@var{file_name}, between @var{file_names} or after the last @var{file_name}.
If @var{mode_switches} are present, they must always be placed after
the last @var{file_name} and all @code{switches}.

If you are using standard file extensions (@file{.adb} and @file{.ads}), then the
extension may be omitted from the @var{file_name} arguments. However, if
you are using non-standard extensions, then it is required that the
extension be given. A relative or absolute directory path can be
specified in a @var{file_name}, in which case, the input source file will
be searched for in the specified directory only. Otherwise, the input
source file will first be searched in the directory where
@command{gnatmake} was invoked and if it is not found, it will be search on
the source path of the compiler as described in
@ref{Search Paths and the Run-Time Library (RTL)}.

All @command{gnatmake} output (except when you specify
@option{-M}) is to
@file{stderr}. The output produced by the
@option{-M} switch is send to
@file{stdout}.

@node Switches for gnatmake
@section Switches for @command{gnatmake}

@noindent
You may specify any of the following switches to @command{gnatmake}:

@table @option
@c !sort!

@item --version
@cindex @option{--version} @command{gnatmake}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatmake}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item --GCC=@var{compiler_name}
@cindex @option{--GCC=compiler_name} (@command{gnatmake})
Program used for compiling. The default is `@command{gcc}'. You need to use
quotes around @var{compiler_name} if @code{compiler_name} contains
spaces or other separator characters. As an example @option{--GCC="foo -x
-y"} will instruct @command{gnatmake} to use @code{foo -x -y} as your
compiler. A limitation of this syntax is that the name and path name of
the executable itself must not include any embedded spaces. Note that
switch @option{-c} is always inserted after your command name. Thus in the
above example the compiler command that will be used by @command{gnatmake}
will be @code{foo -c -x -y}. If several @option{--GCC=compiler_name} are
used, only the last @var{compiler_name} is taken into account. However,
all the additional switches are also taken into account. Thus,
@option{--GCC="foo -x -y" --GCC="bar -z -t"} is equivalent to
@option{--GCC="bar -x -y -z -t"}.

@item --GNATBIND=@var{binder_name}
@cindex @option{--GNATBIND=binder_name} (@command{gnatmake})
Program used for binding. The default is `@code{gnatbind}'. You need to
use quotes around @var{binder_name} if @var{binder_name} contains spaces
or other separator characters. As an example @option{--GNATBIND="bar -x
-y"} will instruct @command{gnatmake} to use @code{bar -x -y} as your
binder. Binder switches that are normally appended by @command{gnatmake}
to `@code{gnatbind}' are now appended to the end of @code{bar -x -y}.
A limitation of this syntax is that the name and path name of the executable
itself must not include any embedded spaces.

@item --GNATLINK=@var{linker_name}
@cindex @option{--GNATLINK=linker_name} (@command{gnatmake})
Program used for linking. The default is `@command{gnatlink}'. You need to
use quotes around @var{linker_name} if @var{linker_name} contains spaces
or other separator characters. As an example @option{--GNATLINK="lan -x
-y"} will instruct @command{gnatmake} to use @code{lan -x -y} as your
linker. Linker switches that are normally appended by @command{gnatmake} to
`@command{gnatlink}' are now appended to the end of @code{lan -x -y}.
A limitation of this syntax is that the name and path name of the executable
itself must not include any embedded spaces.


@item --subdirs=subdir
Actual object directory of each project file is the subdirectory subdir of the
object directory specified or defaulted in the project file.

@item --single-compile-per-obj-dir
Disallow simultaneous compilations in the same object directory when
project files are used.

@item --unchecked-shared-lib-imports
By default, shared library projects are not allowed to import static library
projects. When this switch is used on the command line, this restriction is
relaxed.

@item --source-info=<source info file>
Specify a source info file. This switch is active only when project files
are used. If the source info file is specified as a relative path, then it is
relative to the object directory of the main project. If the source info file
does not exist, then after the Project Manager has successfully parsed and
processed the project files and found the sources, it creates the source info
file. If the source info file already exists and can be read successfully,
then the Project Manager will get all the needed information about the sources
from the source info file and will not look for them. This reduces the time
to process the project files, especially when looking for sources that take a
long time. If the source info file exists but cannot be parsed successfully,
the Project Manager will attempt to recreate it. If the Project Manager fails
to create the source info file, a message is issued, but gnatmake does not
fail. @command{gnatmake} "trusts" the source info file. This means that
if the source files have changed (addition, deletion, moving to a different
source directory), then the source info file need to be deleted and recreated.

@item --create-map-file
When linking an executable, create a map file. The name of the map file
has the same name as the executable with extension ".map".

@item --create-map-file=mapfile
When linking an executable, create a map file. The name of the map file is
"mapfile".


@item -a
@cindex @option{-a} (@command{gnatmake})
Consider all files in the make process, even the GNAT internal system
files (for example, the predefined Ada library files), as well as any
locked files. Locked files are files whose ALI file is write-protected.
By default,
@command{gnatmake} does not check these files,
because the assumption is that the GNAT internal files are properly up
to date, and also that any write protected ALI files have been properly
installed. Note that if there is an installation problem, such that one
of these files is not up to date, it will be properly caught by the
binder.
You may have to specify this switch if you are working on GNAT
itself. The switch @option{-a} is also useful
in conjunction with @option{-f}
if you need to recompile an entire application,
including run-time files, using special configuration pragmas,
such as a @code{Normalize_Scalars} pragma.

By default
@code{gnatmake -a} compiles all GNAT
internal files with
@code{gcc -c -gnatpg} rather than @code{gcc -c}.

@item -b
@cindex @option{-b} (@command{gnatmake})
Bind only. Can be combined with @option{-c} to do
compilation and binding, but no link.
Can be combined with @option{-l}
to do binding and linking. When not combined with
@option{-c}
all the units in the closure of the main program must have been previously
compiled and must be up to date. The root unit specified by @var{file_name}
may be given without extension, with the source extension or, if no GNAT
Project File is specified, with the ALI file extension.

@item -c
@cindex @option{-c} (@command{gnatmake})
Compile only. Do not perform binding, except when @option{-b}
is also specified. Do not perform linking, except if both
@option{-b} and
@option{-l} are also specified.
If the root unit specified by @var{file_name} is not a main unit, this is the
default. Otherwise @command{gnatmake} will attempt binding and linking
unless all objects are up to date and the executable is more recent than
the objects.

@item -C
@cindex @option{-C} (@command{gnatmake})
Use a temporary mapping file. A mapping file is a way to communicate
to the compiler two mappings: from unit names to file names (without
any directory information) and from file names to path names (with
full directory information). A mapping file can make the compiler's
file searches faster, especially if there are many source directories,
or the sources are read over a slow network connection. If
@option{-P} is used, a mapping file is always used, so
@option{-C} is unnecessary; in this case the mapping file
is initially populated based on the project file. If
@option{-C} is used without
@option{-P},
the mapping file is initially empty. Each invocation of the compiler
will add any newly accessed sources to the mapping file.

@item -C=@var{file}
@cindex @option{-C=} (@command{gnatmake})
Use a specific mapping file. The file, specified as a path name (absolute or
relative) by this switch, should already exist, otherwise the switch is
ineffective. The specified mapping file will be communicated to the compiler.
This switch is not compatible with a project file
(-P@var{file}) or with multiple compiling processes
(-jnnn, when nnn is greater than 1).

@item -d
@cindex @option{-d} (@command{gnatmake})
Display progress for each source, up to date or not, as a single line

@smallexample
completed x out of y (zz%)
@end smallexample

If the file needs to be compiled this is displayed after the invocation of
the compiler. These lines are displayed even in quiet output mode.

@item -D @var{dir}
@cindex @option{-D} (@command{gnatmake})
Put all object files and ALI file in directory @var{dir}.
If the @option{-D} switch is not used, all object files
and ALI files go in the current working directory.

This switch cannot be used when using a project file.

@item -eInnn
@cindex @option{-eI} (@command{gnatmake})
Indicates that the main source is a multi-unit source and the rank of the unit
in the source file is nnn. nnn needs to be a positive number and a valid
index in the source. This switch cannot be used when @command{gnatmake} is
invoked for several mains.

@item -eL
@cindex @option{-eL} (@command{gnatmake})
@cindex symbolic links
Follow all symbolic links when processing project files.
This should be used if your project uses symbolic links for files or
directories, but is not needed in other cases.

@cindex naming scheme
This also assumes that no directory matches the naming scheme for files (for
instance that you do not have a directory called "sources.ads" when using the
default GNAT naming scheme).

When you do not have to use this switch (i.e.@: by default), gnatmake is able to
save a lot of system calls (several per source file and object file), which
can result in a significant speed up to load and manipulate a project file,
especially when using source files from a remote system.


@item -eS
@cindex @option{-eS} (@command{gnatmake})
Output the commands for the compiler, the binder and the linker
on standard output,
instead of standard error.

@item -f
@cindex @option{-f} (@command{gnatmake})
Force recompilations. Recompile all sources, even though some object
files may be up to date, but don't recompile predefined or GNAT internal
files or locked files (files with a write-protected ALI file),
unless the @option{-a} switch is also specified.

@item -F
@cindex @option{-F} (@command{gnatmake})
When using project files, if some errors or warnings are detected during
parsing and verbose mode is not in effect (no use of switch
-v), then error lines start with the full path name of the project
file, rather than its simple file name.

@item -g
@cindex @option{-g} (@command{gnatmake})
Enable debugging. This switch is simply passed to the compiler and to the
linker.

@item -i
@cindex @option{-i} (@command{gnatmake})
In normal mode, @command{gnatmake} compiles all object files and ALI files
into the current directory. If the @option{-i} switch is used,
then instead object files and ALI files that already exist are overwritten
in place. This means that once a large project is organized into separate
directories in the desired manner, then @command{gnatmake} will automatically
maintain and update this organization. If no ALI files are found on the
Ada object path (@ref{Search Paths and the Run-Time Library (RTL)}),
the new object and ALI files are created in the
directory containing the source being compiled. If another organization
is desired, where objects and sources are kept in different directories,
a useful technique is to create dummy ALI files in the desired directories.
When detecting such a dummy file, @command{gnatmake} will be forced to
recompile the corresponding source file, and it will be put the resulting
object and ALI files in the directory where it found the dummy file.

@item -j@var{n}
@cindex @option{-j} (@command{gnatmake})
@cindex Parallel make
Use @var{n} processes to carry out the (re)compilations. On a multiprocessor
machine compilations will occur in parallel. If @var{n} is 0, then the
maximum number of parallel compilations is the number of core processors
on the platform. In the event of compilation errors, messages from various
compilations might get interspersed (but @command{gnatmake} will give you the
full ordered list of failing compiles at the end). If this is problematic,
rerun the make process with n set to 1 to get a clean list of messages.

@item -k
@cindex @option{-k} (@command{gnatmake})
Keep going. Continue as much as possible after a compilation error. To
ease the programmer's task in case of compilation errors, the list of
sources for which the compile fails is given when @command{gnatmake}
terminates.

If @command{gnatmake} is invoked with several @file{file_names} and with this
switch, if there are compilation errors when building an executable,
@command{gnatmake} will not attempt to build the following executables.

@item -l
@cindex @option{-l} (@command{gnatmake})
Link only. Can be combined with @option{-b} to binding
and linking. Linking will not be performed if combined with
@option{-c}
but not with @option{-b}.
When not combined with @option{-b}
all the units in the closure of the main program must have been previously
compiled and must be up to date, and the main program needs to have been bound.
The root unit specified by @var{file_name}
may be given without extension, with the source extension or, if no GNAT
Project File is specified, with the ALI file extension.

@item -m
@cindex @option{-m} (@command{gnatmake})
Specify that the minimum necessary amount of recompilations
be performed. In this mode @command{gnatmake} ignores time
stamp differences when the only
modifications to a source file consist in adding/removing comments,
empty lines, spaces or tabs. This means that if you have changed the
comments in a source file or have simply reformatted it, using this
switch will tell @command{gnatmake} not to recompile files that depend on it
(provided other sources on which these files depend have undergone no
semantic modifications). Note that the debugging information may be
out of date with respect to the sources if the @option{-m} switch causes
a compilation to be switched, so the use of this switch represents a
trade-off between compilation time and accurate debugging information.

@item -M
@cindex Dependencies, producing list
@cindex @option{-M} (@command{gnatmake})
Check if all objects are up to date. If they are, output the object
dependences to @file{stdout} in a form that can be directly exploited in
a @file{Makefile}. By default, each source file is prefixed with its
(relative or absolute) directory name. This name is whatever you
specified in the various @option{-aI}
and @option{-I} switches. If you use
@code{gnatmake -M}
@option{-q}
(see below), only the source file names,
without relative paths, are output. If you just specify the
@option{-M}
switch, dependencies of the GNAT internal system files are omitted. This
is typically what you want. If you also specify
the @option{-a} switch,
dependencies of the GNAT internal files are also listed. Note that
dependencies of the objects in external Ada libraries (see switch
@option{-aL}@var{dir} in the following list)
are never reported.

@item -n
@cindex @option{-n} (@command{gnatmake})
Don't compile, bind, or link. Checks if all objects are up to date.
If they are not, the full name of the first file that needs to be
recompiled is printed.
Repeated use of this option, followed by compiling the indicated source
file, will eventually result in recompiling all required units.

@item -o @var{exec_name}
@cindex @option{-o} (@command{gnatmake})
Output executable name. The name of the final executable program will be
@var{exec_name}. If the @option{-o} switch is omitted the default
name for the executable will be the name of the input file in appropriate form
for an executable file on the host system.

This switch cannot be used when invoking @command{gnatmake} with several
@file{file_names}.

@item -p or --create-missing-dirs
@cindex @option{-p} (@command{gnatmake})
When using project files (-P@var{project}), create
automatically missing object directories, library directories and exec
directories.

@item -P@var{project}
@cindex @option{-P} (@command{gnatmake})
Use project file @var{project}. Only one such switch can be used.
@xref{gnatmake and Project Files}.

@item -q
@cindex @option{-q} (@command{gnatmake})
Quiet. When this flag is not set, the commands carried out by
@command{gnatmake} are displayed.

@item -s
@cindex @option{-s} (@command{gnatmake})
Recompile if compiler switches have changed since last compilation.
All compiler switches but -I and -o are taken into account in the
following way:
orders between different ``first letter'' switches are ignored, but
orders between same switches are taken into account. For example,
@option{-O -O2} is different than @option{-O2 -O}, but @option{-g -O}
is equivalent to @option{-O -g}.

This switch is recommended when Integrated Preprocessing is used.

@item -u
@cindex @option{-u} (@command{gnatmake})
Unique. Recompile at most the main files. It implies -c. Combined with
-f, it is equivalent to calling the compiler directly. Note that using
-u with a project file and no main has a special meaning
(@pxref{Project Files and Main Subprograms}).

@item -U
@cindex @option{-U} (@command{gnatmake})
When used without a project file or with one or several mains on the command
line, is equivalent to -u. When used with a project file and no main
on the command line, all sources of all project files are checked and compiled
if not up to date, and libraries are rebuilt, if necessary.

@item -v
@cindex @option{-v} (@command{gnatmake})
Verbose. Display the reason for all recompilations @command{gnatmake}
decides are necessary, with the highest verbosity level.

@item -vl
@cindex @option{-vl} (@command{gnatmake})
Verbosity level Low. Display fewer lines than in verbosity Medium.

@item -vm
@cindex @option{-vm} (@command{gnatmake})
Verbosity level Medium. Potentially display fewer lines than in verbosity High.

@item -vh
@cindex @option{-vm} (@command{gnatmake})
Verbosity level High. Equivalent to -v.

@item -vP@emph{x}
Indicate the verbosity of the parsing of GNAT project files.
@xref{Switches Related to Project Files}.

@item -x
@cindex @option{-x} (@command{gnatmake})
Indicate that sources that are not part of any Project File may be compiled.
Normally, when using Project Files, only sources that are part of a Project
File may be compile. When this switch is used, a source outside of all Project
Files may be compiled. The ALI file and the object file will be put in the
object directory of the main Project. The compilation switches used will only
be those specified on the command line. Even when
@option{-x} is used, mains specified on the
command line need to be sources of a project file.

@item -X@var{name=value}
Indicate that external variable @var{name} has the value @var{value}.
The Project Manager will use this value for occurrences of
@code{external(name)} when parsing the project file.
@xref{Switches Related to Project Files}.

@item -z
@cindex @option{-z} (@command{gnatmake})
No main subprogram. Bind and link the program even if the unit name
given on the command line is a package name. The resulting executable
will execute the elaboration routines of the package and its closure,
then the finalization routines.

@end table

@table @asis
@item @command{gcc} @asis{switches}
Any uppercase or multi-character switch that is not a @command{gnatmake} switch
is passed to @command{gcc} (e.g.@: @option{-O}, @option{-gnato,} etc.)
@end table

@noindent
Source and library search path switches:

@table @option
@c !sort!
@item -aI@var{dir}
@cindex @option{-aI} (@command{gnatmake})
When looking for source files also look in directory @var{dir}.
The order in which source files search is undertaken is
described in @ref{Search Paths and the Run-Time Library (RTL)}.

@item -aL@var{dir}
@cindex @option{-aL} (@command{gnatmake})
Consider @var{dir} as being an externally provided Ada library.
Instructs @command{gnatmake} to skip compilation units whose @file{.ALI}
files have been located in directory @var{dir}. This allows you to have
missing bodies for the units in @var{dir} and to ignore out of date bodies
for the same units. You still need to specify
the location of the specs for these units by using the switches
@option{-aI@var{dir}}
or @option{-I@var{dir}}.
Note: this switch is provided for compatibility with previous versions
of @command{gnatmake}. The easier method of causing standard libraries
to be excluded from consideration is to write-protect the corresponding
ALI files.

@item -aO@var{dir}
@cindex @option{-aO} (@command{gnatmake})
When searching for library and object files, look in directory
@var{dir}. The order in which library files are searched is described in
@ref{Search Paths for gnatbind}.

@item -A@var{dir}
@cindex Search paths, for @command{gnatmake}
@cindex @option{-A} (@command{gnatmake})
Equivalent to @option{-aL@var{dir}
-aI@var{dir}}.

@item -I@var{dir}
@cindex @option{-I} (@command{gnatmake})
Equivalent to @option{-aO@var{dir}
-aI@var{dir}}.

@item -I-
@cindex @option{-I-} (@command{gnatmake})
@cindex Source files, suppressing search
Do not look for source files in the directory containing the source
file named in the command line.
Do not look for ALI or object files in the directory
where @command{gnatmake} was invoked.

@item -L@var{dir}
@cindex @option{-L} (@command{gnatmake})
@cindex Linker libraries
Add directory @var{dir} to the list of directories in which the linker
will search for libraries. This is equivalent to
@option{-largs -L}@var{dir}.
Furthermore, under Windows, the sources pointed to by the libraries path
set in the registry are not searched for.

@item -nostdinc
@cindex @option{-nostdinc} (@command{gnatmake})
Do not look for source files in the system default directory.

@item -nostdlib
@cindex @option{-nostdlib} (@command{gnatmake})
Do not look for library files in the system default directory.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gnatmake})
Specifies the default location of the runtime library. GNAT looks for the
runtime
in the following directories, and stops as soon as a valid runtime is found
(@file{adainclude} or @file{ada_source_path}, and @file{adalib} or
@file{ada_object_path} present):

@itemize @bullet
@item <current directory>/$rts_path

@item <default-search-dir>/$rts_path

@item <default-search-dir>/rts-$rts_path
@end itemize

@noindent
The selected path is handled like a normal RTS path.

@end table

@node Mode Switches for gnatmake
@section Mode Switches for @command{gnatmake}

@noindent
The mode switches (referred to as @code{mode_switches}) allow the
inclusion of switches that are to be passed to the compiler itself, the
binder or the linker. The effect of a mode switch is to cause all
subsequent switches up to the end of the switch list, or up to the next
mode switch, to be interpreted as switches to be passed on to the
designated component of GNAT.

@table @option
@c !sort!
@item -cargs @var{switches}
@cindex @option{-cargs} (@command{gnatmake})
Compiler switches. Here @var{switches} is a list of switches
that are valid switches for @command{gcc}. They will be passed on to
all compile steps performed by @command{gnatmake}.

@item -bargs @var{switches}
@cindex @option{-bargs} (@command{gnatmake})
Binder switches. Here @var{switches} is a list of switches
that are valid switches for @code{gnatbind}. They will be passed on to
all bind steps performed by @command{gnatmake}.

@item -largs @var{switches}
@cindex @option{-largs} (@command{gnatmake})
Linker switches. Here @var{switches} is a list of switches
that are valid switches for @command{gnatlink}. They will be passed on to
all link steps performed by @command{gnatmake}.

@item -margs @var{switches}
@cindex @option{-margs} (@command{gnatmake})
Make switches. The switches are directly interpreted by @command{gnatmake},
regardless of any previous occurrence of @option{-cargs}, @option{-bargs}
or @option{-largs}.
@end table

@node Notes on the Command Line
@section Notes on the Command Line

@noindent
This section contains some additional useful notes on the operation
of the @command{gnatmake} command.

@itemize @bullet
@item
@cindex Recompilation, by @command{gnatmake}
If @command{gnatmake} finds no ALI files, it recompiles the main program
and all other units required by the main program.
This means that @command{gnatmake}
can be used for the initial compile, as well as during subsequent steps of
the development cycle.

@item
If you enter @code{gnatmake @var{file}.adb}, where @file{@var{file}.adb}
is a subunit or body of a generic unit, @command{gnatmake} recompiles
@file{@var{file}.adb} (because it finds no ALI) and stops, issuing a
warning.

@item
In @command{gnatmake} the switch @option{-I}
is used to specify both source and
library file paths. Use @option{-aI}
instead if you just want to specify
source paths only and @option{-aO}
if you want to specify library paths
only.

@item
@command{gnatmake} will ignore any files whose ALI file is write-protected.
This may conveniently be used to exclude standard libraries from
consideration and in particular it means that the use of the
@option{-f} switch will not recompile these files
unless @option{-a} is also specified.

@item
@command{gnatmake} has been designed to make the use of Ada libraries
particularly convenient. Assume you have an Ada library organized
as follows: @i{obj-dir} contains the objects and ALI files for
of your Ada compilation units,
whereas @i{include-dir} contains the
specs of these units, but no bodies. Then to compile a unit
stored in @code{main.adb}, which uses this Ada library you would just type

@smallexample
$ gnatmake -aI@var{include-dir}  -aL@var{obj-dir}  main
@end smallexample

@item
Using @command{gnatmake} along with the
@option{-m (minimal recompilation)}
switch provides a mechanism for avoiding unnecessary recompilations. Using
this switch,
you can update the comments/format of your
source files without having to recompile everything. Note, however, that
adding or deleting lines in a source files may render its debugging
info obsolete. If the file in question is a spec, the impact is rather
limited, as that debugging info will only be useful during the
elaboration phase of your program. For bodies the impact can be more
significant. In all events, your debugger will warn you if a source file
is more recent than the corresponding object, and alert you to the fact
that the debugging information may be out of date.
@end itemize

@node How gnatmake Works
@section How @command{gnatmake} Works

@noindent
Generally @command{gnatmake} automatically performs all necessary
recompilations and you don't need to worry about how it works. However,
it may be useful to have some basic understanding of the @command{gnatmake}
approach and in particular to understand how it uses the results of
previous compilations without incorrectly depending on them.

First a definition: an object file is considered @dfn{up to date} if the
corresponding ALI file exists and if all the source files listed in the
dependency section of this ALI file have time stamps matching those in
the ALI file. This means that neither the source file itself nor any
files that it depends on have been modified, and hence there is no need
to recompile this file.

@command{gnatmake} works by first checking if the specified main unit is up
to date. If so, no compilations are required for the main unit. If not,
@command{gnatmake} compiles the main program to build a new ALI file that
reflects the latest sources. Then the ALI file of the main unit is
examined to find all the source files on which the main program depends,
and @command{gnatmake} recursively applies the above procedure on all these
files.

This process ensures that @command{gnatmake} only trusts the dependencies
in an existing ALI file if they are known to be correct. Otherwise it
always recompiles to determine a new, guaranteed accurate set of
dependencies. As a result the program is compiled ``upside down'' from what may
be more familiar as the required order of compilation in some other Ada
systems. In particular, clients are compiled before the units on which
they depend. The ability of GNAT to compile in any order is critical in
allowing an order of compilation to be chosen that guarantees that
@command{gnatmake} will recompute a correct set of new dependencies if
necessary.

When invoking @command{gnatmake} with several @var{file_names}, if a unit is
imported by several of the executables, it will be recompiled at most once.

Note: when using non-standard naming conventions
(@pxref{Using Other File Names}), changing through a configuration pragmas
file the version of a source and invoking @command{gnatmake} to recompile may
have no effect, if the previous version of the source is still accessible
by @command{gnatmake}. It may be necessary to use the switch
-f.

@node Examples of gnatmake Usage
@section Examples of @command{gnatmake} Usage

@table @code
@item gnatmake hello.adb
Compile all files necessary to bind and link the main program
@file{hello.adb} (containing unit @code{Hello}) and bind and link the
resulting object files to generate an executable file @file{hello}.

@item gnatmake main1 main2 main3
Compile all files necessary to bind and link the main programs
@file{main1.adb} (containing unit @code{Main1}), @file{main2.adb}
(containing unit @code{Main2}) and @file{main3.adb}
(containing unit @code{Main3}) and bind and link the resulting object files
to generate three executable files @file{main1},
@file{main2}
and @file{main3}.

@item gnatmake -q Main_Unit -cargs -O2 -bargs -l

Compile all files necessary to bind and link the main program unit
@code{Main_Unit} (from file @file{main_unit.adb}). All compilations will
be done with optimization level 2 and the order of elaboration will be
listed by the binder. @command{gnatmake} will operate in quiet mode, not
displaying commands it is executing.
@end table

@c *************************
@node Improving Performance
@chapter Improving Performance
@cindex Improving performance

@noindent
This chapter presents several topics related to program performance.
It first describes some of the tradeoffs that need to be considered
and some of the techniques for making your program run faster.
It then documents
@ifclear FSFEDITION
the @command{gnatelim} tool and
@end ifclear
unused subprogram/data
elimination feature, which can reduce the size of program executables.

@ifnottex
@menu
* Performance Considerations::
* Text_IO Suggestions::
@ifclear FSFEDITION
* Reducing Size of Ada Executables with gnatelim::
@end ifclear
* Reducing Size of Executables with unused subprogram/data elimination::
@end menu
@end ifnottex

@c *****************************
@node Performance Considerations
@section Performance Considerations

@noindent
The GNAT system provides a number of options that allow a trade-off
between

@itemize @bullet
@item
performance of the generated code

@item
speed of compilation

@item
minimization of dependences and recompilation

@item
the degree of run-time checking.
@end itemize

@noindent
The defaults (if no options are selected) aim at improving the speed
of compilation and minimizing dependences, at the expense of performance
of the generated code:

@itemize @bullet
@item
no optimization

@item
no inlining of subprogram calls

@item
all run-time checks enabled except overflow and elaboration checks
@end itemize

@noindent
These options are suitable for most program development purposes. This
chapter describes how you can modify these choices, and also provides
some guidelines on debugging optimized code.

@menu
* Controlling Run-Time Checks::
* Use of Restrictions::
* Optimization Levels::
* Debugging Optimized Code::
* Inlining of Subprograms::
* Vectorization of loops::
* Other Optimization Switches::
* Optimization and Strict Aliasing::
* Aliased Variables and Optimization::
* Atomic Variables and Optimization::
* Passive Task Optimization::

@end menu

@node Controlling Run-Time Checks
@subsection Controlling Run-Time Checks

@noindent
By default, GNAT generates all run-time checks, except integer overflow
checks, stack overflow checks, and checks for access before elaboration on
subprogram calls. The latter are not required in default mode, because all
necessary checking is done at compile time.
@cindex @option{-gnatp} (@command{gcc})
@cindex @option{-gnato} (@command{gcc})
Two gnat switches, @option{-gnatp} and @option{-gnato} allow this default to
be modified. @xref{Run-Time Checks}.

Our experience is that the default is suitable for most development
purposes.

We treat integer overflow specially because these
are quite expensive and in our experience are not as important as other
run-time checks in the development process. Note that division by zero
is not considered an overflow check, and divide by zero checks are
generated where required by default.

Elaboration checks are off by default, and also not needed by default, since
GNAT uses a static elaboration analysis approach that avoids the need for
run-time checking. This manual contains a full chapter discussing the issue
of elaboration checks, and if the default is not satisfactory for your use,
you should read this chapter.

For validity checks, the minimal checks required by the Ada Reference
Manual (for case statements and assignments to array elements) are on
by default. These can be suppressed by use of the @option{-gnatVn} switch.
Note that in Ada 83, there were no validity checks, so if the Ada 83 mode
is acceptable (or when comparing GNAT performance with an Ada 83 compiler),
it may be reasonable to routinely use @option{-gnatVn}. Validity checks
are also suppressed entirely if @option{-gnatp} is used.

@cindex Overflow checks
@cindex Checks, overflow
@findex Suppress
@findex Unsuppress
@cindex pragma Suppress
@cindex pragma Unsuppress
Note that the setting of the switches controls the default setting of
the checks. They may be modified using either @code{pragma Suppress} (to
remove checks) or @code{pragma Unsuppress} (to add back suppressed
checks) in the program source.

@node Use of Restrictions
@subsection Use of Restrictions

@noindent
The use of pragma Restrictions allows you to control which features are
permitted in your program. Apart from the obvious point that if you avoid
relatively expensive features like finalization (enforceable by the use
of pragma Restrictions (No_Finalization), the use of this pragma does not
affect the generated code in most cases.

One notable exception to this rule is that the possibility of task abort
results in some distributed overhead, particularly if finalization or
exception handlers are used. The reason is that certain sections of code
have to be marked as non-abortable.

If you use neither the @code{abort} statement, nor asynchronous transfer
of control (@code{select @dots{} then abort}), then this distributed overhead
is removed, which may have a general positive effect in improving
overall performance.  Especially code involving frequent use of tasking
constructs and controlled types will show much improved performance.
The relevant restrictions pragmas are

@smallexample @c ada
   @b{pragma} Restrictions (No_Abort_Statements);
   @b{pragma} Restrictions (Max_Asynchronous_Select_Nesting => 0);
@end smallexample

@noindent
It is recommended that these restriction pragmas be used if possible. Note
that this also means that you can write code without worrying about the
possibility of an immediate abort at any point.

@node Optimization Levels
@subsection Optimization Levels
@cindex @option{-O} (@command{gcc})

@noindent
Without any optimization option,
the compiler's goal is to reduce the cost of
compilation and to make debugging produce the expected results.
Statements are independent: if you stop the program with a breakpoint between
statements, you can then assign a new value to any variable or change
the program counter to any other statement in the subprogram and get exactly
the results you would expect from the source code.

Turning on optimization makes the compiler attempt to improve the
performance and/or code size at the expense of compilation time and
possibly the ability to debug the program.

If you use multiple
-O options, with or without level numbers,
the last such option is the one that is effective.

@noindent
The default is optimization off. This results in the fastest compile
times, but GNAT makes absolutely no attempt to optimize, and the
generated programs are considerably larger and slower than when
optimization is enabled. You can use the
@option{-O} switch (the permitted forms are @option{-O0}, @option{-O1}
@option{-O2}, @option{-O3}, and @option{-Os})
to @command{gcc} to control the optimization level:

@table @option
@item -O0
No optimization (the default);
generates unoptimized code but has
the fastest compilation time.

Note that many other compilers do fairly extensive optimization
even if ``no optimization'' is specified. With gcc, it is
very unusual to use -O0 for production if
execution time is of any concern, since -O0
really does mean no optimization at all. This difference between
gcc and other compilers should be kept in mind when doing
performance comparisons.

@item -O1
Moderate optimization;
optimizes reasonably well but does not
degrade compilation time significantly.

@item -O2
Full optimization;
generates highly optimized code and has
the slowest compilation time.

@item -O3
Full optimization as in @option{-O2};
also uses more aggressive automatic inlining of subprograms within a unit
(@pxref{Inlining of Subprograms}) and attempts to vectorize loops.

@item -Os
Optimize space usage (code and data) of resulting program.
@end table

@noindent
Higher optimization levels perform more global transformations on the
program and apply more expensive analysis algorithms in order to generate
faster and more compact code. The price in compilation time, and the
resulting improvement in execution time,
both depend on the particular application and the hardware environment.
You should experiment to find the best level for your application.

Since the precise set of optimizations done at each level will vary from
release to release (and sometime from target to target), it is best to think
of the optimization settings in general terms.
@xref{Optimize Options,, Options That Control Optimization, gcc, Using
the GNU Compiler Collection (GCC)}, for details about
the @option{-O} settings and a number of @option{-f} options that
individually enable or disable specific optimizations.

Unlike some other compilation systems, @command{gcc} has
been tested extensively at all optimization levels. There are some bugs
which appear only with optimization turned on, but there have also been
bugs which show up only in @emph{unoptimized} code. Selecting a lower
level of optimization does not improve the reliability of the code
generator, which in practice is highly reliable at all optimization
levels.

Note regarding the use of @option{-O3}: The use of this optimization level
is generally discouraged with GNAT, since it often results in larger
executables which may run more slowly. See further discussion of this point
in @ref{Inlining of Subprograms}.

@node Debugging Optimized Code
@subsection Debugging Optimized Code
@cindex Debugging optimized code
@cindex Optimization and debugging

@noindent
Although it is possible to do a reasonable amount of debugging at
nonzero optimization levels,
the higher the level the more likely that
source-level constructs will have been eliminated by optimization.
For example, if a loop is strength-reduced, the loop
control variable may be completely eliminated and thus cannot be
displayed in the debugger.
This can only happen at @option{-O2} or @option{-O3}.
Explicit temporary variables that you code might be eliminated at
level @option{-O1} or higher.

The use of the @option{-g} switch,
@cindex @option{-g} (@command{gcc})
which is needed for source-level debugging,
affects the size of the program executable on disk,
and indeed the debugging information can be quite large.
However, it has no effect on the generated code (and thus does not
degrade performance)

Since the compiler generates debugging tables for a compilation unit before
it performs optimizations, the optimizing transformations may invalidate some
of the debugging data.  You therefore need to anticipate certain
anomalous situations that may arise while debugging optimized code.
These are the most common cases:

@enumerate
@item
@i{The ``hopping Program Counter'':}  Repeated @code{step} or @code{next}
commands show
the PC bouncing back and forth in the code.  This may result from any of
the following optimizations:

@itemize @bullet
@item
@i{Common subexpression elimination:} using a single instance of code for a
quantity that the source computes several times.  As a result you
may not be able to stop on what looks like a statement.

@item
@i{Invariant code motion:} moving an expression that does not change within a
loop, to the beginning of the loop.

@item
@i{Instruction scheduling:} moving instructions so as to
overlap loads and stores (typically) with other code, or in
general to move computations of values closer to their uses. Often
this causes you to pass an assignment statement without the assignment
happening and then later bounce back to the statement when the
value is actually needed.  Placing a breakpoint on a line of code
and then stepping over it may, therefore, not always cause all the
expected side-effects.
@end itemize

@item
@i{The ``big leap'':} More commonly known as @emph{cross-jumping}, in which
two identical pieces of code are merged and the program counter suddenly
jumps to a statement that is not supposed to be executed, simply because
it (and the code following) translates to the same thing as the code
that @emph{was} supposed to be executed.  This effect is typically seen in
sequences that end in a jump, such as a @code{goto}, a @code{return}, or
a @code{break} in a C @code{switch} statement.

@item
@i{The ``roving variable'':} The symptom is an unexpected value in a variable.
There are various reasons for this effect:

@itemize @bullet
@item
In a subprogram prologue, a parameter may not yet have been moved to its
``home''.

@item
A variable may be dead, and its register re-used.  This is
probably the most common cause.

@item
As mentioned above, the assignment of a value to a variable may
have been moved.

@item
A variable may be eliminated entirely by value propagation or
other means.  In this case, GCC may incorrectly generate debugging
information for the variable
@end itemize

@noindent
In general, when an unexpected value appears for a local variable or parameter
you should first ascertain if that value was actually computed by
your program, as opposed to being incorrectly reported by the debugger.
Record fields or
array elements in an object designated by an access value
are generally less of a problem, once you have ascertained that the access
value is sensible.
Typically, this means checking variables in the preceding code and in the
calling subprogram to verify that the value observed is explainable from other
values (one must apply the procedure recursively to those
other values); or re-running the code and stopping a little earlier
(perhaps before the call) and stepping to better see how the variable obtained
the value in question; or continuing to step @emph{from} the point of the
strange value to see if code motion had simply moved the variable's
assignments later.
@end enumerate

@noindent
In light of such anomalies, a recommended technique is to use @option{-O0}
early in the software development cycle, when extensive debugging capabilities
are most needed, and then move to @option{-O1} and later @option{-O2} as
the debugger becomes less critical.
Whether to use the @option{-g} switch in the release version is
a release management issue.
Note that if you use @option{-g} you can then use the @command{strip} program
on the resulting executable,
which removes both debugging information and global symbols.

@node Inlining of Subprograms
@subsection Inlining of Subprograms

@noindent
A call to a subprogram in the current unit is inlined if all the
following conditions are met:

@itemize @bullet
@item
The optimization level is at least @option{-O1}.

@item
The called subprogram is suitable for inlining: It must be small enough
and not contain something that @command{gcc} cannot support in inlined
subprograms.

@item
@cindex pragma Inline
@findex Inline
Any one of the following applies: @code{pragma Inline} is applied to the
subprogram and the @option{-gnatn} switch is specified; the
subprogram is local to the unit and called once from within it; the
subprogram is small and optimization level @option{-O2} is specified;
optimization level @option{-O3} is specified.
@end itemize

@noindent
Calls to subprograms in @code{with}'ed units are normally not inlined.
To achieve actual inlining (that is, replacement of the call by the code
in the body of the subprogram), the following conditions must all be true:

@itemize @bullet
@item
The optimization level is at least @option{-O1}.

@item
The called subprogram is suitable for inlining: It must be small enough
and not contain something that @command{gcc} cannot support in inlined
subprograms.

@item
The call appears in a body (not in a package spec).

@item
There is a @code{pragma Inline} for the subprogram.

@item
The @option{-gnatn} switch is used on the command line.
@end itemize

Even if all these conditions are met, it may not be possible for
the compiler to inline the call, due to the length of the body,
or features in the body that make it impossible for the compiler
to do the inlining.

Note that specifying the @option{-gnatn} switch causes additional
compilation dependencies. Consider the following:

@smallexample @c ada
@cartouche
@b{package} R @b{is}
   @b{procedure} Q;
   @b{pragma} Inline (Q);
@b{end} R;
@b{package} @b{body} R @b{is}
   @dots{}
@b{end} R;

@b{with} R;
@b{procedure} Main @b{is}
@b{begin}
   @dots{}
   R.Q;
@b{end} Main;
@end cartouche
@end smallexample

@noindent
With the default behavior (no @option{-gnatn} switch specified), the
compilation of the @code{Main} procedure depends only on its own source,
@file{main.adb}, and the spec of the package in file @file{r.ads}. This
means that editing the body of @code{R} does not require recompiling
@code{Main}.

On the other hand, the call @code{R.Q} is not inlined under these
circumstances. If the @option{-gnatn} switch is present when @code{Main}
is compiled, the call will be inlined if the body of @code{Q} is small
enough, but now @code{Main} depends on the body of @code{R} in
@file{r.adb} as well as on the spec. This means that if this body is edited,
the main program must be recompiled. Note that this extra dependency
occurs whether or not the call is in fact inlined by @command{gcc}.

The use of front end inlining with @option{-gnatN} generates similar
additional dependencies.

@cindex @option{-fno-inline} (@command{gcc})
Note: The @option{-fno-inline} switch
can be used to prevent
all inlining. This switch overrides all other conditions and ensures
that no inlining occurs. The extra dependences resulting from
@option{-gnatn} will still be active, even if
this switch is used to suppress the resulting inlining actions.

@cindex @option{-fno-inline-functions} (@command{gcc})
Note: The @option{-fno-inline-functions} switch can be used to prevent
automatic inlining of subprograms if @option{-O3} is used.

@cindex @option{-fno-inline-small-functions} (@command{gcc})
Note: The @option{-fno-inline-small-functions} switch can be used to prevent
automatic inlining of small subprograms if @option{-O2} is used.

@cindex @option{-fno-inline-functions-called-once} (@command{gcc})
Note: The @option{-fno-inline-functions-called-once} switch
can be used to prevent inlining of subprograms local to the unit
and called once from within it if @option{-O1} is used.

Note regarding the use of @option{-O3}: @option{-gnatn} is made up of two
sub-switches @option{-gnatn1} and @option{-gnatn2} that can be directly
specified in lieu of it, @option{-gnatn} being translated into one of them
based on the optimization level. With @option{-O2} or below, @option{-gnatn}
is equivalent to @option{-gnatn1} which activates pragma @code{Inline} with
moderate inlining across modules. With @option{-O3}, @option{-gnatn} is
equivalent to @option{-gnatn2} which activates pragma @code{Inline} with
full inlining across modules. If you have used pragma @code{Inline} in appropriate cases, then it is usually much better to use @option{-O2} and @option{-gnatn} and avoid the use of @option{-O3} which has the additional
effect of inlining subprograms you did not think should be inlined. We have
found that the use of @option{-O3} may slow down the compilation and increase
the code size by performing excessive inlining, leading to increased
instruction cache pressure from the increased code size and thus minor
performance improvements. So the bottom line here is that you should not
automatically assume that @option{-O3} is better than @option{-O2}, and
indeed you should use @option{-O3} only if tests show that it actually
improves performance for your program.

@node Vectorization of loops
@subsection Vectorization of loops
@cindex Optimization Switches

You can take advantage of the auto-vectorizer present in the @command{gcc}
back end to vectorize loops with GNAT.  The corresponding command line switch
is @option{-ftree-vectorize} but, as it is enabled by default at @option{-O3}
and other aggressive optimizations helpful for vectorization also are enabled
by default at this level, using @option{-O3} directly is recommended.

You also need to make sure that the target architecture features a supported
SIMD instruction set.  For example, for the x86 architecture, you should at
least specify @option{-msse2} to get significant vectorization (but you don't
need to specify it for x86-64 as it is part of the base 64-bit architecture).
Similarly, for the PowerPC architecture, you should specify @option{-maltivec}.

The preferred loop form for vectorization is the @code{for} iteration scheme.
Loops with a @code{while} iteration scheme can also be vectorized if they are
very simple, but the vectorizer will quickly give up otherwise.  With either
iteration scheme, the flow of control must be straight, in particular no
@code{exit} statement may appear in the loop body.  The loop may however
contain a single nested loop, if it can be vectorized when considered alone:

@smallexample @c ada
@cartouche
   A : @b{array} (1..4, 1..4) @b{of} Long_Float;
   S : @b{array} (1..4) @b{of} Long_Float;

   @b{procedure} Sum @b{is}
   @b{begin}
      @b{for} I @b{in} A'Range(1) @b{loop}
         @b{for} J @b{in} A'Range(2) @b{loop}
            S (I) := S (I) + A (I, J);
         @b{end} @b{loop};
      @b{end} @b{loop};
   @b{end} Sum;
@end cartouche
@end smallexample

The vectorizable operations depend on the targeted SIMD instruction set, but
the adding and some of the multiplying operators are generally supported, as
well as the logical operators for modular types.  Note that, in the former
case, enabling overflow checks, for example with @option{-gnato}, totally
disables vectorization.  The other checks are not supposed to have the same
definitive effect, although compiling with @option{-gnatp} might well reveal
cases where some checks do thwart vectorization.

Type conversions may also prevent vectorization if they involve semantics that
are not directly supported by the code generator or the SIMD instruction set.
A typical example is direct conversion from floating-point to integer types.
The solution in this case is to use the following idiom:

@smallexample @c ada
   Integer (S'Truncation (F))
@end smallexample

@noindent
if @code{S} is the subtype of floating-point object @code{F}.

In most cases, the vectorizable loops are loops that iterate over arrays.
All kinds of array types are supported, i.e. constrained array types with
static bounds:

@smallexample @c ada
   @b{type} Array_Type @b{is} @b{array} (1 .. 4) @b{of} Long_Float;
@end smallexample

@noindent
constrained array types with dynamic bounds:

@smallexample @c ada
   @b{type} Array_Type @b{is} @b{array} (1 .. Q.N) @b{of} Long_Float;

   @b{type} Array_Type @b{is} @b{array} (Q.K .. 4) @b{of} Long_Float;

   @b{type} Array_Type @b{is} @b{array} (Q.K .. Q.N) @b{of} Long_Float;
@end smallexample

@noindent
or unconstrained array types:

@smallexample @c ada
  @b{type} Array_Type @b{is} @b{array} (Positive @b{range} <>) @b{of} Long_Float;
@end smallexample

@noindent
The quality of the generated code decreases when the dynamic aspect of the
array type increases, the worst code being generated for unconstrained array
types.  This is so because, the less information the compiler has about the
bounds of the array, the more fallback code it needs to generate in order to
fix things up at run time.

It is possible to specify that a given loop should be subject to vectorization
preferably to other optimizations by means of pragma @code{Loop_Optimize}:

@smallexample @c ada
  @b{pragma} Loop_Optimize (Vector);
@end smallexample

@noindent
placed immediately within the loop will convey the appropriate hint to the
compiler for this loop.

It is also possible to help the compiler generate better vectorized code
for a given loop by asserting that there are no loop-carried dependencies
in the loop.  Consider for example the procedure:

@smallexample @c ada
  @b{type} Arr @b{is} @b{array} (1 .. 4) @b{of} Long_Float;

  @b{procedure} Add (X, Y : @b{not} @b{null} @b{access} Arr; R : @b{not} @b{null} @b{access} Arr) @b{is}
  @b{begin}
    @b{for} I @b{in} Arr'Range @b{loop}
      R(I) := X(I) + Y(I);
    @b{end} @b{loop};
  @b{end};
@end smallexample

@noindent
By default, the compiler cannot unconditionally vectorize the loop because
assigning to a component of the array designated by R in one iteration could
change the value read from the components of the array designated by X or Y
in a later iteration.  As a result, the compiler will generate two versions
of the loop in the object code, one vectorized and the other not vectorized,
as well as a test to select the appropriate version at run time.  This can
be overcome by another hint:

@smallexample @c ada
  @b{pragma} Loop_Optimize (Ivdep);
@end smallexample

@noindent
placed immediately within the loop will tell the compiler that it can safely
omit the non-vectorized version of the loop as well as the run-time test.

@node Other Optimization Switches
@subsection Other Optimization Switches
@cindex Optimization Switches

Since @code{GNAT} uses the @command{gcc} back end, all the specialized
@command{gcc} optimization switches are potentially usable. These switches
have not been extensively tested with GNAT but can generally be expected
to work. Examples of switches in this category are @option{-funroll-loops}
and the various target-specific @option{-m} options (in particular, it has
been observed that @option{-march=xxx} can significantly improve performance
on appropriate machines). For full details of these switches, see
@ref{Submodel Options,, Hardware Models and Configurations, gcc, Using
the GNU Compiler Collection (GCC)}.

@node Optimization and Strict Aliasing
@subsection Optimization and Strict Aliasing
@cindex Aliasing
@cindex Strict Aliasing
@cindex No_Strict_Aliasing

@noindent
The strong typing capabilities of Ada allow an optimizer to generate
efficient code in situations where other languages would be forced to
make worst case assumptions preventing such optimizations. Consider
the following example:

@smallexample @c ada
@cartouche
@b{procedure} R @b{is}
   @b{type} Int1 @b{is} @b{new} Integer;
   @b{type} Int2 @b{is} @b{new} Integer;
   @b{type} Int1A @b{is} @b{access} Int1;
   @b{type} Int2A @b{is} @b{access} Int2;
   Int1V : Int1A;
   Int2V : Int2A;
   @dots{}

@b{begin}
   @dots{}
   @b{for} J @b{in} Data'Range @b{loop}
      @b{if} Data (J) = Int1V.@b{all} @b{then}
         Int2V.@b{all} := Int2V.@b{all} + 1;
      @b{end} @b{if};
   @b{end} @b{loop};
   @dots{}
@b{end} R;
@end cartouche
@end smallexample

@noindent
In this example, since the variable @code{Int1V} can only access objects
of type @code{Int1}, and @code{Int2V} can only access objects of type
@code{Int2}, there is no possibility that the assignment to
@code{Int2V.all} affects the value of @code{Int1V.all}. This means that
the compiler optimizer can "know" that the value @code{Int1V.all} is constant
for all iterations of the loop and avoid the extra memory reference
required to dereference it each time through the loop.

This kind of optimization, called strict aliasing analysis, is
triggered by specifying an optimization level of @option{-O2} or
higher or @option{-Os} and allows @code{GNAT} to generate more efficient code
when access values are involved.

However, although this optimization is always correct in terms of
the formal semantics of the Ada Reference Manual, difficulties can
arise if features like @code{Unchecked_Conversion} are used to break
the typing system. Consider the following complete program example:

@smallexample @c ada
@cartouche
@b{package} p1 @b{is}
   @b{type} int1 @b{is} @b{new} integer;
   @b{type} int2 @b{is} @b{new} integer;
   @b{type} a1 @b{is} @b{access} int1;
   @b{type} a2 @b{is} @b{access} int2;
@b{end} p1;

@b{with} p1; @b{use} p1;
@b{package} p2 @b{is}
   @b{function} to_a2 (Input : a1) @b{return} a2;
@b{end} p2;

@b{with} Unchecked_Conversion;
@b{package} @b{body} p2 @b{is}
   @b{function} to_a2 (Input : a1) @b{return} a2 @b{is}
      @b{function} to_a2u @b{is}
        @b{new} Unchecked_Conversion (a1, a2);
   @b{begin}
      @b{return} to_a2u (Input);
   @b{end} to_a2;
@b{end} p2;

@b{with} p2; @b{use} p2;
@b{with} p1; @b{use} p1;
@b{with} Text_IO; @b{use} Text_IO;
@b{procedure} m @b{is}
   v1 : a1 := @b{new} int1;
   v2 : a2 := to_a2 (v1);
@b{begin}
   v1.@b{all} := 1;
   v2.@b{all} := 0;
   put_line (int1'image (v1.@b{all}));
@b{end};
@end cartouche
@end smallexample

@noindent
This program prints out 0 in @option{-O0} or @option{-O1}
mode, but it prints out 1 in @option{-O2} mode. That's
because in strict aliasing mode, the compiler can and
does assume that the assignment to @code{v2.all} could not
affect the value of @code{v1.all}, since different types
are involved.

This behavior is not a case of non-conformance with the standard, since
the Ada RM specifies that an unchecked conversion where the resulting
bit pattern is not a correct value of the target type can result in an
abnormal value and attempting to reference an abnormal value makes the
execution of a program erroneous.  That's the case here since the result
does not point to an object of type @code{int2}.  This means that the
effect is entirely unpredictable.

However, although that explanation may satisfy a language
lawyer, in practice an applications programmer expects an
unchecked conversion involving pointers to create true
aliases and the behavior of printing 1 seems plain wrong.
In this case, the strict aliasing optimization is unwelcome.

Indeed the compiler recognizes this possibility, and the
unchecked conversion generates a warning:

@smallexample
p2.adb:5:07: warning: possible aliasing problem with type "a2"
p2.adb:5:07: warning: use -fno-strict-aliasing switch for references
p2.adb:5:07: warning:  or use "pragma No_Strict_Aliasing (a2);"
@end smallexample

@noindent
Unfortunately the problem is recognized when compiling the body of
package @code{p2}, but the actual "bad" code is generated while
compiling the body of @code{m} and this latter compilation does not see
the suspicious @code{Unchecked_Conversion}.

As implied by the warning message, there are approaches you can use to
avoid the unwanted strict aliasing optimization in a case like this.

One possibility is to simply avoid the use of @option{-O2}, but
that is a bit drastic, since it throws away a number of useful
optimizations that do not involve strict aliasing assumptions.

A less drastic approach is to compile the program using the
option @option{-fno-strict-aliasing}. Actually it is only the
unit containing the dereferencing of the suspicious pointer
that needs to be compiled. So in this case, if we compile
unit @code{m} with this switch, then we get the expected
value of zero printed. Analyzing which units might need
the switch can be painful, so a more reasonable approach
is to compile the entire program with options @option{-O2}
and @option{-fno-strict-aliasing}. If the performance is
satisfactory with this combination of options, then the
advantage is that the entire issue of possible "wrong"
optimization due to strict aliasing is avoided.

To avoid the use of compiler switches, the configuration
pragma @code{No_Strict_Aliasing} with no parameters may be
used to specify that for all access types, the strict
aliasing optimization should be suppressed.

However, these approaches are still overkill, in that they causes
all manipulations of all access values to be deoptimized. A more
refined approach is to concentrate attention on the specific
access type identified as problematic.

First, if a careful analysis of uses of the pointer shows
that there are no possible problematic references, then
the warning can be suppressed by bracketing the
instantiation of @code{Unchecked_Conversion} to turn
the warning off:

@smallexample @c ada
   @b{pragma} Warnings (Off);
   @b{function} to_a2u @b{is}
     @b{new} Unchecked_Conversion (a1, a2);
   @b{pragma} Warnings (On);
@end smallexample

@noindent
Of course that approach is not appropriate for this particular
example, since indeed there is a problematic reference. In this
case we can take one of two other approaches.

The first possibility is to move the instantiation of unchecked
conversion to the unit in which the type is declared. In
this example, we would move the instantiation of
@code{Unchecked_Conversion} from the body of package
@code{p2} to the spec of package @code{p1}. Now the
warning disappears. That's because any use of the
access type knows there is a suspicious unchecked
conversion, and the strict aliasing optimization
is automatically suppressed for the type.

If it is not practical to move the unchecked conversion to the same unit
in which the destination access type is declared (perhaps because the
source type is not visible in that unit), you may use pragma
@code{No_Strict_Aliasing} for the type. This pragma must occur in the
same declarative sequence as the declaration of the access type:

@smallexample @c ada
   @b{type} a2 @b{is} @b{access} int2;
   @b{pragma} No_Strict_Aliasing (a2);
@end smallexample

@noindent
Here again, the compiler now knows that the strict aliasing optimization
should be suppressed for any reference to type @code{a2} and the
expected behavior is obtained.

Finally, note that although the compiler can generate warnings for
simple cases of unchecked conversions, there are tricker and more
indirect ways of creating type incorrect aliases which the compiler
cannot detect. Examples are the use of address overlays and unchecked
conversions involving composite types containing access types as
components. In such cases, no warnings are generated, but there can
still be aliasing problems. One safe coding practice is to forbid the
use of address clauses for type overlaying, and to allow unchecked
conversion only for primitive types. This is not really a significant
restriction since any possible desired effect can be achieved by
unchecked conversion of access values.

The aliasing analysis done in strict aliasing mode can certainly
have significant benefits. We have seen cases of large scale
application code where the time is increased by up to 5% by turning
this optimization off. If you have code that includes significant
usage of unchecked conversion, you might want to just stick with
@option{-O1} and avoid the entire issue. If you get adequate
performance at this level of optimization level, that's probably
the safest approach. If tests show that you really need higher
levels of optimization, then you can experiment with @option{-O2}
and @option{-O2 -fno-strict-aliasing} to see how much effect this
has on size and speed of the code. If you really need to use
@option{-O2} with strict aliasing in effect, then you should
review any uses of unchecked conversion of access types,
particularly if you are getting the warnings described above.

@node Aliased Variables and Optimization
@subsection Aliased Variables and Optimization
@cindex Aliasing
There are scenarios in which programs may
use low level techniques to modify variables
that otherwise might be considered to be unassigned. For example,
a variable can be passed to a procedure by reference, which takes
the address of the parameter and uses the address to modify the
variable's value, even though it is passed as an IN parameter.
Consider the following example:

@smallexample @c ada
@b{procedure} P @b{is}
   Max_Length : @b{constant} Natural := 16;
   @b{type} Char_Ptr @b{is} @b{access} @b{all} Character;

   @b{procedure} Get_String(Buffer: Char_Ptr; Size : Integer);
   @b{pragma} Import (C, Get_String, "get_string");

   Name : @b{aliased} String (1 .. Max_Length) := (@b{others} => ' ');
   Temp : Char_Ptr;

   @b{function} Addr (S : String) @b{return} Char_Ptr @b{is}
      @b{function} To_Char_Ptr @b{is}
        @b{new} Ada.Unchecked_Conversion (System.Address, Char_Ptr);
   @b{begin}
      @b{return} To_Char_Ptr (S (S'First)'Address);
   @b{end};

@b{begin}
   Temp := Addr (Name);
   Get_String (Temp, Max_Length);
@b{end};
@end smallexample

@noindent
where Get_String is a C function that uses the address in Temp to
modify the variable @code{Name}. This code is dubious, and arguably
erroneous, and the compiler would be entitled to assume that
@code{Name} is never modified, and generate code accordingly.

However, in practice, this would cause some existing code that
seems to work with no optimization to start failing at high
levels of optimzization.

What the compiler does for such cases is to assume that marking
a variable as aliased indicates that some "funny business" may
be going on. The optimizer recognizes the aliased keyword and
inhibits optimizations that assume the value cannot be assigned.
This means that the above example will in fact "work" reliably,
that is, it will produce the expected results.

@node Atomic Variables and Optimization
@subsection Atomic Variables and Optimization
@cindex Atomic
There are two considerations with regard to performance when
atomic variables are used.

First, the RM only guarantees that access to atomic variables
be atomic, it has nothing to say about how this is achieved,
though there is a strong implication that this should not be
achieved by explicit locking code. Indeed GNAT will never
generate any locking code for atomic variable access (it will
simply reject any attempt to make a variable or type atomic
if the atomic access cannot be achieved without such locking code).

That being said, it is important to understand that you cannot
assume that the entire variable will always be accessed. Consider
this example:

@smallexample @c ada
@b{type} R @b{is} @b{record}
   A,B,C,D : Character;
@b{end} @b{record};
@b{for} R'Size @b{use} 32;
@b{for} R'Alignment @b{use} 4;

RV : R;
@b{pragma} Atomic (RV);
X : Character;
...
X := RV.B;
@end smallexample

@noindent
You cannot assume that the reference to @code{RV.B}
will read the entire 32-bit
variable with a single load instruction. It is perfectly legitimate if
the hardware allows it to do a byte read of just the B field. This read
is still atomic, which is all the RM requires. GNAT can and does take
advantage of this, depending on the architecture and optimization level.
Any assumption to the contrary is non-portable and risky. Even if you
examine the assembly language and see a full 32-bit load, this might
change in a future version of the compiler.

If your application requires that all accesses to @code{RV} in this
example be full 32-bit loads, you need to make a copy for the access
as in:

@smallexample @c ada
@b{declare}
   RV_Copy : @b{constant} R := RV;
@b{begin}
   X := RV_Copy.B;
@b{end};
@end smallexample


@noindent
Now the reference to RV must read the whole variable.
Actually one can imagine some compiler which figures
out that the whole copy is not required (because only
the B field is actually accessed), but GNAT
certainly won't do that, and we don't know of any
compiler that would not handle this right, and the
above code will in practice work portably across
all architectures (that permit the Atomic declaration).

The second issue with atomic variables has to do with
the possible requirement of generating synchronization
code. For more details on this, consult the sections on
the pragmas Enable/Disable_Atomic_Synchronization in the
GNAT Reference Manual. If performance is critical, and
such synchronization code is not required, it may be
useful to disable it.

@node Passive Task Optimization
@subsection Passive Task Optimization
@cindex Passive Task

A passive task is one which is sufficiently simple that
in theory a compiler could recognize it an implement it
efficiently without creating a new thread. The original design
of Ada 83 had in mind this kind of passive task optimization, but
only a few Ada 83 compilers attempted it. The problem was that
it was difficult to determine the exact conditions under which
the optimization was possible. The result is a very fragile
optimization where a very minor change in the program can
suddenly silently make a task non-optimizable.

With the revisiting of this issue in Ada 95, there was general
agreement that this approach was fundamentally flawed, and the
notion of protected types was introduced. When using protected
types, the restrictions are well defined, and you KNOW that the
operations will be optimized, and furthermore this optimized
performance is fully portable.

Although it would theoretically be possible for GNAT to attempt to
do this optimization, but it really doesn't make sense in the
context of Ada 95, and none of the Ada 95 compilers implement
this optimization as far as we know. In particular GNAT never
attempts to perform this optimization.

In any new Ada 95 code that is written, you should always
use protected types in place of tasks that might be able to
be optimized in this manner.
Of course this does not help if you have legacy Ada 83 code
that depends on this optimization, but it is unusual to encounter
a case where the performance gains from this optimization
are significant.

Your program should work correctly without this optimization. If
you have performance problems, then the most practical
approach is to figure out exactly where these performance problems
arise, and update those particular tasks to be protected types. Note
that typically clients of the tasks who call entries, will not have
to be modified, only the task definition itself.



@node Text_IO Suggestions
@section @code{Text_IO} Suggestions
@cindex @code{Text_IO} and performance

@noindent
The @code{Ada.Text_IO} package has fairly high overheads due in part to
the requirement of maintaining page and line counts. If performance
is critical, a recommendation is to use @code{Stream_IO} instead of
@code{Text_IO} for volume output, since this package has less overhead.

If @code{Text_IO} must be used, note that by default output to the standard
output and standard error files is unbuffered (this provides better
behavior when output statements are used for debugging, or if the
progress of a program is observed by tracking the output, e.g. by
using the Unix @command{tail -f} command to watch redirected output.

If you are generating large volumes of output with @code{Text_IO} and
performance is an important factor, use a designated file instead
of the standard output file, or change the standard output file to
be buffered using @code{Interfaces.C_Streams.setvbuf}.


@ifclear FSFEDITION
@node Reducing Size of Ada Executables with gnatelim
@section Reducing Size of Ada Executables with @code{gnatelim}
@findex gnatelim

@noindent
This section describes @command{gnatelim}, a tool which detects unused
subprograms and helps the compiler to create a smaller executable for your
program.

@menu
* About gnatelim::
* Running gnatelim::
* Processing Precompiled Libraries::
* Correcting the List of Eliminate Pragmas::
* Making Your Executables Smaller::
* Summary of the gnatelim Usage Cycle::
@end menu

@node About gnatelim
@subsection About @code{gnatelim}

@noindent
When a program shares a set of Ada
packages with other programs, it may happen that this program uses
only a fraction of the subprograms defined in these packages. The code
created for these unused subprograms increases the size of the executable.

@code{gnatelim} tracks unused subprograms in an Ada program and
outputs a list of GNAT-specific pragmas @code{Eliminate} marking all the
subprograms that are declared but never called. By placing the list of
@code{Eliminate} pragmas in the GNAT configuration file @file{gnat.adc} and
recompiling your program, you may decrease the size of its executable,
because the compiler will not generate the code for 'eliminated' subprograms.
@xref{Pragma Eliminate,,, gnat_rm, GNAT Reference Manual}, for more
information about this pragma.

@code{gnatelim} needs as its input data the name of the main subprogram.

If a set of source files is specified as @code{gnatelim} arguments, it
treats these files as a complete set of sources making up a program to
analyse, and analyses only these sources.

After a full successful build of the main subprogram @code{gnatelim} can be
called without  specifying sources to analyse, in this case it computes
the source closure of the main unit from the @file{ALI} files.

If the set of sources to be processed by @code{gnatelim} contains sources with
preprocessing directives
then the needed options should be provided to run preprocessor as a part of
the @command{gnatelim} call, and the generated set of pragmas @code{Eliminate}
will correspond to preprocessed sources.

The following command will create the set of @file{ALI} files needed for
@code{gnatelim}:

@smallexample
$ gnatmake -c Main_Prog
@end smallexample

Note that @code{gnatelim} does not need object files.

@node Running gnatelim
@subsection Running @code{gnatelim}

@noindent
@code{gnatelim} has the following command-line interface:

@smallexample
$ gnatelim [@var{switches}] -main=@var{main_unit_name} @{@var{filename}@} @r{[}-cargs @var{gcc_switches}@r{]}
@end smallexample

@noindent
@var{main_unit_name} should be a name of a source file that contains the main
subprogram of a program (partition).

Each @var{filename} is the name (including the extension) of a source
file to process. ``Wildcards'' are allowed, and
the file name may contain path information.

@samp{@var{gcc_switches}} is a list of switches for
@command{gcc}. They will be passed on to all compiler invocations made by
@command{gnatelim} to generate the ASIS trees. Here you can provide
@option{-I} switches to form the source search path,
use the @option{-gnatec} switch to set the configuration file,
use the @option{-gnat05} switch if sources should be compiled in
Ada 2005 mode etc.

@code{gnatelim} has the following switches:

@table @option
@c !sort!
@item --version
@cindex @option{--version} @command{gnatelim}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatelim}
Display usage, then exit disregarding all other options.

@item -P @var{file}
@cindex @option{-P} @command{gnatelim}
Indicates the name of the project file that describes the set of sources
to be processed.

@item -X@var{name}=@var{value}
@cindex @option{-X} @command{gnatelim}
Indicates that external variable @var{name} in the argument project
has the value @var{value}. Has no effect if no project is specified as
tool argument.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gnatelim})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item -files=@var{filename}
@cindex @option{-files} (@code{gnatelim})
Take the argument source files from the specified file. This file should be an
ordinary text file containing file names separated by spaces or
line breaks. You can use this switch more than once in the same call to
@command{gnatelim}. You also can combine this switch with
an explicit list of files.

@item -log
@cindex @option{-log} (@command{gnatelim})
Duplicate all the output sent to @file{stderr} into a log file. The log file
is named @file{gnatelim.log} and is located in the current directory.

@ignore
@item -log=@var{filename}
@cindex @option{-log} (@command{gnatelim})
Duplicate all the output sent to @file{stderr} into a specified log file.
@end ignore

@cindex @option{--no-elim-dispatch} (@command{gnatelim})
@item --no-elim-dispatch
Do not generate pragmas for dispatching operations.

@item --ignore=@var{filename}
@cindex @option{--ignore} (@command{gnatelim})
Do not generate pragmas for subprograms declared in the sources
listed in a specified file

@cindex @option{-o} (@command{gnatelim})
@item -o=@var{report_file}
Put @command{gnatelim} output into a specified file. If this file already exists,
it is overridden. If this switch is not used, @command{gnatelim} outputs its results
into @file{stderr}

@item -j@var{n}
@cindex @option{-j} (@command{gnatelim})
Use @var{n} processes to carry out the tree creations (internal representations
of the argument sources). On a multiprocessor machine this speeds up processing
of big sets of argument sources. If @var{n} is 0, then the maximum number of
parallel tree creations is the number of core processors on the platform.

@item -q
@cindex @option{-q} (@command{gnatelim})
Quiet mode: by default @code{gnatelim} outputs to the standard error
stream the number of program units left to be processed. This option turns
this trace off.

@cindex @option{-t} (@command{gnatelim})
@item -t
Print out execution time.

@item -v
@cindex @option{-v} (@command{gnatelim})
Verbose mode: @code{gnatelim} version information is printed as Ada
comments to the standard output stream. Also, in addition to the number of
program units left @code{gnatelim} will output the name of the current unit
being processed.

@item -wq
@cindex @option{-wq} (@command{gnatelim})
Quiet warning mode - some warnings are suppressed. In particular warnings that
indicate that the analysed set of sources is incomplete to make up a
partition and that some subprogram bodies are missing are not generated.
@end table

@noindent
Note: to invoke @command{gnatelim} with a project file, use the @code{gnat}
driver (see @ref{The GNAT Driver and Project Files}).

@node Processing Precompiled Libraries
@subsection Processing Precompiled Libraries

@noindent
If some program uses a precompiled Ada library, it can be processed by
@code{gnatelim} in a usual way. @code{gnatelim} will newer generate an
Eliminate pragma for a subprogram if the body of this subprogram has not
been analysed, this is a typical case for subprograms from precompiled
libraries. Switch @option{-wq} may be used to suppress
warnings about missing source files and non-analyzed subprogram bodies
that can be generated when processing precompiled Ada libraries.

@node Correcting the List of Eliminate Pragmas
@subsection Correcting the List of Eliminate Pragmas

@noindent
In some rare cases @code{gnatelim} may try to eliminate
subprograms that are actually called in the program. In this case, the
compiler will generate an error message of the form:

@smallexample
main.adb:4:08: cannot reference subprogram "P" eliminated at elim.out:5
@end smallexample

@noindent
You will need to manually remove the wrong @code{Eliminate} pragmas from
the configuration file indicated in the error message. You should recompile
your program from scratch after that, because you need a consistent
configuration file(s) during the entire compilation.

@node Making Your Executables Smaller
@subsection Making Your Executables Smaller

@noindent
In order to get a smaller executable for your program you now have to
recompile the program completely with the configuration file containing
pragmas Eliminate generated by gnatelim. If these pragmas are placed in
@file{gnat.adc} file located in your current directory, just do:

@smallexample
$ gnatmake -f main_prog
@end smallexample

@noindent
(Use the @option{-f} option for @command{gnatmake} to
recompile everything
with the set of pragmas @code{Eliminate} that you have obtained with
@command{gnatelim}).

Be aware that the set of @code{Eliminate} pragmas is specific to each
program. It is not recommended to merge sets of @code{Eliminate}
pragmas created for different programs in one configuration file.

@node Summary of the gnatelim Usage Cycle
@subsection Summary of the @code{gnatelim} Usage Cycle

@noindent
Here is a quick summary of the steps to be taken in order to reduce
the size of your executables with @code{gnatelim}. You may use
other GNAT options to control the optimization level,
to produce the debugging information, to set search path, etc.

@enumerate
@item
Create a complete set of @file{ALI} files (if the program has not been
built already)

@smallexample
$ gnatmake -c main_prog
@end smallexample

@item
Generate a list of @code{Eliminate} pragmas in default configuration file
@file{gnat.adc} in the current directory
@smallexample
$ gnatelim main_prog >@r{[}>@r{]} gnat.adc
@end smallexample

@item
Recompile the application

@smallexample
$ gnatmake -f main_prog
@end smallexample

@end enumerate
@end ifclear

@node Reducing Size of Executables with unused subprogram/data elimination
@section Reducing Size of Executables with Unused Subprogram/Data Elimination
@findex unused subprogram/data elimination

@noindent
This section describes how you can eliminate unused subprograms and data from
your executable just by setting options at compilation time.

@menu
* About unused subprogram/data elimination::
* Compilation options::
* Example of unused subprogram/data elimination::
@end menu

@node About unused subprogram/data elimination
@subsection About unused subprogram/data elimination

@noindent
By default, an executable contains all code and data of its composing objects
(directly linked or coming from statically linked libraries), even data or code
never used by this executable.

This feature will allow you to eliminate such unused code from your
executable, making it smaller (in disk and in memory).

This functionality is available on all Linux platforms except for the IA-64
architecture and on all cross platforms using the ELF binary file format.
In both cases GNU binutils version 2.16 or later are required to enable it.

@node Compilation options
@subsection Compilation options

@noindent
The operation of eliminating the unused code and data from the final executable
is directly performed by the linker.

In order to do this, it has to work with objects compiled with the
following options:
@option{-ffunction-sections} @option{-fdata-sections}.
@cindex @option{-ffunction-sections} (@command{gcc})
@cindex @option{-fdata-sections} (@command{gcc})
These options are usable with C and Ada files.
They will place respectively each
function or data in a separate section in the resulting object file.

Once the objects and static libraries are created with these options, the
linker can perform the dead code elimination. You can do this by setting
the @option{-Wl,--gc-sections} option to gcc command or in the
@option{-largs} section of @command{gnatmake}. This will perform a
garbage collection of code and data never referenced.

If the linker performs a partial link (@option{-r} linker option), then you
will need to provide the entry point using the @option{-e} / @option{--entry}
linker option.

Note that objects compiled without the @option{-ffunction-sections} and
@option{-fdata-sections} options can still be linked with the executable.
However, no dead code elimination will be performed on those objects (they will
be linked as is).

The GNAT static library is now compiled with -ffunction-sections and
-fdata-sections on some platforms. This allows you to eliminate the unused code
and data of the GNAT library from your executable.

@node Example of unused subprogram/data elimination
@subsection Example of unused subprogram/data elimination

@noindent
Here is a simple example:

@smallexample @c ada
@b{with} Aux;

@b{procedure} Test @b{is}
@b{begin}
   Aux.Used (10);
@b{end} Test;

@b{package} Aux @b{is}
   Used_Data   : Integer;
   Unused_Data : Integer;

   @b{procedure} Used   (Data : Integer);
   @b{procedure} Unused (Data : Integer);
@b{end} Aux;

@b{package} @b{body} Aux @b{is}
   @b{procedure} Used (Data : Integer) @b{is}
   @b{begin}
      Used_Data := Data;
   @b{end} Used;

   @b{procedure} Unused (Data : Integer) @b{is}
   @b{begin}
      Unused_Data := Data;
   @b{end} Unused;
@b{end} Aux;
@end smallexample

@noindent
@code{Unused} and @code{Unused_Data} are never referenced in this code
excerpt, and hence they may be safely removed from the final executable.

@smallexample
$ gnatmake test

$ nm test | grep used
020015f0 T aux__unused
02005d88 B aux__unused_data
020015cc T aux__used
02005d84 B aux__used_data

$ gnatmake test -cargs -fdata-sections -ffunction-sections \
     -largs -Wl,--gc-sections

$ nm test | grep used
02005350 T aux__used
0201ffe0 B aux__used_data
@end smallexample

@noindent
It can be observed that the procedure @code{Unused} and the object
@code{Unused_Data} are removed by the linker when using the
appropriate options.

@c ********************************
@node Renaming Files with gnatchop
@chapter Renaming Files with @code{gnatchop}
@findex gnatchop

@noindent
This chapter discusses how to handle files with multiple units by using
the @code{gnatchop} utility. This utility is also useful in renaming
files to meet the standard GNAT default file naming conventions.

@menu
* Handling Files with Multiple Units::
* Operating gnatchop in Compilation Mode::
* Command Line for gnatchop::
* Switches for gnatchop::
* Examples of gnatchop Usage::
@end menu

@node Handling Files with Multiple Units
@section Handling Files with Multiple Units

@noindent
The basic compilation model of GNAT requires that a file submitted to the
compiler have only one unit and there be a strict correspondence
between the file name and the unit name.

The @code{gnatchop} utility allows both of these rules to be relaxed,
allowing GNAT to process files which contain multiple compilation units
and files with arbitrary file names. @code{gnatchop}
reads the specified file and generates one or more output files,
containing one unit per file. The unit and the file name correspond,
as required by GNAT.

If you want to permanently restructure a set of ``foreign'' files so that
they match the GNAT rules, and do the remaining development using the
GNAT structure, you can simply use @command{gnatchop} once, generate the
new set of files and work with them from that point on.

Alternatively, if you want to keep your files in the ``foreign'' format,
perhaps to maintain compatibility with some other Ada compilation
system, you can set up a procedure where you use @command{gnatchop} each
time you compile, regarding the source files that it writes as temporary
files that you throw away.

Note that if your file containing multiple units starts with a byte order
mark (BOM) specifying UTF-8 encoding, then the files generated by gnatchop
will each start with a copy of this BOM, meaning that they can be compiled
automatically in UTF-8 mode without needing to specify an explicit encoding.

@node Operating gnatchop in Compilation Mode
@section Operating gnatchop in Compilation Mode

@noindent
The basic function of @code{gnatchop} is to take a file with multiple units
and split it into separate files. The boundary between files is reasonably
clear, except for the issue of comments and pragmas. In default mode, the
rule is that any pragmas between units belong to the previous unit, except
that configuration pragmas always belong to the following unit. Any comments
belong to the following unit. These rules
almost always result in the right choice of
the split point without needing to mark it explicitly and most users will
find this default to be what they want. In this default mode it is incorrect to
submit a file containing only configuration pragmas, or one that ends in
configuration pragmas, to @code{gnatchop}.

However, using a special option to activate ``compilation mode'',
@code{gnatchop}
can perform another function, which is to provide exactly the semantics
required by the RM for handling of configuration pragmas in a compilation.
In the absence of configuration pragmas (at the main file level), this
option has no effect, but it causes such configuration pragmas to be handled
in a quite different manner.

First, in compilation mode, if @code{gnatchop} is given a file that consists of
only configuration pragmas, then this file is appended to the
@file{gnat.adc} file in the current directory. This behavior provides
the required behavior described in the RM for the actions to be taken
on submitting such a file to the compiler, namely that these pragmas
should apply to all subsequent compilations in the same compilation
environment. Using GNAT, the current directory, possibly containing a
@file{gnat.adc} file is the representation
of a compilation environment. For more information on the
@file{gnat.adc} file, see @ref{Handling of Configuration Pragmas}.

Second, in compilation mode, if @code{gnatchop}
is given a file that starts with
configuration pragmas, and contains one or more units, then these
configuration pragmas are prepended to each of the chopped files. This
behavior provides the required behavior described in the RM for the
actions to be taken on compiling such a file, namely that the pragmas
apply to all units in the compilation, but not to subsequently compiled
units.

Finally, if configuration pragmas appear between units, they are appended
to the previous unit. This results in the previous unit being illegal,
since the compiler does not accept configuration pragmas that follow
a unit. This provides the required RM behavior that forbids configuration
pragmas other than those preceding the first compilation unit of a
compilation.

For most purposes, @code{gnatchop} will be used in default mode. The
compilation mode described above is used only if you need exactly
accurate behavior with respect to compilations, and you have files
that contain multiple units and configuration pragmas. In this
circumstance the use of @code{gnatchop} with the compilation mode
switch provides the required behavior, and is for example the mode
in which GNAT processes the ACVC tests.

@node Command Line for gnatchop
@section Command Line for @code{gnatchop}

@noindent
The @code{gnatchop} command has the form:

@smallexample
@c $ gnatchop switches @var{file name} @r{[}@var{file name} @dots{}@r{]}
@c      @ovar{directory}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatchop switches @var{file name} @r{[}@var{file name} @dots{}@r{]}
      @r{[}@var{directory}@r{]}
@end smallexample

@noindent
The only required argument is the file name of the file to be chopped.
There are no restrictions on the form of this file name. The file itself
contains one or more Ada units, in normal GNAT format, concatenated
together. As shown, more than one file may be presented to be chopped.

When run in default mode, @code{gnatchop} generates one output file in
the current directory for each unit in each of the files.

@var{directory}, if specified, gives the name of the directory to which
the output files will be written. If it is not specified, all files are
written to the current directory.

For example, given a
file called @file{hellofiles} containing

@smallexample @c ada
@group
@cartouche
@b{procedure} hello;

@b{with} Text_IO; @b{use} Text_IO;
@b{procedure} hello @b{is}
@b{begin}
   Put_Line ("Hello");
@b{end} hello;
@end cartouche
@end group
@end smallexample

@noindent
the command

@smallexample
$ gnatchop hellofiles
@end smallexample

@noindent
generates two files in the current directory, one called
@file{hello.ads} containing the single line that is the procedure spec,
and the other called @file{hello.adb} containing the remaining text. The
original file is not affected. The generated files can be compiled in
the normal manner.

@noindent
When gnatchop is invoked on a file that is empty or that contains only empty
lines and/or comments, gnatchop will not fail, but will not produce any
new sources.

For example, given a
file called @file{toto.txt} containing

@smallexample @c ada
@group
@cartouche
--@i{  Just a comment}
@end cartouche
@end group
@end smallexample

@noindent
the command

@smallexample
$ gnatchop toto.txt
@end smallexample

@noindent
will not produce any new file and will result in the following warnings:

@smallexample
toto.txt:1:01: warning: empty file, contains no compilation units
no compilation units found
no source files written
@end smallexample

@node Switches for gnatchop
@section Switches for @code{gnatchop}

@noindent
@command{gnatchop} recognizes the following switches:

@table @option
@c !sort!

@item --version
@cindex @option{--version} @command{gnatchop}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatchop}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item -c
@cindex @option{-c} (@code{gnatchop})
Causes @code{gnatchop} to operate in compilation mode, in which
configuration pragmas are handled according to strict RM rules. See
previous section for a full description of this mode.

@item -gnat@var{xxx}
This passes the given @option{-gnat@var{xxx}} switch to @code{gnat} which is
used to parse the given file. Not all @var{xxx} options make sense,
but for example, the use of @option{-gnati2} allows @code{gnatchop} to
process a source file that uses Latin-2 coding for identifiers.

@item -h
Causes @code{gnatchop} to generate a brief help summary to the standard
output file showing usage information.

@item -k@var{mm}
@cindex @option{-k} (@code{gnatchop})
Limit generated file names to the specified number @code{mm}
of characters.
This is useful if the
resulting set of files is required to be interoperable with systems
which limit the length of file names.
No space is allowed between the @option{-k} and the numeric value. The numeric
value may be omitted in which case a default of @option{-k8},
suitable for use
with DOS-like file systems, is used. If no @option{-k} switch
is present then
there is no limit on the length of file names.

@item -p
@cindex @option{-p} (@code{gnatchop})
Causes the file modification time stamp of the input file to be
preserved and used for the time stamp of the output file(s). This may be
useful for preserving coherency of time stamps in an environment where
@code{gnatchop} is used as part of a standard build process.

@item -q
@cindex @option{-q} (@code{gnatchop})
Causes output of informational messages indicating the set of generated
files to be suppressed. Warnings and error messages are unaffected.

@item -r
@cindex @option{-r} (@code{gnatchop})
@findex Source_Reference
Generate @code{Source_Reference} pragmas. Use this switch if the output
files are regarded as temporary and development is to be done in terms
of the original unchopped file. This switch causes
@code{Source_Reference} pragmas to be inserted into each of the
generated files to refers back to the original file name and line number.
The result is that all error messages refer back to the original
unchopped file.
In addition, the debugging information placed into the object file (when
the @option{-g} switch of @command{gcc} or @command{gnatmake} is
specified)
also refers back to this original file so that tools like profilers and
debuggers will give information in terms of the original unchopped file.

If the original file to be chopped itself contains
a @code{Source_Reference}
pragma referencing a third file, then gnatchop respects
this pragma, and the generated @code{Source_Reference} pragmas
in the chopped file refer to the original file, with appropriate
line numbers. This is particularly useful when @code{gnatchop}
is used in conjunction with @code{gnatprep} to compile files that
contain preprocessing statements and multiple units.

@item -v
@cindex @option{-v} (@code{gnatchop})
Causes @code{gnatchop} to operate in verbose mode. The version
number and copyright notice are output, as well as exact copies of
the gnat1 commands spawned to obtain the chop control information.

@item -w
@cindex @option{-w} (@code{gnatchop})
Overwrite existing file names. Normally @code{gnatchop} regards it as a
fatal error if there is already a file with the same name as a
file it would otherwise output, in other words if the files to be
chopped contain duplicated units. This switch bypasses this
check, and causes all but the last instance of such duplicated
units to be skipped.

@item --GCC=@var{xxxx}
@cindex @option{--GCC=} (@code{gnatchop})
Specify the path of the GNAT parser to be used. When this switch is used,
no attempt is made to add the prefix to the GNAT parser executable.
@end table

@node Examples of gnatchop Usage
@section Examples of @code{gnatchop} Usage

@table @code
@item gnatchop -w hello_s.ada prerelease/files

Chops the source file @file{hello_s.ada}. The output files will be
placed in the directory @file{prerelease/files},
overwriting any
files with matching names in that directory (no files in the current
directory are modified).

@item gnatchop archive
Chops the source file @file{archive}
into the current directory. One
useful application of @code{gnatchop} is in sending sets of sources
around, for example in email messages. The required sources are simply
concatenated (for example, using a Unix @code{cat}
command), and then
@command{gnatchop} is used at the other end to reconstitute the original
file names.

@item gnatchop file1 file2 file3 direc
Chops all units in files @file{file1}, @file{file2}, @file{file3}, placing
the resulting files in the directory @file{direc}. Note that if any units
occur more than once anywhere within this set of files, an error message
is generated, and no files are written. To override this check, use the
@option{-w} switch,
in which case the last occurrence in the last file will
be the one that is output, and earlier duplicate occurrences for a given
unit will be skipped.
@end table

@node Configuration Pragmas
@chapter Configuration Pragmas
@cindex Configuration pragmas
@cindex Pragmas, configuration

@menu
* Handling of Configuration Pragmas::
* The Configuration Pragmas Files::
@end menu

@noindent
Configuration pragmas include those pragmas described as
such in the Ada Reference Manual, as well as
implementation-dependent pragmas that are configuration pragmas.
@xref{Implementation Defined Pragmas,,, gnat_rm, GNAT Reference Manual},
for details on these additional GNAT-specific configuration pragmas.
Most notably, the pragma @code{Source_File_Name}, which allows
specifying non-default names for source files, is a configuration
pragma. The following is a complete list of configuration pragmas
recognized by GNAT:

@smallexample
   Ada_83
   Ada_95
   Ada_05
   Ada_2005
   Ada_12
   Ada_2012
   Allow_Integer_Address
   Annotate
   Assertion_Policy
   Assume_No_Invalid_Values
   C_Pass_By_Copy
   Check_Name
   Check_Policy
   Compile_Time_Error
   Compile_Time_Warning
   Compiler_Unit
   Component_Alignment
   Convention_Identifier
   Debug_Policy
   Detect_Blocking
   Default_Storage_Pool
   Discard_Names
   Elaboration_Checks
   Eliminate
   Extend_System
   Extensions_Allowed
   External_Name_Casing
   Fast_Math
   Favor_Top_Level
   Float_Representation
   Implicit_Packing
   Initialize_Scalars
   Interrupt_State
   License
   Locking_Policy
   Long_Float
   No_Run_Time
   No_Strict_Aliasing
   Normalize_Scalars
   Optimize_Alignment
   Persistent_BSS
   Polling
   Priority_Specific_Dispatching
   Profile
   Profile_Warnings
   Propagate_Exceptions
   Queuing_Policy
   Ravenscar
   Restricted_Run_Time
   Restrictions
   Restrictions_Warnings
   Reviewable
   Short_Circuit_And_Or
   Source_File_Name
   Source_File_Name_Project
   SPARK_Mode
   Style_Checks
   Suppress
   Suppress_Exception_Locations
   Task_Dispatching_Policy
   Universal_Data
   Unsuppress
   Use_VADS_Size
   Validity_Checks
   Warnings
   Wide_Character_Encoding
@end smallexample

@node Handling of Configuration Pragmas
@section Handling of Configuration Pragmas

Configuration pragmas may either appear at the start of a compilation
unit, or they can appear in a configuration pragma file to apply to
all compilations performed in a given compilation environment.

GNAT also provides the @code{gnatchop} utility to provide an automatic
way to handle configuration pragmas following the semantics for
compilations (that is, files with multiple units), described in the RM.
See @ref{Operating gnatchop in Compilation Mode} for details.
However, for most purposes, it will be more convenient to edit the
@file{gnat.adc} file that contains configuration pragmas directly,
as described in the following section.

In the case of @code{Restrictions} pragmas appearing as configuration
pragmas in individual compilation units, the exact handling depends on
the type of restriction.

Restrictions that require partition-wide consistency (like
@code{No_Tasking}) are
recognized wherever they appear
and can be freely inherited, e.g. from a with'ed unit to the with'ing
unit. This makes sense since the binder will in any case insist on seeing
consistent use, so any unit not conforming to any restrictions that are
anywhere in the partition will be rejected, and you might as well find
that out at compile time rather than at bind time.

For restrictions that do not require partition-wide consistency, e.g.
SPARK or No_Implementation_Attributes, in general the restriction applies
only to the unit in which the pragma appears, and not to any other units.

The exception is No_Elaboration_Code which always applies to the entire
object file from a compilation, i.e. to the body, spec, and all subunits.
This restriction can be specified in a configuration pragma file, or it
can be on the body and/or the spec (in eithe case it applies to all the
relevant units). It can appear on a subunit only if it has previously
appeared in the body of spec.

@node The Configuration Pragmas Files
@section The Configuration Pragmas Files
@cindex @file{gnat.adc}

@noindent
In GNAT a compilation environment is defined by the current
directory at the time that a compile command is given. This current
directory is searched for a file whose name is @file{gnat.adc}. If
this file is present, it is expected to contain one or more
configuration pragmas that will be applied to the current compilation.
However, if the switch @option{-gnatA} is used, @file{gnat.adc} is not
considered. When taken into account, @file{gnat.adc} is added to the
dependencies, so that if @file{gnat.adc} is modified later, an invocation of
@command{gnatmake} will recompile the source.

Configuration pragmas may be entered into the @file{gnat.adc} file
either by running @code{gnatchop} on a source file that consists only of
configuration pragmas, or more conveniently by direct editing of the
@file{gnat.adc} file, which is a standard format source file.

In addition to @file{gnat.adc}, additional files containing configuration
pragmas may be applied to the current compilation using the switch
@option{-gnatec=}@var{path}. @var{path} must designate an existing file that
contains only configuration pragmas. These configuration pragmas are
in addition to those found in @file{gnat.adc} (provided @file{gnat.adc}
is present and switch @option{-gnatA} is not used).

It is allowable to specify several switches @option{-gnatec=}, all of which
will be taken into account.

Files containing configuration pragmas specified with switches
@option{-gnatec=} are added to the dependencies, unless they are
temporary files. A file is considered temporary if its name ends in
@file{.tmp} or @file{.TMP}. Certain tools follow this naming
convention because they pass information to @command{gcc} via
temporary files that are immediately deleted; it doesn't make sense to
depend on a file that no longer exists. Such tools include
@command{gprbuild}, @command{gnatmake}, and @command{gnatcheck}.

If you are using project file, a separate mechanism is provided using
project attributes, see @ref{Specifying Configuration Pragmas} for more
details.


@node Handling Arbitrary File Naming Conventions with gnatname
@chapter Handling Arbitrary File Naming Conventions with @code{gnatname}
@cindex Arbitrary File Naming Conventions

@menu
* Arbitrary File Naming Conventions::
* Running gnatname::
* Switches for gnatname::
* Examples of gnatname Usage::
@end menu

@node Arbitrary File Naming Conventions
@section Arbitrary File Naming Conventions

@noindent
The GNAT compiler must be able to know the source file name of a compilation
unit.  When using the standard GNAT default file naming conventions
(@code{.ads} for specs, @code{.adb} for bodies), the GNAT compiler
does not need additional information.

@noindent
When the source file names do not follow the standard GNAT default file naming
conventions, the GNAT compiler must be given additional information through
a configuration pragmas file (@pxref{Configuration Pragmas})
or a project file.
When the non-standard file naming conventions are well-defined,
a small number of pragmas @code{Source_File_Name} specifying a naming pattern
(@pxref{Alternative File Naming Schemes}) may be sufficient. However,
if the file naming conventions are irregular or arbitrary, a number
of pragma @code{Source_File_Name} for individual compilation units
must be defined.
To help maintain the correspondence between compilation unit names and
source file names within the compiler,
GNAT provides a tool @code{gnatname} to generate the required pragmas for a
set of files.

@node Running gnatname
@section Running @code{gnatname}

@noindent
The usual form of the @code{gnatname} command is

@smallexample
@c $ gnatname @ovar{switches} @var{naming_pattern} @ovar{naming_patterns}
@c       @r{[}--and @ovar{switches} @var{naming_pattern} @ovar{naming_patterns}@r{]}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatname @r{[}@var{switches}@r{]} @var{naming_pattern} @r{[}@var{naming_patterns}@r{]}
      @r{[}--and @r{[}@var{switches}@r{]} @var{naming_pattern} @r{[}@var{naming_patterns}@r{]}@r{]}
@end smallexample

@noindent
All of the arguments are optional. If invoked without any argument,
@code{gnatname} will display its usage.

@noindent
When used with at least one naming pattern, @code{gnatname} will attempt to
find all the compilation units in files that follow at least one of the
naming patterns. To find these compilation units,
@code{gnatname} will use the GNAT compiler in syntax-check-only mode on all
regular files.

@noindent
One or several Naming Patterns may be given as arguments to @code{gnatname}.
Each Naming Pattern is enclosed between double quotes (or single
quotes on Windows).
A Naming Pattern is a regular expression similar to the wildcard patterns
used in file names by the Unix shells or the DOS prompt.

@noindent
@code{gnatname} may be called with several sections of directories/patterns.
Sections are separated by switch @code{--and}. In each section, there must be
at least one pattern. If no directory is specified in a section, the current
directory (or the project directory is @code{-P} is used) is implied.
The options other that the directory switches and the patterns apply globally
even if they are in different sections.

@noindent
Examples of Naming Patterns are

@smallexample
   "*.[12].ada"
   "*.ad[sb]*"
   "body_*"    "spec_*"
@end smallexample

@noindent
For a more complete description of the syntax of Naming Patterns,
see the second kind of regular expressions described in @file{g-regexp.ads}
(the ``Glob'' regular expressions).

@noindent
When invoked with no switch @code{-P}, @code{gnatname} will create a
configuration pragmas file @file{gnat.adc} in the current working directory,
with pragmas @code{Source_File_Name} for each file that contains a valid Ada
unit.

@node Switches for gnatname
@section Switches for @code{gnatname}

@noindent
Switches for @code{gnatname} must precede any specified Naming Pattern.

@noindent
You may specify any of the following switches to @code{gnatname}:

@table @option
@c !sort!

@item --version
@cindex @option{--version} @command{gnatname}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatname}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item --subdirs=<dir>
Real object, library or exec directories are subdirectories <dir> of the
specified ones.

@item --no-backup
Do not create a backup copy of an existing project file.

@item --and
Start another section of directories/patterns.

@item -c@file{file}
@cindex @option{-c} (@code{gnatname})
Create a configuration pragmas file @file{file} (instead of the default
@file{gnat.adc}).
There may be zero, one or more space between @option{-c} and
@file{file}.
@file{file} may include directory information. @file{file} must be
writable. There may be only one switch @option{-c}.
When a switch @option{-c} is
specified, no switch @option{-P} may be specified (see below).

@item -d@file{dir}
@cindex @option{-d} (@code{gnatname})
Look for source files in directory @file{dir}. There may be zero, one or more
spaces between @option{-d} and @file{dir}.
@file{dir} may end with @code{/**}, that is it may be of the form
@code{root_dir/**}. In this case, the directory @code{root_dir} and all of its
subdirectories, recursively, have to be searched for sources.
When a switch @option{-d}
is specified, the current working directory will not be searched for source
files, unless it is explicitly specified with a @option{-d}
or @option{-D} switch.
Several switches @option{-d} may be specified.
If @file{dir} is a relative path, it is relative to the directory of
the configuration pragmas file specified with switch
@option{-c},
or to the directory of the project file specified with switch
@option{-P} or,
if neither switch @option{-c}
nor switch @option{-P} are specified, it is relative to the
current working directory. The directory
specified with switch @option{-d} must exist and be readable.

@item -D@file{file}
@cindex @option{-D} (@code{gnatname})
Look for source files in all directories listed in text file @file{file}.
There may be zero, one or more spaces between @option{-D}
and @file{file}.
@file{file} must be an existing, readable text file.
Each nonempty line in @file{file} must be a directory.
Specifying switch @option{-D} is equivalent to specifying as many
switches @option{-d} as there are nonempty lines in
@file{file}.

@item -eL
Follow symbolic links when processing project files.

@item -f@file{pattern}
@cindex @option{-f} (@code{gnatname})
Foreign patterns. Using this switch, it is possible to add sources of languages
other than Ada to the list of sources of a project file.
It is only useful if a -P switch is used.
For example,
@smallexample
gnatname -Pprj -f"*.c" "*.ada"
@end smallexample
@noindent
will look for Ada units in all files with the @file{.ada} extension,
and will add to the list of file for project @file{prj.gpr} the C files
with extension @file{.c}.

@item -h
@cindex @option{-h} (@code{gnatname})
Output usage (help) information. The output is written to @file{stdout}.

@item -P@file{proj}
@cindex @option{-P} (@code{gnatname})
Create or update project file @file{proj}. There may be zero, one or more space
between @option{-P} and @file{proj}. @file{proj} may include directory
information. @file{proj} must be writable.
There may be only one switch @option{-P}.
When a switch @option{-P} is specified,
no switch @option{-c} may be specified.
On all platforms, except on VMS, when @code{gnatname} is invoked for an
existing project file <proj>.gpr, a backup copy of the project file is created
in the project directory with file name <proj>.gpr.saved_x. 'x' is the first
non negative number that makes this backup copy a new file.

@item -v
@cindex @option{-v} (@code{gnatname})
Verbose mode. Output detailed explanation of behavior to @file{stdout}.
This includes name of the file written, the name of the directories to search
and, for each file in those directories whose name matches at least one of
the Naming Patterns, an indication of whether the file contains a unit,
and if so the name of the unit.

@item -v -v
@cindex @option{-v -v} (@code{gnatname})
Very Verbose mode. In addition to the output produced in verbose mode,
for each file in the searched directories whose name matches none of
the Naming Patterns, an indication is given that there is no match.

@item -x@file{pattern}
@cindex @option{-x} (@code{gnatname})
Excluded patterns. Using this switch, it is possible to exclude some files
that would match the name patterns. For example,
@smallexample
gnatname -x "*_nt.ada" "*.ada"
@end smallexample
@noindent
will look for Ada units in all files with the @file{.ada} extension,
except those whose names end with @file{_nt.ada}.

@end table

@node Examples of gnatname Usage
@section Examples of @code{gnatname} Usage


@smallexample
$ gnatname -c /home/me/names.adc -d sources "[a-z]*.ada*"
@end smallexample

@noindent
In this example, the directory @file{/home/me} must already exist
and be writable. In addition, the directory
@file{/home/me/sources} (specified by
@option{-d sources}) must exist and be readable.

Note the optional spaces after @option{-c} and @option{-d}.

@smallexample
$ gnatname -P/home/me/proj -x "*_nt_body.ada"
  -dsources -dsources/plus -Dcommon_dirs.txt "body_*" "spec_*"
@end smallexample

Note that several switches @option{-d} may be used,
even in conjunction with one or several switches
@option{-D}. Several Naming Patterns and one excluded pattern
are used in this example.

@c *****************************************
@c * G N A T  P r o j e c t  M a n a g e r *
@c *****************************************

@c ------ macros for projects.texi
@c These macros are needed when building the gprbuild documentation, but
@c should have no effect in the gnat user's guide

@macro CODESAMPLE{TXT}
@smallexample
@group
\TXT\
@end group
@end smallexample
@end macro

@macro PROJECTFILE{TXT}
@CODESAMPLE{\TXT\}
@end macro

@c simulates a newline when in a @CODESAMPLE
@macro NL{}
@end macro

@macro TIP{TXT}
@quotation
@noindent
\TXT\
@end quotation
@end macro

@macro TIPHTML{TXT}
\TXT\
@end macro

@macro IMPORTANT{TXT}
@quotation
@noindent
\TXT\
@end quotation

@end macro

@macro NOTE{TXT}
@quotation
@noindent
\TXT\
@end quotation
@end macro

@include projects.texi

@c ---------------------------------------------
@c Tools Supporting Project Files
@c ---------------------------------------------

@node Tools Supporting Project Files
@chapter Tools Supporting Project Files

@noindent

@menu
* gnatmake and Project Files::
* The GNAT Driver and Project Files::
@end menu

@c ---------------------------------------------
@node gnatmake and Project Files
@section gnatmake and Project Files
@c ---------------------------------------------

@noindent
This section covers several topics related to @command{gnatmake} and
project files: defining switches for @command{gnatmake}
and for the tools that it invokes; specifying configuration pragmas;
the use of the @code{Main} attribute; building and rebuilding library project
files.

@menu
* Switches Related to Project Files::
* Switches and Project Files::
* Specifying Configuration Pragmas::
* Project Files and Main Subprograms::
* Library Project Files::
@end menu

@c ---------------------------------------------
@node Switches Related to Project Files
@subsection Switches Related to Project Files
@c ---------------------------------------------

@noindent
The following switches are used by GNAT tools that support project files:

@table @option

@item -P@var{project}
@cindex @option{-P} (any project-aware tool)
Indicates the name of a project file. This project file will be parsed with
the verbosity indicated by @option{-vP@emph{x}},
if any, and using the external references indicated
by @option{-X} switches, if any.
There may zero, one or more spaces between @option{-P} and @var{project}.

There must be only one @option{-P} switch on the command line.

Since the Project Manager parses the project file only after all the switches
on the command line are checked, the order of the switches
@option{-P},
@option{-vP@emph{x}}
or @option{-X} is not significant.

@item -X@var{name=value}
@cindex @option{-X} (any project-aware tool)
Indicates that external variable @var{name} has the value @var{value}.
The Project Manager will use this value for occurrences of
@code{external(name)} when parsing the project file.

If @var{name} or @var{value} includes a space, then @var{name=value} should be
put between quotes.
@smallexample
  -XOS=NT
  -X"user=John Doe"
@end smallexample

Several @option{-X} switches can be used simultaneously.
If several @option{-X} switches specify the same
@var{name}, only the last one is used.

An external variable specified with a @option{-X} switch
takes precedence over the value of the same name in the environment.

@item -vP@emph{x}
@cindex @option{-vP} (any project-aware tool)
Indicates the verbosity of the parsing of GNAT project files.

@option{-vP0} means Default;
@option{-vP1} means Medium;
@option{-vP2} means High.


The default is Default: no output for syntactically correct
project files.
If several @option{-vP@emph{x}} switches are present,
only the last one is used.

@item -aP<dir>
@cindex @option{-aP} (any project-aware tool)
Add directory <dir> at the beginning of the project search path, in order,
after the current working directory.

@item -eL
@cindex @option{-eL} (any project-aware tool)
Follow all symbolic links when processing project files.

@item --subdirs=<subdir>
@cindex @option{--subdirs=} (gnatmake and gnatclean)
This switch is recognized by @command{gnatmake} and @command{gnatclean}. It
indicate that the real directories (except the source directories) are the
subdirectories <subdir> of the directories specified in the project files.
This applies in particular to object directories, library directories and
exec directories. If the subdirectories do not exist, they are created
automatically.

@end table

@c ---------------------------------------------
@node Switches and Project Files
@subsection Switches and Project Files
@c ---------------------------------------------

@noindent

For each of the packages @code{Builder}, @code{Compiler}, @code{Binder}, and
@code{Linker}, you can specify a @code{Default_Switches}
attribute, a @code{Switches} attribute, or both;
as their names imply, these switch-related
attributes affect the switches that are used for each of these GNAT
components when
@command{gnatmake} is invoked.  As will be explained below, these
component-specific switches precede
the switches provided on the @command{gnatmake} command line.

The @code{Default_Switches} attribute is an attribute
indexed by language name (case insensitive) whose value is a string list.
For example:

@smallexample @c projectfile
@group
@b{package} Compiler @b{is}
  @b{for} Default_Switches ("Ada")
      @b{use} ("-gnaty",
           "-v");
@b{end} Compiler;
@end group
@end smallexample

@noindent
The @code{Switches} attribute is indexed on a file name (which may or may
not be case sensitive, depending
on the operating system) whose value is a string list.  For example:

@smallexample @c projectfile
@group
@b{package} Builder @b{is}
   @b{for} Switches ("main1.adb")
       @b{use} ("-O2");
   @b{for} Switches ("main2.adb")
       @b{use} ("-g");
@b{end} Builder;
@end group
@end smallexample

@noindent
For the @code{Builder} package, the file names must designate source files
for main subprograms.  For the @code{Binder} and @code{Linker} packages, the
file names must designate @file{ALI} or source files for main subprograms.
In each case just the file name without an explicit extension is acceptable.

For each tool used in a program build (@command{gnatmake}, the compiler, the
binder, and the linker), the corresponding package @dfn{contributes} a set of
switches for each file on which the tool is invoked, based on the
switch-related attributes defined in the package.
In particular, the switches
that each of these packages contributes for a given file @var{f} comprise:

@itemize @bullet
@item the value of attribute @code{Switches (@var{f})},
  if it is specified in the package for the given file,
@item otherwise, the value of @code{Default_Switches ("Ada")},
  if it is specified in the package.

@end itemize

@noindent
If neither of these attributes is defined in the package, then the package does
not contribute any switches for the given file.

When @command{gnatmake} is invoked on a file, the switches comprise
two sets, in the following order: those contributed for the file
by the @code{Builder} package;
and the switches passed on the command line.

When @command{gnatmake} invokes a tool (compiler, binder, linker) on a file,
the switches passed to the tool comprise three sets,
in the following order:

@enumerate
@item
the applicable switches contributed for the file
by the @code{Builder} package in the project file supplied on the command line;

@item
those contributed for the file by the package (in the relevant project file --
see below) corresponding to the tool; and

@item
the applicable switches passed on the command line.
@end enumerate

The term @emph{applicable switches} reflects the fact that
@command{gnatmake} switches may or may not be passed to individual
tools, depending on the individual switch.

@command{gnatmake} may invoke the compiler on source files from different
projects. The Project Manager will use the appropriate project file to
determine the @code{Compiler} package for each source file being compiled.
Likewise for the @code{Binder} and @code{Linker} packages.

As an example, consider the following package in a project file:

@smallexample @c projectfile
@group
@b{project} Proj1 @b{is}
   @b{package} Compiler @b{is}
      @b{for} Default_Switches ("Ada")
          @b{use} ("-g");
      @b{for} Switches ("a.adb")
          @b{use} ("-O1");
      @b{for} Switches ("b.adb")
          @b{use} ("-O2",
               "-gnaty");
   @b{end} Compiler;
@b{end} Proj1;
@end group
@end smallexample

@noindent
If @command{gnatmake} is invoked with this project file, and it needs to
compile, say, the files @file{a.adb}, @file{b.adb}, and @file{c.adb}, then
@file{a.adb} will be compiled with the switch
@option{-O1},
@file{b.adb} with switches
@option{-O2}
and @option{-gnaty},
and @file{c.adb} with @option{-g}.

The following example illustrates the ordering of the switches
contributed by different packages:

@smallexample @c projectfile
@group
@b{project} Proj2 @b{is}
   @b{package} Builder @b{is}
      @b{for} Switches ("main.adb")
          @b{use} ("-g",
               "-O1",
               "-f");
   @b{end} Builder;
@end group

@group
   @b{package} Compiler @b{is}
      @b{for} Switches ("main.adb")
          @b{use} ("-O2");
   @b{end} Compiler;
@b{end} Proj2;
@end group
@end smallexample

@noindent
If you issue the command:

@smallexample
    gnatmake -Pproj2 -O0 main
@end smallexample

@noindent
then the compiler will be invoked on @file{main.adb} with the following
sequence of switches

@smallexample
   -g -O1 -O2 -O0
@end smallexample

@noindent
with the last @option{-O}
switch having precedence over the earlier ones;
several other switches
(such as @option{-c}) are added implicitly.

The switches
@option{-g}
and @option{-O1} are contributed by package
@code{Builder},  @option{-O2} is contributed
by the package @code{Compiler}
and @option{-O0} comes from the command line.

The @option{-g}
switch will also be passed in the invocation of
@command{Gnatlink.}

A final example illustrates switch contributions from packages in different
project files:

@smallexample @c projectfile
@group
@b{project} Proj3 @b{is}
   @b{for} Source_Files @b{use} ("pack.ads", "pack.adb");
   @b{package} Compiler @b{is}
      @b{for} Default_Switches ("Ada")
          @b{use} ("-gnata");
   @b{end} Compiler;
@b{end} Proj3;
@end group

@group
@b{with} "Proj3";
@b{project} Proj4 @b{is}
   @b{for} Source_Files @b{use} ("foo_main.adb", "bar_main.adb");
   @b{package} Builder @b{is}
      @b{for} Switches ("foo_main.adb")
          @b{use} ("-s",
               "-g");
   @b{end} Builder;
@b{end} Proj4;
@end group

@group
--@i{ Ada source file:}
@b{with} Pack;
@b{procedure} Foo_Main @b{is}
   @dots{}
@b{end} Foo_Main;
@end group
@end smallexample

@noindent
If the command is
@smallexample
gnatmake -PProj4 foo_main.adb -cargs -gnato
@end smallexample

@noindent
then the switches passed to the compiler for @file{foo_main.adb} are
@option{-g} (contributed by the package @code{Proj4.Builder}) and
@option{-gnato} (passed on the command line).
When the imported package @code{Pack} is compiled, the switches used
are @option{-g} from @code{Proj4.Builder},
@option{-gnata} (contributed from package @code{Proj3.Compiler},
and @option{-gnato} from the command line.

When using @command{gnatmake} with project files, some switches or
arguments may be expressed as relative paths. As the working directory where
compilation occurs may change, these relative paths are converted to absolute
paths. For the switches found in a project file, the relative paths
are relative to the project file directory, for the switches on the command
line, they are relative to the directory where @command{gnatmake} is invoked.
The switches for which this occurs are:
-I,
-A,
-L,
-aO,
-aL,
-aI, as well as all arguments that are not switches (arguments to
switch
-o, object files specified in package @code{Linker} or after
-largs on the command line). The exception to this rule is the switch
--RTS= for which a relative path argument is never converted.

@c ---------------------------------------------
@node Specifying Configuration Pragmas
@subsection Specifying Configuration Pragmas
@c ---------------------------------------------

@noindent
When using @command{gnatmake} with project files, if there exists a file
@file{gnat.adc} that contains configuration pragmas, this file will be
ignored.

Configuration pragmas can be defined by means of the following attributes in
project files: @code{Global_Configuration_Pragmas} in package @code{Builder}
and @code{Local_Configuration_Pragmas} in package @code{Compiler}.

Both these attributes are single string attributes. Their values is the path
name of a file containing configuration pragmas. If a path name is relative,
then it is relative to the project directory of the project file where the
attribute is defined.

When compiling a source, the configuration pragmas used are, in order,
those listed in the file designated by attribute
@code{Global_Configuration_Pragmas} in package @code{Builder} of the main
project file, if it is specified, and those listed in the file designated by
attribute @code{Local_Configuration_Pragmas} in package @code{Compiler} of
the project file of the source, if it exists.

@c ---------------------------------------------
@node Project Files and Main Subprograms
@subsection Project Files and Main Subprograms
@c ---------------------------------------------

@noindent
When using a project file, you can invoke @command{gnatmake}
with one or several main subprograms, by specifying their source files on the
command line.

@smallexample
    gnatmake -Pprj main1.adb main2.adb main3.adb
@end smallexample

@noindent
Each of these needs to be a source file of the same project, except
when the switch -u is used.

When -u is not used, all the mains need to be sources of the
same project, one of the project in the tree rooted at the project specified
on the command line. The package @code{Builder} of this common project, the
"main project" is the one that is considered by @command{gnatmake}.

When -u is used, the specified source files may be in projects
imported directly or indirectly by the project specified on the command line.
Note that if such a source file is not part of the project specified on the
command line, the switches found in package @code{Builder} of the
project specified on the command line, if any, that are transmitted
to the compiler will still be used, not those found in the project file of
the source file.

When using a project file, you can also invoke @command{gnatmake} without
explicitly specifying any main, and the effect depends on whether you have
defined the @code{Main} attribute.  This attribute has a string list value,
where each element in the list is the name of a source file (the file
extension is optional) that contains a unit that can be a main subprogram.

If the @code{Main} attribute is defined in a project file as a non-empty
string list and the switch @option{-u} is not used on the command
line, then invoking @command{gnatmake} with this project file but without any
main on the command line is equivalent to invoking @command{gnatmake} with all
the file names in the @code{Main} attribute on the command line.

Example:
@smallexample @c projectfile
@group
   @b{project} Prj @b{is}
      @b{for} Main @b{use} ("main1.adb", "main2.adb", "main3.adb");
   @b{end} Prj;
@end group
@end smallexample

@noindent
With this project file, @code{"gnatmake -Pprj"}
is equivalent to
@code{"gnatmake -Pprj main1.adb main2.adb main3.adb"}.

When the project attribute @code{Main} is not specified, or is specified
as an empty string list, or when the switch @option{-u} is used on the command
line, then invoking @command{gnatmake} with no main on the command line will
result in all immediate sources of the project file being checked, and
potentially recompiled. Depending on the presence of the switch @option{-u},
sources from other project files on which the immediate sources of the main
project file depend are also checked and potentially recompiled. In other
words, the @option{-u} switch is applied to all of the immediate sources of the
main project file.

When no main is specified on the command line and attribute @code{Main} exists
and includes several mains, or when several mains are specified on the
command line, the default switches in package @code{Builder} will
be used for all mains, even if there are specific switches
specified for one or several mains.

But the switches from package @code{Binder} or @code{Linker} will be
the specific switches for each main, if they are specified.

@c ---------------------------------------------
@node Library Project Files
@subsection Library Project Files
@c ---------------------------------------------

@noindent
When @command{gnatmake} is invoked with a main project file that is a library
project file, it is not allowed to specify one or more mains on the command
line.

When a library project file is specified, switches -b and
-l have special meanings.

@itemize @bullet
@item -b is only allowed for stand-alone libraries. It indicates
  to @command{gnatmake} that @command{gnatbind} should be invoked for the
  library.

@item -l may be used for all library projects. It indicates
  to @command{gnatmake} that the binder generated file should be compiled
  (in the case of a stand-alone library) and that the library should be built.
@end itemize

@c ---------------------------------------------
@node The GNAT Driver and Project Files
@section The GNAT Driver and Project Files
@c ---------------------------------------------

@noindent
A number of GNAT tools, other than @command{gnatmake}
can benefit from project files:
(@command{gnatbind},
@ifclear FSFEDITION
@command{gnatcheck},
@end ifclear
@command{gnatclean},
@ifclear FSFEDITION
@command{gnatelim},
@end ifclear
@command{gnatfind},
@command{gnatlink},
@command{gnatls},
@ifclear FSFEDITION
@command{gnatmetric},
@command{gnatpp},
@command{gnatstub},
@end ifclear
and @command{gnatxref}). However, none of these tools can be invoked
directly with a project file switch (@option{-P}).
They must be invoked through the @command{gnat} driver.

The @command{gnat} driver is a wrapper that accepts a number of commands and
calls the corresponding tool. It was designed initially for VMS platforms (to
convert VMS qualifiers to Unix-style switches), but it is now available on all
GNAT platforms.

On non-VMS platforms, the @command{gnat} driver accepts the following commands
(case insensitive):

@itemize @bullet
@item BIND to invoke @command{gnatbind}
@item CHOP to invoke @command{gnatchop}
@item CLEAN to invoke @command{gnatclean}
@item COMP or COMPILE to invoke the compiler
@ifclear FSFEDITION
@item ELIM to invoke @command{gnatelim}
@end ifclear
@item FIND to invoke @command{gnatfind}
@item KR or KRUNCH to invoke @command{gnatkr}
@item LINK to invoke @command{gnatlink}
@item LS or LIST to invoke @command{gnatls}
@item MAKE to invoke @command{gnatmake}
@item NAME to invoke @command{gnatname}
@item PREP or PREPROCESS to invoke @command{gnatprep}
@ifclear FSFEDITION
@item PP or PRETTY to invoke @command{gnatpp}
@item METRIC to invoke @command{gnatmetric}
@item STUB to invoke @command{gnatstub}
@end ifclear
@item XREF to invoke @command{gnatxref}

@end itemize

@noindent
(note that the compiler is invoked using the command
@command{gnatmake -f -u -c}).

On non-VMS platforms, between @command{gnat} and the command, two
special switches may be used:

@itemize @bullet
@item @command{-v} to display the invocation of the tool.
@item @command{-dn} to prevent the @command{gnat} driver from removing
  the temporary files it has created. These temporary files are
  configuration files and temporary file list files.

@end itemize

@noindent
The command may be followed by switches and arguments for the invoked
tool.

@smallexample
  gnat bind -C main.ali
  gnat ls -a main
  gnat chop foo.txt
@end smallexample

@noindent
Switches may also be put in text files, one switch per line, and the text
files may be specified with their path name preceded by '@@'.

@smallexample
   gnat bind @@args.txt main.ali
@end smallexample

@noindent
In addition, for commands BIND, COMP or COMPILE, FIND,
@ifclear FSFEDITION
ELIM,
@end ifclear
LS or LIST, LINK,
@ifclear FSFEDITION
METRIC,
PP or PRETTY,
STUB,
@end ifclear
and XREF, the project file related switches
(@option{-P},
@option{-X} and
@option{-vPx}) may be used in addition to
the switches of the invoking tool.

@ifclear FSFEDITION
When GNAT PP or GNAT PRETTY is used with a project file, but with no source
specified on the command line, it invokes @command{gnatpp} with all
the immediate sources of the specified project file.
@end ifclear

@ifclear FSFEDITION
When GNAT METRIC is used with a project file, but with no source
specified on the command line, it invokes @command{gnatmetric}
with all the immediate sources of the specified project file and with
@option{-d} with the parameter pointing to the object directory
of the project.
@end ifclear

@ifclear FSFEDITION
In addition, when GNAT PP, GNAT PRETTY or GNAT METRIC is used with
a project file, no source is specified on the command line and
switch -U is specified on the command line, then
the underlying tool (gnatpp or
gnatmetric) is invoked for all sources of all projects,
not only for the immediate sources of the main project.
(-U stands for Universal or Union of the project files of the project tree)
@end ifclear

For each of the following commands, there is optionally a corresponding
package in the main project.

@itemize @bullet
@item package @code{Binder} for command BIND (invoking @code{gnatbind})

@ifclear FSFEDITION
@item package @code{Check} for command CHECK (invoking
  @code{gnatcheck})
@end ifclear

@item package @code{Compiler} for command COMP or COMPILE (invoking the compiler)

@item package @code{Cross_Reference} for command XREF (invoking
  @code{gnatxref})

@ifclear FSFEDITION
@item package @code{Eliminate} for command ELIM (invoking
  @code{gnatelim})
@end ifclear

@item package @code{Finder} for command FIND (invoking @code{gnatfind})

@item package @code{Gnatls} for command LS or LIST (invoking @code{gnatls})

@ifclear FSFEDITION
@item package @code{Gnatstub} for command STUB
  (invoking @code{gnatstub})
@end ifclear

@item package @code{Linker} for command LINK (invoking @code{gnatlink})

@ifclear FSFEDITION
@item package @code{Check} for command CHECK
  (invoking @code{gnatcheck})
@end ifclear

@ifclear FSFEDITION
@item package @code{Metrics} for command METRIC
  (invoking @code{gnatmetric})
@end ifclear

@ifclear FSFEDITION
@item package @code{Pretty_Printer} for command PP or PRETTY
  (invoking @code{gnatpp})
@end ifclear

@end itemize

@noindent
Package @code{Gnatls} has a unique attribute @code{Switches},
a simple variable with a string list value. It contains switches
for the invocation of @code{gnatls}.

@smallexample @c projectfile
@group
@b{project} Proj1 @b{is}
   @b{package} gnatls @b{is}
      @b{for} Switches
          @b{use} ("-a",
               "-v");
   @b{end} gnatls;
@b{end} Proj1;
@end group
@end smallexample

@noindent
All other packages have two attribute @code{Switches} and
@code{Default_Switches}.

@code{Switches} is an indexed attribute, indexed by the
source file name, that has a string list value: the switches to be
used when the tool corresponding to the package is invoked for the specific
source file.

@code{Default_Switches} is an attribute,
indexed by  the programming language that has a string list value.
@code{Default_Switches ("Ada")} contains the
switches for the invocation of the tool corresponding
to the package, except if a specific @code{Switches} attribute
is specified for the source file.

@smallexample @c projectfile
@group
@b{project} Proj @b{is}

   @b{for} Source_Dirs @b{use} ("**");

   @b{package} gnatls @b{is}
      @b{for} Switches @b{use}
          ("-a",
           "-v");
   @b{end} gnatls;
@end group
@group

   @b{package} Compiler @b{is}
      @b{for} Default_Switches ("Ada")
          @b{use} ("-gnatv",
               "-gnatwa");
   @b{end} Binder;
@end group
@group

   @b{package} Binder @b{is}
      @b{for} Default_Switches ("Ada")
          @b{use} ("-C",
               "-e");
   @b{end} Binder;
@end group
@group

   @b{package} Linker @b{is}
      @b{for} Default_Switches ("Ada")
          @b{use} ("-C");
      @b{for} Switches ("main.adb")
          @b{use} ("-C",
               "-v",
               "-v");
   @b{end} Linker;
@end group
@group

   @b{package} Finder @b{is}
      @b{for} Default_Switches ("Ada")
           @b{use} ("-a",
                "-f");
   @b{end} Finder;
@end group
@group

   @b{package} Cross_Reference @b{is}
      @b{for} Default_Switches ("Ada")
          @b{use} ("-a",
               "-f",
               "-d",
               "-u");
   @b{end} Cross_Reference;
@b{end} Proj;
@end group
@end smallexample

@noindent
With the above project file, commands such as

@smallexample
   gnat comp -Pproj main
   gnat ls -Pproj main
   gnat xref -Pproj main
   gnat bind -Pproj main.ali
   gnat link -Pproj main.ali
@end smallexample

@noindent
will set up the environment properly and invoke the tool with the switches
found in the package corresponding to the tool:
@code{Default_Switches ("Ada")} for all tools,
except @code{Switches ("main.adb")}
for @code{gnatlink}.
@ifclear FSFEDITION
It is also possible to invoke some of the tools,
(@code{gnatcheck},
@code{gnatmetric},
and @code{gnatpp})
on a set of project units thanks to the combination of the switches
@option{-P}, @option{-U} and possibly the main unit when one is interested
in its closure. For instance,
@smallexample
gnat metric -Pproj
@end smallexample

@noindent
will compute the metrics for all the immediate units of project
@code{proj}.
@smallexample
gnat metric -Pproj -U
@end smallexample

@noindent
will compute the metrics for all the units of the closure of projects
rooted at @code{proj}.
@smallexample
gnat metric -Pproj -U main_unit
@end smallexample

@noindent
will compute the metrics for the closure of units rooted at
@code{main_unit}. This last possibility relies implicitly
on @command{gnatbind}'s option @option{-R}. But if the argument files for the
tool invoked by the @command{gnat} driver are explicitly  specified
either directly or through the tool @option{-files} option, then the tool
is called only for these explicitly specified files.
@end ifclear

@c *****************************************
@c * Cross-referencing tools
@c *****************************************

@node The Cross-Referencing Tools gnatxref and gnatfind
@chapter  The Cross-Referencing Tools @code{gnatxref} and @code{gnatfind}
@findex gnatxref
@findex gnatfind

@noindent
The compiler generates cross-referencing information (unless
you set the @samp{-gnatx} switch), which are saved in the @file{.ali} files.
This information indicates where in the source each entity is declared and
referenced. Note that entities in package Standard are not included, but
entities in all other predefined units are included in the output.

Before using any of these two tools, you need to compile successfully your
application, so that GNAT gets a chance to generate the cross-referencing
information.

The two tools @code{gnatxref} and @code{gnatfind} take advantage of this
information to provide the user with the capability to easily locate the
declaration and references to an entity. These tools are quite similar,
the difference being that @code{gnatfind} is intended for locating
definitions and/or references to a specified entity or entities, whereas
@code{gnatxref} is oriented to generating a full report of all
cross-references.

To use these tools, you must not compile your application using the
@option{-gnatx} switch on the @command{gnatmake} command line
(@pxref{The GNAT Make Program gnatmake}). Otherwise, cross-referencing
information will not be generated.

Note: to invoke @code{gnatxref} or @code{gnatfind} with a project file,
use the @code{gnat} driver (see @ref{The GNAT Driver and Project Files}).

@menu
* Switches for gnatxref::
* Switches for gnatfind::
* Project Files for gnatxref and gnatfind::
* Regular Expressions in gnatfind and gnatxref::
* Examples of gnatxref Usage::
* Examples of gnatfind Usage::
@end menu

@node Switches for gnatxref
@section @code{gnatxref} Switches

@noindent
The command invocation for @code{gnatxref} is:
@smallexample
@c $ gnatxref @ovar{switches} @var{sourcefile1} @r{[}@var{sourcefile2} @dots{}@r{]}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatxref @r{[}@var{switches}@r{]} @var{sourcefile1} @r{[}@var{sourcefile2} @dots{}@r{]}
@end smallexample

@noindent
where

@table @var
@item sourcefile1
@itemx sourcefile2
identifies the source files for which a report is to be generated. The
``with''ed units will be processed too. You must provide at least one file.

These file names are considered to be regular expressions, so for instance
specifying @file{source*.adb} is the same as giving every file in the current
directory whose name starts with @file{source} and whose extension is
@file{adb}.

You shouldn't specify any directory name, just base names. @command{gnatxref}
and @command{gnatfind} will be able to locate these files by themselves using
the source path. If you specify directories, no result is produced.

@end table

@noindent
The switches can be:
@table @option
@c !sort!
@item --version
@cindex @option{--version} @command{gnatxref}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatxref}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item -a
@cindex @option{-a} (@command{gnatxref})
If this switch is present, @code{gnatfind} and @code{gnatxref} will parse
the read-only files found in the library search path. Otherwise, these files
will be ignored. This option can be used to protect Gnat sources or your own
libraries from being parsed, thus making @code{gnatfind} and @code{gnatxref}
much faster, and their output much smaller. Read-only here refers to access
or permissions status in the file system for the current user.

@item -aIDIR
@cindex @option{-aIDIR} (@command{gnatxref})
When looking for source files also look in directory DIR. The order in which
source file search is undertaken is the same as for @command{gnatmake}.

@item -aODIR
@cindex @option{-aODIR} (@command{gnatxref})
When searching for library and object files, look in directory
DIR. The order in which library files are searched is the same as for
@command{gnatmake}.

@item -nostdinc
@cindex @option{-nostdinc} (@command{gnatxref})
Do not look for sources in the system default directory.

@item -nostdlib
@cindex @option{-nostdlib} (@command{gnatxref})
Do not look for library files in the system default directory.

@item --ext=@var{extension}
@cindex @option{--ext} (@command{gnatxref})
Specify an alternate ali file extension. The default is @code{ali} and other
extensions (e.g. @code{gli} for C/C++ sources when using @option{-fdump-xref})
may be specified via this switch. Note that if this switch overrides the
default, which means that only the new extension will be considered.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gnatxref})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item -d
@cindex @option{-d} (@command{gnatxref})
If this switch is set @code{gnatxref} will output the parent type
reference for each matching derived types.

@item -f
@cindex @option{-f} (@command{gnatxref})
If this switch is set, the output file names will be preceded by their
directory (if the file was found in the search path). If this switch is
not set, the directory will not be printed.

@item -g
@cindex @option{-g} (@command{gnatxref})
If this switch is set, information is output only for library-level
entities, ignoring local entities. The use of this switch may accelerate
@code{gnatfind} and @code{gnatxref}.

@item -IDIR
@cindex @option{-IDIR} (@command{gnatxref})
Equivalent to @samp{-aODIR -aIDIR}.

@item -pFILE
@cindex @option{-pFILE} (@command{gnatxref})
Specify a project file to use @xref{GNAT Project Manager}.
If you need to use the @file{.gpr}
project files, you should use gnatxref through the GNAT driver
(@command{gnat xref -Pproject}).

By default, @code{gnatxref} and @code{gnatfind} will try to locate a
project file in the current directory.

If a project file is either specified or found by the tools, then the content
of the source directory and object directory lines are added as if they
had been specified respectively by @samp{-aI}
and @samp{-aO}.
@item -u
Output only unused symbols. This may be really useful if you give your
main compilation unit on the command line, as @code{gnatxref} will then
display every unused entity and 'with'ed package.

@item -v
Instead of producing the default output, @code{gnatxref} will generate a
@file{tags} file that can be used by vi. For examples how to use this
feature, see @ref{Examples of gnatxref Usage}. The tags file is output
to the standard output, thus you will have to redirect it to a file.

@end table

@noindent
All these switches may be in any order on the command line, and may even
appear after the file names. They need not be separated by spaces, thus
you can say @samp{gnatxref -ag} instead of
@samp{gnatxref -a -g}.

@node Switches for gnatfind
@section @code{gnatfind} Switches

@noindent
The command line for @code{gnatfind} is:

@smallexample
@c $ gnatfind @ovar{switches} @var{pattern}@r{[}:@var{sourcefile}@r{[}:@var{line}@r{[}:@var{column}@r{]]]}
@c       @r{[}@var{file1} @var{file2} @dots{}]
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatfind @r{[}@var{switches}@r{]} @var{pattern}@r{[}:@var{sourcefile}@r{[}:@var{line}@r{[}:@var{column}@r{]]]}
      @r{[}@var{file1} @var{file2} @dots{}@r{]}
@end smallexample

@noindent
where

@table @var
@item pattern
An entity will be output only if it matches the regular expression found
in @var{pattern}, see @ref{Regular Expressions in gnatfind and gnatxref}.

Omitting the pattern is equivalent to specifying @samp{*}, which
will match any entity. Note that if you do not provide a pattern, you
have to provide both a sourcefile and a line.

Entity names are given in Latin-1, with uppercase/lowercase equivalence
for matching purposes. At the current time there is no support for
8-bit codes other than Latin-1, or for wide characters in identifiers.

@item sourcefile
@code{gnatfind} will look for references, bodies or declarations
of symbols referenced in @file{@var{sourcefile}}, at line @var{line}
and column @var{column}. See @ref{Examples of gnatfind Usage}
for syntax examples.

@item line
is a decimal integer identifying the line number containing
the reference to the entity (or entities) to be located.

@item column
is a decimal integer identifying the exact location on the
line of the first character of the identifier for the
entity reference. Columns are numbered from 1.

@item file1 file2 @dots{}
The search will be restricted to these source files. If none are given, then
the search will be done for every library file in the search path.
These file must appear only after the pattern or sourcefile.

These file names are considered to be regular expressions, so for instance
specifying @file{source*.adb} is the same as giving every file in the current
directory whose name starts with @file{source} and whose extension is
@file{adb}.

The location of the spec of the entity will always be displayed, even if it
isn't in one of @file{@var{file1}}, @file{@var{file2}},@enddots{}  The
occurrences of the entity in the separate units of the ones given on the
command line will also be displayed.

Note that if you specify at least one file in this part, @code{gnatfind} may
sometimes not be able to find the body of the subprograms.

@end table

@noindent
At least one of 'sourcefile' or 'pattern' has to be present on
the command line.

The following switches are available:
@table @option
@c !sort!

@cindex @option{--version} @command{gnatfind}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatfind}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item -a
@cindex @option{-a} (@command{gnatfind})
If this switch is present, @code{gnatfind} and @code{gnatxref} will parse
the read-only files found in the library search path. Otherwise, these files
will be ignored. This option can be used to protect Gnat sources or your own
libraries from being parsed, thus making @code{gnatfind} and @code{gnatxref}
much faster, and their output much smaller. Read-only here refers to access
or permission status in the file system for the current user.

@item -aIDIR
@cindex @option{-aIDIR} (@command{gnatfind})
When looking for source files also look in directory DIR. The order in which
source file search is undertaken is the same as for @command{gnatmake}.

@item -aODIR
@cindex @option{-aODIR} (@command{gnatfind})
When searching for library and object files, look in directory
DIR. The order in which library files are searched is the same as for
@command{gnatmake}.

@item -nostdinc
@cindex @option{-nostdinc} (@command{gnatfind})
Do not look for sources in the system default directory.

@item -nostdlib
@cindex @option{-nostdlib} (@command{gnatfind})
Do not look for library files in the system default directory.

@item --ext=@var{extension}
@cindex @option{--ext} (@command{gnatfind})
Specify an alternate ali file extension. The default is @code{ali} and other
extensions (e.g. @code{gli} for C/C++ sources when using @option{-fdump-xref})
may be specified via this switch. Note that if this switch overrides the
default, which means that only the new extension will be considered.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gnatfind})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item -d
@cindex @option{-d} (@code{gnatfind})
If this switch is set, then @code{gnatfind} will output the parent type
reference for each matching derived types.

@item -e
@cindex @option{-e} (@command{gnatfind})
By default, @code{gnatfind} accept the simple regular expression set for
@samp{pattern}. If this switch is set, then the pattern will be
considered as full Unix-style regular expression.

@item -f
@cindex @option{-f} (@command{gnatfind})
If this switch is set, the output file names will be preceded by their
directory (if the file was found in the search path). If this switch is
not set, the directory will not be printed.

@item -g
@cindex @option{-g} (@command{gnatfind})
If this switch is set, information is output only for library-level
entities, ignoring local entities. The use of this switch may accelerate
@code{gnatfind} and @code{gnatxref}.

@item -IDIR
@cindex @option{-IDIR} (@command{gnatfind})
Equivalent to @samp{-aODIR -aIDIR}.

@item -pFILE
@cindex @option{-pFILE} (@command{gnatfind})
Specify a project file (@pxref{GNAT Project Manager}) to use.
By default, @code{gnatxref} and @code{gnatfind} will try to locate a
project file in the current directory.

If a project file is either specified or found by the tools, then the content
of the source directory and object directory lines are added as if they
had been specified respectively by @samp{-aI} and
@samp{-aO}.

@item -r
@cindex @option{-r} (@command{gnatfind})
By default, @code{gnatfind} will output only the information about the
declaration, body or type completion of the entities. If this switch is
set, the @code{gnatfind} will locate every reference to the entities in
the files specified on the command line (or in every file in the search
path if no file is given on the command line).

@item -s
@cindex @option{-s} (@command{gnatfind})
If this switch is set, then @code{gnatfind} will output the content
of the Ada source file lines were the entity was found.

@item -t
@cindex @option{-t} (@command{gnatfind})
If this switch is set, then @code{gnatfind} will output the type hierarchy for
the specified type. It act like -d option but recursively from parent
type to parent type. When this switch is set it is not possible to
specify more than one file.

@end table

@noindent
All these switches may be in any order on the command line, and may even
appear after the file names. They need not be separated by spaces, thus
you can say @samp{gnatxref -ag} instead of
@samp{gnatxref -a -g}.

As stated previously, gnatfind will search in every directory in the
search path. You can force it to look only in the current directory if
you specify @code{*} at the end of the command line.

@node Project Files for gnatxref and gnatfind
@section Project Files for @command{gnatxref} and @command{gnatfind}

@noindent
Project files allow a programmer to specify how to compile its
application, where to find sources, etc.  These files are used
primarily by GPS, but they can also be used
by the two tools
@code{gnatxref} and @code{gnatfind}.

A project file name must end with @file{.gpr}. If a single one is
present in the current directory, then @code{gnatxref} and @code{gnatfind} will
extract the information from it. If multiple project files are found, none of
them is read, and you have to use the @samp{-p} switch to specify the one
you want to use.

The following lines can be included, even though most of them have default
values which can be used in most cases.
The lines can be entered in any order in the file.
Except for @file{src_dir} and @file{obj_dir}, you can only have one instance of
each line. If you have multiple instances, only the last one is taken into
account.

@table @code
@item src_dir=DIR
[default: @code{"./"}]
specifies a directory where to look for source files. Multiple @code{src_dir}
lines can be specified and they will be searched in the order they
are specified.

@item obj_dir=DIR
[default: @code{"./"}]
specifies a directory where to look for object and library files. Multiple
@code{obj_dir} lines can be specified, and they will be searched in the order
they are specified

@item comp_opt=SWITCHES
[default: @code{""}]
creates a variable which can be referred to subsequently by using
the @code{$@{comp_opt@}} notation. This is intended to store the default
switches given to @command{gnatmake} and @command{gcc}.

@item bind_opt=SWITCHES
[default: @code{""}]
creates a variable which can be referred to subsequently by using
the @samp{$@{bind_opt@}} notation. This is intended to store the default
switches given to @command{gnatbind}.

@item link_opt=SWITCHES
[default: @code{""}]
creates a variable which can be referred to subsequently by using
the @samp{$@{link_opt@}} notation. This is intended to store the default
switches given to @command{gnatlink}.

@item main=EXECUTABLE
[default: @code{""}]
specifies the name of the executable for the application. This variable can
be referred to in the following lines by using the @samp{$@{main@}} notation.

@item comp_cmd=COMMAND
[default: @code{"gcc -c -I$@{src_dir@} -g -gnatq"}]
specifies the command used to compile a single file in the application.

@item make_cmd=COMMAND
[default: @code{"gnatmake $@{main@} -aI$@{src_dir@}
-aO$@{obj_dir@} -g -gnatq -cargs $@{comp_opt@}
-bargs $@{bind_opt@} -largs $@{link_opt@}"}]
specifies the command used to recompile the whole application.

@item run_cmd=COMMAND
[default: @code{"$@{main@}"}]
specifies the command used to run the application.

@item debug_cmd=COMMAND
[default: @code{"gdb $@{main@}"}]
specifies the command used to debug the application

@end table

@noindent
@command{gnatxref} and @command{gnatfind} only take into account the
@code{src_dir} and @code{obj_dir} lines, and ignore the others.

@node Regular Expressions in gnatfind and gnatxref
@section  Regular Expressions in @code{gnatfind} and @code{gnatxref}

@noindent
As specified in the section about @command{gnatfind}, the pattern can be a
regular expression. Actually, there are to set of regular expressions
which are recognized by the program:

@table @code
@item globbing patterns
These are the most usual regular expression. They are the same that you
generally used in a Unix shell command line, or in a DOS session.

Here is a more formal grammar:
@smallexample
@group
@iftex
@leftskip=.5cm
@end iftex
regexp ::= term
term   ::= elmt            -- matches elmt
term   ::= elmt elmt       -- concatenation (elmt then elmt)
term   ::= *               -- any string of 0 or more characters
term   ::= ?               -- matches any character
term   ::= [char @{char@}]   -- matches any character listed
term   ::= [char - char]   -- matches any character in range
@end group
@end smallexample

@item full regular expression
The second set of regular expressions is much more powerful. This is the
type of regular expressions recognized by utilities such a @file{grep}.

The following is the form of a regular expression, expressed in Ada
reference manual style BNF is as follows

@smallexample
@iftex
@leftskip=.5cm
@end iftex
@group
regexp ::= term @{| term@}   -- alternation (term or term @dots{})

term ::= item @{item@}       -- concatenation (item then item)

item ::= elmt              -- match elmt
item ::= elmt *            -- zero or more elmt's
item ::= elmt +            -- one or more elmt's
item ::= elmt ?            -- matches elmt or nothing
@end group
@group
elmt ::= nschar            -- matches given character
elmt ::= [nschar @{nschar@}]   -- matches any character listed
elmt ::= [^ nschar @{nschar@}] -- matches any character not listed
elmt ::= [char - char]     -- matches chars in given range
elmt ::= \ char            -- matches given character
elmt ::= .                 -- matches any single character
elmt ::= ( regexp )        -- parens used for grouping

char ::= any character, including special characters
nschar ::= any character except ()[].*+?^
@end group
@end smallexample

Following are a few examples:

@table @samp
@item abcde|fghi
will match any of the two strings @samp{abcde} and @samp{fghi},

@item abc*d
will match any string like @samp{abd}, @samp{abcd}, @samp{abccd},
@samp{abcccd}, and so on,

@item [a-z]+
will match any string which has only lowercase characters in it (and at
least one character.

@end table
@end table

@node Examples of gnatxref Usage
@section Examples of @code{gnatxref} Usage

@subsection General Usage

@noindent
For the following examples, we will consider the following units:

@smallexample @c ada
@group
@cartouche
main.ads:
1: @b{with} Bar;
2: @b{package} Main @b{is}
3:     @b{procedure} Foo (B : @b{in} Integer);
4:     C : Integer;
5: @b{private}
6:     D : Integer;
7: @b{end} Main;

main.adb:
1: @b{package} @b{body} Main @b{is}
2:     @b{procedure} Foo (B : @b{in} Integer) @b{is}
3:     @b{begin}
4:        C := B;
5:        D := B;
6:        Bar.Print (B);
7:        Bar.Print (C);
8:     @b{end} Foo;
9: @b{end} Main;

bar.ads:
1: @b{package} Bar @b{is}
2:     @b{procedure} Print (B : Integer);
3: @b{end} bar;
@end cartouche
@end group
@end smallexample

@table @code

@noindent
The first thing to do is to recompile your application (for instance, in
that case just by doing a @samp{gnatmake main}, so that GNAT generates
the cross-referencing information.
You can then issue any of the following commands:

@item gnatxref main.adb
@code{gnatxref} generates cross-reference information for main.adb
and every unit 'with'ed by main.adb.

The output would be:
@smallexample
@iftex
@leftskip=0cm
@end iftex
B                                                      Type: Integer
  Decl: bar.ads           2:22
B                                                      Type: Integer
  Decl: main.ads          3:20
  Body: main.adb          2:20
  Ref:  main.adb          4:13     5:13     6:19
Bar                                                    Type: Unit
  Decl: bar.ads           1:9
  Ref:  main.adb          6:8      7:8
       main.ads           1:6
C                                                      Type: Integer
  Decl: main.ads          4:5
  Modi: main.adb          4:8
  Ref:  main.adb          7:19
D                                                      Type: Integer
  Decl: main.ads          6:5
  Modi: main.adb          5:8
Foo                                                    Type: Unit
  Decl: main.ads          3:15
  Body: main.adb          2:15
Main                                                    Type: Unit
  Decl: main.ads          2:9
  Body: main.adb          1:14
Print                                                   Type: Unit
  Decl: bar.ads           2:15
  Ref:  main.adb          6:12     7:12
@end smallexample

@noindent
that is the entity @code{Main} is declared in main.ads, line 2, column 9,
its body is in main.adb, line 1, column 14 and is not referenced any where.

The entity @code{Print} is declared in bar.ads, line 2, column 15 and it
is referenced in main.adb, line 6 column 12 and line 7 column 12.

@item gnatxref package1.adb package2.ads
@code{gnatxref} will generates cross-reference information for
package1.adb, package2.ads and any other package 'with'ed by any
of these.

@end table

@subsection Using gnatxref with vi

@code{gnatxref} can generate a tags file output, which can be used
directly from @command{vi}. Note that the standard version of @command{vi}
will not work properly with overloaded symbols. Consider using another
free implementation of @command{vi}, such as @command{vim}.

@smallexample
$ gnatxref -v gnatfind.adb > tags
@end smallexample

@noindent
will generate the tags file for @code{gnatfind} itself (if the sources
are in the search path!).

From @command{vi}, you can then use the command @samp{:tag @var{entity}}
(replacing @var{entity} by whatever you are looking for), and vi will
display a new file with the corresponding declaration of entity.

@node Examples of gnatfind Usage
@section Examples of @code{gnatfind} Usage

@table @code

@item gnatfind -f xyz:main.adb
Find declarations for all entities xyz referenced at least once in
main.adb. The references are search in every library file in the search
path.

The directories will be printed as well (as the @samp{-f}
switch is set)

The output will look like:
@smallexample
directory/main.ads:106:14: xyz <= declaration
directory/main.adb:24:10: xyz <= body
directory/foo.ads:45:23: xyz <= declaration
@end smallexample

@noindent
that is to say, one of the entities xyz found in main.adb is declared at
line 12 of main.ads (and its body is in main.adb), and another one is
declared at line 45 of foo.ads

@item gnatfind -fs xyz:main.adb
This is the same command as the previous one, instead @code{gnatfind} will
display the content of the Ada source file lines.

The output will look like:

@smallexample
directory/main.ads:106:14: xyz <= declaration
   procedure xyz;
directory/main.adb:24:10: xyz <= body
   procedure xyz is
directory/foo.ads:45:23: xyz <= declaration
   xyz : Integer;
@end smallexample

@noindent
This can make it easier to find exactly the location your are looking
for.

@item gnatfind -r "*x*":main.ads:123 foo.adb
Find references to all entities containing an x that are
referenced on line 123 of main.ads.
The references will be searched only in main.ads and foo.adb.

@item gnatfind main.ads:123
Find declarations and bodies for all entities that are referenced on
line 123 of main.ads.

This is the same as @code{gnatfind "*":main.adb:123}.

@item gnatfind mydir/main.adb:123:45
Find the declaration for the entity referenced at column 45 in
line 123 of file main.adb in directory mydir. Note that it
is usual to omit the identifier name when the column is given,
since the column position identifies a unique reference.

The column has to be the beginning of the identifier, and should not
point to any character in the middle of the identifier.

@end table

@ifclear FSFEDITION
@c *********************************
@node The GNAT Pretty-Printer gnatpp
@chapter The GNAT Pretty-Printer @command{gnatpp}
@findex gnatpp
@cindex Pretty-Printer

@menu
* Switches for gnatpp::
* Formatting Rules::
@end menu

@noindent
The @command{gnatpp} tool is an ASIS-based utility
for source reformatting / pretty-printing.
It takes an Ada source file as input and generates a reformatted
version as output.
You can specify various style directives via switches; e.g.,
identifier case conventions, rules of indentation, and comment layout.

Note: A newly-redesigned set of formatting algorithms used by gnatpp
is now available.
To invoke the old formatting algorithms, use the @option{--pp-old} switch.
Support for @option{--pp-old} will be removed in some future version.

To produce a reformatted file, @command{gnatpp} invokes the Ada
compiler and generates and uses the ASIS tree for the input source;
thus the input must be legal Ada code, and the tool should have all the
information needed to compile the input source. To provide this information,
you may specify as a tool parameter the project file the input source belongs to
(or you may call @command{gnatpp}
through the @command{gnat} driver (see @ref{The GNAT Driver and
Project Files}). Another possibility is to specify the source search
path and needed configuration files in @option{-cargs} section of @command{gnatpp}
call, see the description of the @command{gnatpp} switches below.

@command{gnatpp} cannot process sources that contain
preprocessing directives.

The @command{gnatpp} command has the form

@smallexample
@c $ gnatpp @ovar{switches} @var{filename}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatpp @r{[}@var{switches}@r{]} @var{filename} @r{[}-cargs @var{gcc_switches}@r{]}
@end smallexample

@noindent
where
@itemize @bullet
@item
@var{switches} is an optional sequence of switches defining such properties as
the formatting rules, the source search path, and the destination for the
output source file

@item
@var{filename} is the name (including the extension) of the source file to
reformat; wildcards or several file names on the same gnatpp command are
allowed. The file name may contain path information; it does not have to
follow the GNAT file naming rules

@item
@samp{@var{gcc_switches}} is a list of switches for
@command{gcc}. They will be passed on to all compiler invocations made by
@command{gnatpp} to generate the ASIS trees. Here you can provide
@option{-I} switches to form the source search path,
use the @option{-gnatec} switch to set the configuration file, etc.
@end itemize

@node Switches for gnatpp
@section Switches for @command{gnatpp}

@noindent
The following subsections describe the various switches accepted by
@command{gnatpp}, organized by category.

You specify a switch by supplying a name and generally also a value.
In many cases the values for a switch with a given name are incompatible with
each other
(for example the switch that controls the casing of a reserved word may have
exactly one value: upper case, lower case, or
mixed case) and thus exactly one such switch can be in effect for an
invocation of @command{gnatpp}.
If more than one is supplied, the last one is used.
However, some values for the same switch are mutually compatible.
You may supply several such switches to @command{gnatpp}, but then
each must be specified in full, with both the name and the value.
Abbreviated forms (the name appearing once, followed by each value) are
not permitted.


@menu
* Alignment Control::
* Casing Control::
* General Text Layout Control::
* Other Formatting Options::
* Setting the Source Search Path::
* Output File Control::
* Other gnatpp Switches::
@end menu

@node Alignment Control
@subsection Alignment Control
@cindex Alignment control in @command{gnatpp}

@noindent
Programs can be easier to read if certain constructs are vertically aligned.
By default alignment of the following constructs is set ON:
@code{:} in declarations, @code{:=} in initializations in declarations
@code{:=} in assignment statements, @code{=>} in associations, and
@code{at} keywords in the component clauses in record
representation clauses.

@table @option
@cindex @option{-A@var{n}} (@command{gnatpp})

@item -A0
Set alignment to OFF

@item -A1
Set alignment to ON
@end table

@node Casing Control
@subsection Casing Control
@cindex Casing control in @command{gnatpp}

@noindent
@command{gnatpp} allows you to specify the casing for reserved words,
pragma names, attribute designators and identifiers.
For identifiers you may define a
general rule for name casing but also override this rule
via a set of dictionary files.

Three types of casing are supported: lower case, upper case, and mixed case.
``Mixed case'' means that the first letter, and also each letter immediately
following an underscore, are converted to their uppercase forms;
all the other letters are converted to their lowercase forms.

@table @option
@cindex @option{-a@var{x}} (@command{gnatpp})
@item -aL
Attribute designators are lower case

@item -aU
Attribute designators are upper case

@item -aM
Attribute designators are mixed case (this is the default)

@cindex @option{-k@var{x}} (@command{gnatpp})
@item -kL
Keywords (technically, these are known in Ada as @emph{reserved words}) are
lower case (this is the default)

@item -kU
Keywords are upper case

@cindex @option{-n@var{x}} (@command{gnatpp})
@item -nD
Name casing for defining occurrences are as they appear in the source file
(this is the default)

@item -nU
Names are in upper case

@item -nL
Names are in lower case

@item -nM
Names are in mixed case

@cindex @option{-ne@var{x}} (@command{gnatpp})
@item -neD
Enumeration literal casing for defining occurrences are as they appear in the
source file. Overrides -n casing setting.

@item -neU
Enumeration literals are in upper case.  Overrides -n casing
setting.

@item -neL
Enumeration literals are in lower case. Overrides -n casing
setting.

@item -neM
Enumeration literals are in mixed case. Overrides -n casing
setting.

@cindex @option{-nt@var{x}} (@command{gnatpp})
@item -neD
Names introduced by type and subtype declarations are always
cased as they appear in the declaration in the source file.
Overrides -n casing setting.

@item -ntU
Names introduced by type and subtype declarations are always in
upper case. Overrides -n casing setting.

@item -ntL
Names introduced by type and subtype declarations are always in
lower case. Overrides -n casing setting.

@item -ntM
Names introduced by type and subtype declarations are always in
mixed case. Overrides -n casing setting.

@item -nnU
Names introduced by number declarations are always in
upper case. Overrides -n casing setting.

@item -nnL
Names introduced by number declarations are always in
lower case. Overrides -n casing setting.

@item -nnM
Names introduced by number declarations are always in
mixed case. Overrides -n casing setting.

@cindex @option{-p@var{x}} (@command{gnatpp})
@item -pL
Pragma names are lower case

@item -pU
Pragma names are upper case

@item -pM
Pragma names are mixed case (this is the default)

@item -D@var{file}
@cindex @option{-D} (@command{gnatpp})
Use @var{file} as a @emph{dictionary file} that defines
the casing for a set of specified names,
thereby overriding the effect on these names by
any explicit or implicit
-n switch.
To supply more than one dictionary file,
use several @option{-D} switches.

@noindent
@option{gnatpp} implicitly uses a @emph{default dictionary file}
to define the casing for the Ada predefined names and
the names declared in the GNAT libraries.

@item -D-
@cindex @option{-D-} (@command{gnatpp})
Do not use the default dictionary file;
instead, use the casing
defined by a @option{-n} switch and any explicit
dictionary file(s)
@end table

@noindent
The structure of a dictionary file, and details on the conventions
used in the default dictionary file, are defined in @ref{Name Casing}.

The @option{-D-} and
@option{-D@var{file}} switches are mutually
compatible.

@noindent
This group of @command{gnatpp} switches controls the layout of comments and
complex syntactic constructs.  See @ref{Formatting Comments} for details
on their effect.

@table @option
@cindex @option{-c@var{n}} (@command{gnatpp})
@item -c0
All comments remain unchanged.

@item -c1
GNAT-style comment line indentation.
This is the default.

@item -c3
GNAT-style comment beginning.

@item -c4
Fill comment blocks.

@item -c5
Keep unchanged special form comments.
This is the default.

@item --comments-only
@cindex @option{--comments-only} @command{gnatpp}
Format just the comments.

@cindex @option{--no-separate-is} (@command{gnatpp})
@item --no-separate-is
Do not place the keyword @code{is} on a separate line in a subprogram body in
case if the spec occupies more than one line.

@cindex @option{--separate-loop-then} (@command{gnatpp})
@item --separate-loop-then
Place the keyword @code{loop} in FOR and WHILE loop statements and the
keyword @code{then} in IF statements on a separate line.

@cindex @option{--no-separate-loop-then} (@command{gnatpp})
@item --no-separate-loop-then
Do not place the keyword @code{loop} in FOR and WHILE loop statements and the
keyword @code{then} in IF statements on a separate line. This option is
incompatible with @option{--separate-loop-then} option.

@cindex @option{--use-on-new-line} (@command{gnatpp})
@item --use-on-new-line
Start each USE clause in a context clause from a separate line.

@cindex @option{--insert-blank-lines} (@command{gnatpp})
@item --insert-blank-lines
Insert blank lines where appropriate (between bodies and other large
constructs).

@cindex @option{--preserve-blank-lines} (@command{gnatpp})
@item --preserve-blank-lines
Preserve blank lines in the input. By default, gnatpp will squeeze
multiple blank lines down to one.

@end table

@noindent
The @option{-c} switches are compatible with one another, except that
the @option{-c0} switch disables all other comment formatting
switches.


@node General Text Layout Control
@subsection General Text Layout Control

@noindent
These switches allow control over line length and indentation.

@table @option
@item -M@var{nnn}
@cindex @option{-M} (@command{gnatpp})
Maximum line length, @var{nnn} from 32@dots{}256, the default value is 79

@item -i@var{nnn}
@cindex @option{-i} (@command{gnatpp})
Indentation level, @var{nnn} from 1@dots{}9, the default value is 3

@item -cl@var{nnn}
@cindex @option{-cl} (@command{gnatpp})
Indentation level for continuation lines (relative to the line being
continued), @var{nnn} from 1@dots{}9.
The default
value is one less than the (normal) indentation level, unless the
indentation is set to 1 (in which case the default value for continuation
line indentation is also 1)
@end table

@node Other Formatting Options
@subsection Other Formatting Options

@noindent
These switches control other formatting not listed above.

@table @option
@item --decimal-grouping=@var{n}
@cindex @option{--decimal-grouping} @command{gnatpp}
Put underscores in decimal literals (numeric literals without a base)
every @var{n} characters. If a literal already has one or more
underscores, it is not modified. For example, with
@code{--decimal-grouping=3}, @code{1000000} will be changed to
@code{1_000_000}.

@item --based-grouping=@var{n}
@cindex @option{--based-grouping} @command{gnatpp}
Same as @code{--decimal-grouping}, but for based literals. For
example, with @code{--based-grouping=4}, @code{16#0001FFFE#} will be
changed to @code{16#0001_FFFE#}.

@item --split-line-before-op
@cindex @option{--split-line-before-op} (@command{gnatpp})
If it is necessary to split a line at a binary operator, by default
the line is split after the operator. With this option, it is split
before the operator.

@item --RM-style-spacing
@cindex @option{--RM-style-spacing} (@command{gnatpp})
Do not insert an extra blank before various occurrences of
`(' and `:'. This also turns off alignment.

@item -ff
@cindex @option{-ff} (@command{gnatpp})
Insert a Form Feed character after a pragma Page.

@item --call_threshold=@var{nnn}
@cindex @option{--call_threshold} (@command{gnatpp})
If the number of parameter associations is greater than @var{nnn} and if at
least one association uses named notation, start each association from
a new line. If @var{nnn} is 0, no check for the number of associations
is made; this is the default.

@item --par_threshold=@var{nnn}
@cindex @option{--par_threshold} (@command{gnatpp})
If the number of parameter specifications is greater than @var{nnn}
(or equal to @var{nnn} in case of a function), start each specification from
a new line. This feature is disabled by default.
@end table

@node Setting the Source Search Path
@subsection Setting the Source Search Path

@noindent
To define the search path for the input source file, @command{gnatpp}
uses the same switches as the GNAT compiler, with the same effects:

@table @option
@item -I@var{dir}
@cindex @option{-I} (@command{gnatpp})

@item -I-
@cindex @option{-I-} (@command{gnatpp})

@item -gnatec=@var{path}
@cindex @option{-gnatec} (@command{gnatpp})

@end table

@node Output File Control
@subsection Output File Control

@noindent
By default the output is sent to a file whose name is obtained by appending
the @file{.pp} suffix to the name of the input file.
If the file with this name already exists, it is overwritten.
Thus if the input file is @file{my_ada_proc.adb} then
@command{gnatpp} will produce @file{my_ada_proc.adb.pp}
as output file.
The output may be redirected by the following switches:

@table @option
@item --output-dir=@var{dir}
@cindex @option{--output-dir} (@command{gnatpp})
Generate output file in directory @file{dir} with the same name as the input
file. If @file{dir} is the same as the directory containing the input file,
the input file is not processed; use @option{-rnb}
if you want to update the input file in place.

@item -pipe
@cindex @option{-pipe} (@command{gnatpp})
Send the output to @code{Standard_Output}

@item -o @var{output_file}
@cindex @option{-o} (@code{gnatpp})
Write the output into @var{output_file}.
If @var{output_file} already exists, @command{gnatpp} terminates without
reading or processing the input file.

@item -of @var{output_file}
@cindex @option{-of} (@command{gnatpp})
Write the output into @var{output_file}, overwriting the existing file
(if one is present).

@item -r
@cindex @option{-r} (@command{gnatpp})
Replace the input source file with the reformatted output, and copy the
original input source into the file whose name is obtained by appending the
@file{.npp} suffix to the name of the input file.
If a file with this name already exists, @command{gnatpp} terminates without
reading or processing the input file.

@item -rf
@cindex @option{-rf} (@code{gnatpp})
Like @option{-r} except that if the file with the specified name
already exists, it is overwritten.

@item -rnb
@cindex @option{-rnb} (@command{gnatpp})
Replace the input source file with the reformatted output without
creating any backup copy of the input source.

@item --eol=@var{xxx}
@cindex @option{--eol} (@code{gnatpp})
Specifies the line-ending style of the reformatted output file. The @var{xxx}
string specified with the switch may be:
@itemize @bullet
@item ``@option{dos}'' MS DOS style, lines end with CR LF characters
@item ``@option{crlf}''
the same as @option{dos}
@item ``@option{unix}'' UNIX style, lines end with LF character
@item ``@option{lf}''
the same as @option{unix}
@end itemize

@item -W@var{e}
@cindex @option{-W} (@command{gnatpp})
Specify the wide character encoding method for the input and output files.
@var{e} is one of the following:

@itemize @bullet

@item h
Hex encoding

@item u
Upper half encoding

@item s
Shift/JIS encoding

@item e
EUC encoding

@item 8
UTF-8 encoding

@item b
Brackets encoding (default value)
@end itemize

@end table

@noindent
Options @option{-o} and
@option{-of} are allowed only if the call to gnatpp
contains only one file to reformat.
Option
@option{--eol}
and
@option{-W}
cannot be used together
with @option{-pipe} option.

@node Other gnatpp Switches
@subsection Other @code{gnatpp} Switches

@noindent
The additional @command{gnatpp} switches are defined in this subsection.

@table @option
@item --version
@cindex @option{--version} @command{gnatpp}
Display copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatpp}
Display usage, then exit disregarding all other options.

@item -P @var{file}
@cindex @option{-P} @command{gnatpp}
Indicates the name of the project file that describes the set of sources
to be processed. The exact set of argument sources depends on other options
specified; see below.

@item -U
@cindex @option{-U} @command{gnatpp}
If a project file is specified and no argument source is explicitly
specified (either directly or by means of @option{-files} option), process
all the units of the closure of the argument project. Otherwise this option
has no effect.

@item -U @var{main_unit}
If a project file is specified and no argument source is explicitly
specified (either directly or by means of @option{-files} option), process
the closure of units rooted at @var{main_unit}. Otherwise this option
has no effect.

@item -X@var{name}=@var{value}
@cindex @option{-X} @command{gnatpp}
Indicates that external variable @var{name} in the argument project
has the value @var{value}. Has no effect if no project is specified as
tool argument.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gnatpp})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item --incremental
@cindex @option{--incremental} @command{gnatpp}
Incremental processing on a per-file basis. Source files are only
processed if they have been modified, or if files they depend on have
been modified. This is similar to the way gnatmake/gprbuild only
compiles files that need to be recompiled. A project file is required
in this mode, and the gnat driver (as in @command{gnat pretty}) is not
supported.

@item --pp-off=@var{xxx}
@cindex @option{--pp-off} @command{gnatpp}
Use @code{--xxx} as the command to turn off pretty printing, instead
of the default @code{--!pp off}.

@item --pp-on=@var{xxx}
@cindex @option{--pp-on} @command{gnatpp}
Use @code{--xxx} as the command to turn pretty printing back on, instead
of the default @code{--!pp on}.

@item --pp-old
@cindex @option{--pp-old} @command{gnatpp}
Use the old formatting algorithms.

@item -files @var{filename}
@cindex @option{-files} (@code{gnatpp})
Take the argument source files from the specified file. This file should be an
ordinary text file containing file names separated by spaces or
line breaks. You can use this switch more than once in the same call to
@command{gnatpp}. You also can combine this switch with an explicit list of
files.

@item -j@var{n}
@cindex @option{-j} (@command{gnatpp})
Without @option{--incremental}, use @var{n} processes to carry out the
tree creations (internal representations of the argument sources). On
a multiprocessor machine this speeds up processing of big sets of
argument sources. If @var{n} is 0, then the maximum number of parallel
tree creations is the number of core processors on the platform. This
option cannot be used together with @option{-r},
@option{-rf} or
@option{-rnb} option.

With @option{--incremental}, use @var{n} @command{gnatpp} processes to
perform pretty-printing in parallel. @var{n} = 0 means the same as
above. In this case, @option{-r},
@option{-rf} or
@option{-rnb} options are allowed.

@cindex @option{-t} (@command{gnatpp})
@item -t
Print out execution time.

@item -v
@cindex @option{-v} (@command{gnatpp})
Verbose mode

@item -q
@cindex @option{-q} (@command{gnatpp})
Quiet mode
@end table

@noindent
If a project file is specified and no argument source is explicitly
specified (either directly or by means of @option{-files} option), and no
@option{-U} is specified, then the set of processed sources is
all the immediate units of the argument project.


@node Formatting Rules
@section Formatting Rules

@noindent
The following subsections show how @command{gnatpp} treats white space,
comments, program layout, and name casing.
They provide detailed descriptions of the switches shown above.

@menu
* Disabling Pretty Printing::
* White Space and Empty Lines::
* Formatting Comments::
* Name Casing::
@end menu

@node Disabling Pretty Printing
@subsection Disabling Pretty Printing

@noindent
Pretty printing is highly heuristic in nature, and sometimes doesn't
do exactly what you want. If you wish to format a certain region of
code by hand, you can turn off pretty printing in that region by
surrounding it with special comments that start with @code{--!pp off}
and @code{--!pp on}. The text in that region will then be reproduced
verbatim in the output with no formatting.

To disable pretty printing for the whole file, put @code{--!pp off} at
the top, with no following @code{--!pp on}.

The comments must appear on a line by themselves, with nothing
preceding except spaces. The initial text of the comment must be
exactly @code{--!pp off} or @code{--!pp on} (case sensitive), but may
be followed by arbitrary additional text. For example:

@smallexample @c ada
@cartouche
@b{package} Interrupts @b{is}
   --@i{!pp off -- turn off pretty printing so "Interrupt_Kind" lines up}
   @b{type}            Interrupt_Kind @b{is}
     (Asynchronous_Interrupt_Kind,
       Synchronous_Interrupt_Kind,
             Green_Interrupt_Kind);
   --@i{!pp on -- reenable pretty printing}

   ...
@end cartouche
@end smallexample

You can specify different comment strings using the @code{--pp-off}
and @code{--pp-on} switches. For example, if you say @code{gnatpp
--pp-off=' pp-' *.ad?} then gnatpp will recognize comments of the form
@code{-- pp-} instead of @code{--!pp off} for disabling pretty
printing. Note that the leading @code{--} of the comment is not
included in the argument to these switches.

@node White Space and Empty Lines
@subsection White Space and Empty Lines

@noindent
@command{gnatpp} does not have an option to control space characters.
It will add or remove spaces according to the style illustrated by the
examples in the @cite{Ada Reference Manual}.
The output file will contain no lines with trailing white space.

By default, a sequence of one or more blank lines in the input is
converted to a single blank line in the output; multiple blank lines
are squeezed down to one.
The @option{--preserve-blank-lines} option
turns off the squeezing; each blank line in the input is copied
to the output.
The @option{--insert-blank-lines} option
causes additional blank lines to be inserted if not already
present in the input (e.g. between bodies).

@node Formatting Comments
@subsection Formatting Comments

@noindent
Comments in Ada code are of two kinds:
@itemize @bullet
@item
a @emph{whole-line comment}, which appears by itself (possibly preceded by
white space) on a line

@item
an @emph{end-of-line comment}, which follows some other Ada code on
the same line.
@end itemize

@noindent
A whole-line comment is indented according to the surrounding code,
with some exceptions.
Comments that start in column 1 are kept there.
If possible, comments are not moved so far to the right that the maximum
line length is exceeded.
The @option{-c0} option
turns off comment formatting.
Special-form comments such as SPARK-style @code{--#...} are left alone.

For an end-of-line comment, @command{gnatpp} tries to leave the same
number of spaces between the end of the preceding Ada code and the
beginning of the comment as appear in the original source.

@noindent
The @option{-c3} switch
(GNAT style comment beginning) has the following
effect:

@itemize @bullet
@item
For each whole-line comment that does not end with two hyphens,
@command{gnatpp} inserts spaces if necessary after the starting two hyphens
to ensure that there are at least two spaces between these hyphens and the
first non-blank character of the comment.
@end itemize

@noindent
The @option{-c4} switch specifies that
whole-line comments that form a paragraph will be filled in typical
word processor style (that is, moving words between lines to make the
lines other than the last similar in length ).

@noindent
The @option{--comments-only} switch specifies that only the comments
are formatted; the rest of the program text is left alone. The
comments are formatted according to the -c3 and -c4 switches; other
formatting switches are ignored. For example, @option{--comments-only
-c4} means to fill comment paragraphs, and do nothing else. Likewise,
@option{--comments-only -c3} ensures comments start with at least two
spaces after @code{--}, and @option{--comments-only -c3 -c4} does
both. If @option{--comments-only} is given without @option{-c3} or
@option{-c4}, then gnatpp doesn't format anything.

@node Name Casing
@subsection Name Casing

@noindent
@command{gnatpp} always converts the usage occurrence of a (simple) name to
the same casing as the corresponding defining identifier.

You control the casing for defining occurrences via the
@option{-n} switch.
With @option{-nD} (``as declared'', which is the default),
defining occurrences appear exactly as in the source file
where they are declared.
The other values for this switch ---
@option{-nU},
@option{-nL},
@option{-nM} ---
result in
upper, lower, or mixed case, respectively.
If @command{gnatpp} changes the casing of a defining
occurrence, it analogously changes the casing of all the
usage occurrences of this name.

If the defining occurrence of a name is not in the source compilation unit
currently being processed by @command{gnatpp}, the casing of each reference to
this name is changed according to the value of the @option{-n}
switch (subject to the dictionary file mechanism described below).
Thus @command{gnatpp} acts as though the @option{-n} switch
had affected the
casing for the defining occurrence of the name.

The options
@option{-a@var{x}},
@option{-k@var{x}},
@option{-ne@var{x}},
@option{-nt@var{x}},
@option{-nn@var{x}}, and
@option{-p@var{x}}
allow finer-grained control over casing for
attributes, keywords, enumeration literals,
types, named numbers and pragmas, respectively.
@option{-nt@var{x}} covers subtypes and
task and protected bodies as well.

Some names may need to be spelled with casing conventions that are not
covered by the upper-, lower-, and mixed-case transformations.
You can arrange correct casing by placing such names in a
@emph{dictionary file},
and then supplying a @option{-D} switch.
The casing of names from dictionary files overrides
any @option{-n} switch.

To handle the casing of Ada predefined names and the names from GNAT libraries,
@command{gnatpp} assumes a default dictionary file.
The name of each predefined entity is spelled with the same casing as is used
for the entity in the @cite{Ada Reference Manual} (usually mixed case).
The name of each entity in the GNAT libraries is spelled with the same casing
as is used in the declaration of that entity.

The @w{@option{-D-}} switch suppresses the use of
the default dictionary file. Instead, the casing for predefined and
GNAT-defined names will be established by the
@option{-n} switch or explicit dictionary files. For
example, by default the names @code{Ada.Text_IO} and
@code{GNAT.OS_Lib} will appear as just shown, even in the presence of
a @option{-nU} switch.  To ensure that even
such names are rendered in uppercase, additionally supply the
@w{@option{-D-}} switch (or else place these names
in upper case in a dictionary file).

A dictionary file is a plain text file; each line in this file can be
either a blank line (containing only space characters), an Ada comment
line, or the specification of exactly one @emph{casing schema}.

A casing schema is a string that has the following syntax:

@smallexample
@cartouche
  @var{casing_schema} ::= @var{identifier} | *@var{simple_identifier}*

  @var{simple_identifier} ::= @var{letter}@{@var{letter_or_digit}@}
@end cartouche
@end smallexample

@noindent
(See @cite{Ada Reference Manual}, Section 2.3) for the definition of the
@var{identifier} lexical element and the @var{letter_or_digit} category.)

The casing schema string can be followed by white space and/or an Ada-style
comment; any amount of white space is allowed before the string.

If a dictionary file is passed as
the value of a @option{-D@var{file}} switch
then for every
simple name and every identifier, @command{gnatpp} checks if the dictionary
defines the casing for the name or for some of its parts (the term ``subword''
is used below to denote the part of a name which is delimited by ``_'' or by
the beginning or end of the word and which does not contain any ``_'' inside):

@itemize @bullet
@item
if the whole name is in the dictionary, @command{gnatpp} uses for this name
the casing defined by the dictionary; no subwords are checked for this word

@item
for every subword @command{gnatpp} checks if the dictionary contains the
corresponding string of the form @code{*@var{simple_identifier}*},
and if it does, the casing of this @var{simple_identifier} is used
for this subword

@item
if the whole name does not contain any ``_'' inside, and if for this name
the dictionary contains two entries - one of the form @var{identifier},
and another - of the form *@var{simple_identifier}*, then the first one
is applied to define the casing of this name

@item
if more than one dictionary file is passed as @command{gnatpp} switches, each
dictionary adds new casing exceptions and overrides all the existing casing
exceptions set by the previous dictionaries

@item
when @command{gnatpp} checks if the word or subword is in the dictionary,
this check is not case sensitive
@end itemize

@noindent
For example, suppose we have the following source to reformat:

@smallexample @c ada
@cartouche
@b{procedure} test @b{is}
   name1 : integer := 1;
   name4_name3_name2 : integer := 2;
   name2_name3_name4 : Boolean;
   name1_var : Float;
@b{begin}
   name2_name3_name4 := name4_name3_name2 > name1;
@b{end};
@end cartouche
@end smallexample

@noindent
And suppose we have two dictionaries:

@smallexample
@cartouche
@i{dict1:}
   NAME1
   *NaMe3*
   *Name1*
@end cartouche

@cartouche
@i{dict2:}
  *NAME3*
@end cartouche
@end smallexample

@noindent
If @command{gnatpp} is called with the following switches:

@smallexample
@command{gnatpp -nM -D dict1 -D dict2 test.adb}
@end smallexample

@noindent
then we will get the following name casing in the @command{gnatpp} output:

@smallexample @c ada
@cartouche
@b{procedure} Test @b{is}
   NAME1             : Integer := 1;
   Name4_NAME3_Name2 : Integer := 2;
   Name2_NAME3_Name4 : Boolean;
   Name1_Var         : Float;
@b{begin}
   Name2_NAME3_Name4 := Name4_NAME3_Name2 > NAME1;
@b{end} Test;
@end cartouche
@end smallexample
@end ifclear

@ifclear FSFEDITION
@c *********************************
@node The Ada-to-XML converter gnat2xml
@chapter The Ada-to-XML converter @command{gnat2xml}
@findex gnat2xml
@cindex XML generation

@noindent
The @command{gnat2xml} tool is an ASIS-based utility that converts
Ada source code into XML.

@menu
* Switches for gnat2xml::
* Other Programs::
* Structure of the XML::
* Generating Representation Clauses::
@end menu

@node Switches for gnat2xml
@section Switches for @command{gnat2xml}

@noindent
@command{gnat2xml} takes Ada source code as input, and produces XML
that conforms to the schema.

Usage:

@smallexample
gnat2xml [options] filenames [-files filename] [-cargs gcc_switches]
@end smallexample

@noindent
options:
@smallexample
-h
--help -- generate usage information and quit, ignoring all other options
--version -- print version and quit, ignoring all other options

-P @file{file} -- indicates the name of the project file that describes
      the set of sources to be processed. The exact set of argument
      sources depends on other options specified, see below.

-U -- if a project file is specified and no argument source is explicitly
      specified, process all the units of the closure of the argument project.
      Otherwise this option has no effect.

-U @var{main_unit} -- if a project file is specified and no argument source
      is explicitly specified (either directly or by means of @option{-files}
      option), process the closure of units rooted at @var{main_unit}.
      Otherwise this option has no effect.

-X@var{name}=@var{value} -- indicates that external variable @var{name} in
      the argument project has the value @var{value}. Has no effect if no
      project is specified as tool argument.

--RTS=@var{rts-path}  -- Specifies the default location of the runtime
      library. Same meaning as the equivalent @command{gnatmake} flag
      (@pxref{Switches for gnatmake}).

--incremental -- incremental processing on a per-file basis. Source files are
      only processed if they have been modified, or if files they depend
      on have been modified. This is similar to the way gnatmake/gprbuild
      only compiles files that need to be recompiled. A project file
      is required in this mode.

-j@var{n} -- In @option{--incremental} mode, use @var{n} @command{gnat2xml}
      processes to perform XML generation in parallel. If @var{n} is 0, then
      the maximum number of parallel tree creations is the number of core
      processors on the platform.

--output-dir=@var{dir} -- generate one .xml file for each Ada source file, in
      directory @file{dir}. (Default is to generate the XML to standard
      output.)

-I <include-dir>
      directories to search for dependencies
      You can also set the ADA_INCLUDE_PATH environment variable for this.

--compact -- debugging version, with interspersed source, and a more
      compact representation of "sloc". This version does not conform
      to any schema.

--rep-clauses -- generate representation clauses (see ``Generating
      Representation Clauses'' below).

-files=filename - the name of a text file containing a list
                  of Ada source files to process

-q -- quiet
-v -- verbose

-cargs ... -- options to pass to gcc
@end smallexample

@noindent
If a project file is specified and no argument source is explicitly
specified, and no @option{-U} is specified, then the set of processed
sources is all the immediate units of the argument project.

Example:

@smallexample
gnat2xml -v -output-dir=xml-files *.ad[sb]
@end smallexample

@noindent
The above will create *.xml files in the @file{xml-files} subdirectory.
For example, if there is an Ada package Mumble.Dumble, whose spec and
body source code lives in mumble-dumble.ads and mumble-dumble.adb,
the above will produce xml-files/mumble-dumble.ads.xml and
xml-files/mumble-dumble.adb.xml.

@node Other Programs
@section Other Programs

@noindent
The distribution includes two other programs that are related to
@command{gnat2xml}:

@command{gnat2xsd} is the schema generator, which generates the schema
to standard output, based on the structure of Ada as encoded by
ASIS. You don't need to run @command{gnat2xsd} in order to use
@command{gnat2xml}. To generate the schema, type:

@smallexample
gnat2xsd > ada-schema.xsd
@end smallexample

@noindent
@command{gnat2xml} generates XML files that will validate against
@file{ada-schema.xsd}.

@command{xml2gnat} is a back-translator that translates the XML back
into Ada source code. The Ada generated by @command{xml2gnat} has
identical semantics to the original Ada code passed to
@command{gnat2xml}. It is not textually identical, however --- for
example, no attempt is made to preserve the original indentation.

@node Structure of the XML
@section Structure of the XML

@noindent
The primary documentation for the structure of the XML generated by
@command{gnat2xml} is the schema (see @command{gnat2xsd} above). The
following documentation gives additional details needed to understand
the schema and therefore the XML.

The elements listed under Defining Occurrences, Usage Occurrences, and
Other Elements represent the syntactic structure of the Ada program.
Element names are given in lower case, with the corresponding element
type Capitalized_Like_This. The element and element type names are
derived directly from the ASIS enumeration type Flat_Element_Kinds,
declared in Asis.Extensions.Flat_Kinds, with the leading ``An_'' or ``A_''
removed. For example, the ASIS enumeration literal
An_Assignment_Statement corresponds to the XML element
assignment_statement of XML type Assignment_Statement.

To understand the details of the schema and the corresponding XML, it is
necessary to understand the ASIS standard, as well as the GNAT-specific
extension to ASIS.

A defining occurrence is an identifier (or character literal or operator
symbol) declared by a declaration. A usage occurrence is an identifier
(or ...) that references such a declared entity. For example, in:

@smallexample
type T is range 1..10;
X, Y : constant T := 1;
@end smallexample

@noindent
The first ``T'' is the defining occurrence of a type. The ``X'' is the
defining occurrence of a constant, as is the ``Y'', and the second ``T'' is
a usage occurrence referring to the defining occurrence of T.

Each element has a 'sloc' (source location), and subelements for each
syntactic subtree, reflecting the Ada grammar as implemented by ASIS.
The types of subelements are as defined in the ASIS standard. For
example, for the right-hand side of an assignment_statement we have
the following comment in asis-statements.ads:

@smallexample
------------------------------------------------------------------------------
--  18.3  function Assignment_Expression
------------------------------------------------------------------------------

   function Assignment_Expression
     (Statement : Asis.Statement)
      return      Asis.Expression;

------------------------------------------------------------------------------
...
--  Returns the expression from the right hand side of the assignment.
...
--  Returns Element_Kinds:
--       An_Expression
@end smallexample

@noindent
The corresponding sub-element of type Assignment_Statement is:

@smallexample
<xsd:element name="assignment_expression_q" type="Expression_Class"/>
@end smallexample

@noindent
where Expression_Class is defined by an xsd:choice of all the
various kinds of expression.

The 'sloc' of each element indicates the starting and ending line and
column numbers. Column numbers are character counts; that is, a tab
counts as 1, not as however many spaces it might expand to.

Subelements of type Element have names ending in ``_q'' (for ASIS
``Query''), and those of type Element_List end in ``_ql'' (``Query returning
List'').

Some subelements are ``Boolean''. For example, Private_Type_Definition
has has_abstract_q and has_limited_q, to indicate whether those
keywords are present, as in @code{type T is abstract limited
private;}. False is represented by a Nil_Element. True is represented
by an element type specific to that query (for example, Abstract and
Limited).

The root of the tree is a Compilation_Unit, with attributes:

@itemize @bullet
@item
unit_kind, unit_class, and unit_origin. These are strings that match the
enumeration literals of types Unit_Kinds, Unit_Classes, and Unit_Origins
in package Asis.

@item
unit_full_name is the full expanded name of the unit, starting from a
root library unit. So for @code{package P.Q.R is ...},
@code{unit_full_name="P.Q.R"}. Same for @code{separate (P.Q) package R is ...}.

@item
def_name is the same as unit_full_name for library units; for subunits,
it is just the simple name.

@item
source_file is the name of the Ada source file. For example, for
the spec of @code{P.Q.R}, @code{source_file="p-q-r.ads"}. This allows one to
interpret the source locations --- the ``sloc'' of all elements
within this Compilation_Unit refers to line and column numbers
within the named file.
@end itemize

@noindent
Defining occurrences have these attributes:

@itemize @bullet
@item
def_name is the simple name of the declared entity, as written in the Ada
source code.

@item
def is a unique URI of the form:

 ada://kind/fully/qualified/name

where:

 kind indicates the kind of Ada entity being declared (see below), and

 fully/qualified/name, is the fully qualified name of the Ada
 entity, with each of ``fully'', ``qualified'', and ``name'' being
 mangled for uniqueness. We do not document the mangling
 algorithm, which is subject to change; we just guarantee that the
 names are unique in the face of overloading.

@item
type is the type of the declared object, or @code{null} for
declarations of things other than objects.
@end itemize

@noindent
Usage occurrences have these attributes:

@itemize @bullet
@item
ref_name is the same as the def_name of the corresponding defining
occurrence. This attribute is not of much use, because of
overloading; use ref for lookups, instead.

@item
ref is the same as the def of the corresponding defining
occurrence.
@end itemize

@noindent
In summary, @code{def_name} and @code{ref_name} are as in the source
code of the declaration, possibly overloaded, whereas @code{def} and
@code{ref} are unique-ified.

Literal elements have this attribute:

@itemize @bullet
@item
lit_val is the value of the literal as written in the source text,
appropriately escaped (e.g. @code{"} ---> @code{&quot;}). This applies
only to numeric and string literals. Enumeration literals in Ada are
not really "literals" in the usual sense; they are usage occurrences,
and have ref_name and ref as described above. Note also that string
literals used as operator symbols are treated as defining or usage
occurrences, not as literals.
@end itemize

@noindent
Elements that can syntactically represent names and expressions (which
includes usage occurrences, plus function calls and so forth) have this
attribute:

@itemize @bullet
@item
type. If the element represents an expression or the name of an object,
'type' is the 'def' for the defining occurrence of the type of that
expression or name. Names of other kinds of entities, such as package
names and type names, do not have a type in Ada; these have type="null"
in the XML.
@end itemize

@noindent
Pragma elements have this attribute:

@itemize @bullet
@item
pragma_name is the name of the pragma. For language-defined pragmas, the
pragma name is redundant with the element kind (for example, an
assert_pragma element necessarily has pragma_name="Assert"). However, all
implementation-defined pragmas are lumped together in ASIS as a single
element kind (for example, the GNAT-specific pragma Unreferenced is
represented by an implementation_defined_pragma element with
pragma_name="Unreferenced").
@end itemize

@noindent
Defining occurrences of formal parameters and generic formal objects have this
attribute:

@itemize @bullet
@item
mode indicates that the parameter is of mode 'in', 'in out', or 'out'.
@end itemize

@noindent
All elements other than Not_An_Element have this attribute:

@itemize @bullet
@item
checks is a comma-separated list of run-time checks that are needed
for that element. The possible checks are: do_accessibility_check,
do_discriminant_check,do_division_check,do_length_check,
do_overflow_check,do_range_check,do_storage_check,do_tag_check.
@end itemize

@noindent
The "kind" part of the "def" and "ref" attributes is taken from the ASIS
enumeration type Flat_Declaration_Kinds, declared in
Asis.Extensions.Flat_Kinds, with the leading "An_" or "A_" removed, and
any trailing "_Declaration" or "_Specification" removed. Thus, the
possible kinds are as follows:

@smallexample
ordinary_type
task_type
protected_type
incomplete_type
tagged_incomplete_type
private_type
private_extension
subtype
variable
constant
deferred_constant
single_task
single_protected
integer_number
real_number
enumeration_literal
discriminant
component
loop_parameter
generalized_iterator
element_iterator
procedure
function
parameter
procedure_body
function_body
return_variable
return_constant
null_procedure
expression_function
package
package_body
object_renaming
exception_renaming
package_renaming
procedure_renaming
function_renaming
generic_package_renaming
generic_procedure_renaming
generic_function_renaming
task_body
protected_body
entry
entry_body
entry_index
procedure_body_stub
function_body_stub
package_body_stub
task_body_stub
protected_body_stub
exception
choice_parameter
generic_procedure
generic_function
generic_package
package_instantiation
procedure_instantiation
function_instantiation
formal_object
formal_type
formal_incomplete_type
formal_procedure
formal_function
formal_package
formal_package_declaration_with_box
@end smallexample

@node Generating Representation Clauses
@section Generating Representation Clauses

@noindent
If the @option{--rep-clauses} switch is given, @command{gnat2xml} will
generate representation clauses for certain types showing the
representation chosen by the compiler. The information is produced by
the ASIS ``Data Decomposition'' facility --- see the
@code{Asis.Data_Decomposition} package for details.

Not all types are supported. For example, @code{Type_Model_Kind} must
be @code{A_Simple_Static_Model}. Types declared within generic units
have no representation. The clauses that are generated include
@code{attribute_definition_clauses} for @code{Size} and
@code{Component_Size}, as well as
@code{record_representation_clauses}.

There is no guarantee that the generated representation clauses could
have actually come from legal Ada code; Ada has some restrictions that
are not necessarily obeyed by the generated clauses.

The representation clauses are surrounded by comment elements to
indicate that they are automatically generated, something like this:

@smallexample
<comment text="--gen+">
...
<attribute_definition_clause>
...
<comment text="--gen-">
...
@end smallexample

@end ifclear

@ifclear FSFEDITION
@c *********************************
@node The GNAT Metrics Tool gnatmetric
@chapter The GNAT Metrics Tool @command{gnatmetric}
@findex gnatmetric
@cindex Metric tool

@noindent
The @command{gnatmetric} tool is an ASIS-based utility
for computing various program metrics.
It takes an Ada source file as input and generates a file containing the
metrics data as output. Various switches control which
metrics are computed and output.

@menu
* Switches for gnatmetric::
@end menu

To compute program metrics, @command{gnatmetric} invokes the Ada
compiler and generates and uses the ASIS tree for the input source;
thus the input must be legal Ada code, and the tool should have all the
information needed to compile the input source. To provide this information,
you may specify as a tool parameter the project file the input source belongs to
(or you may call @command{gnatmetric}
through the @command{gnat} driver (see @ref{The GNAT Driver and
Project Files}). Another possibility is to specify the source search
path and needed configuration files in @option{-cargs} section of @command{gnatmetric}
call, see the description of the @command{gnatmetric} switches below.

If the set of sources to be processed by @code{gnatmetric} contains sources with
preprocessing directives
then the needed options should be provided to run preprocessor as a part of
the @command{gnatmetric} call, and the computed metrics
will correspond to preprocessed sources.


The @command{gnatmetric} command has the form

@smallexample
@c $ gnatmetric @ovar{switches} @{@var{filename}@} @r{[}-cargs @var{gcc_switches}@r{]}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatmetric @r{[}@var{switches}@r{]} @{@var{filename}@} @r{[}-cargs @var{gcc_switches}@r{]}
@end smallexample

@noindent
where
@itemize @bullet
@item
@var{switches} specify the metrics to compute and define the destination for
the output

@item
Each @var{filename} is the name (including the extension) of a source
file to process. ``Wildcards'' are allowed, and
the file name may contain path information.
If no @var{filename} is supplied, then the @var{switches} list must contain
at least one
@option{-files} switch (@pxref{Other gnatmetric Switches}).
Including both a @option{-files} switch and one or more
@var{filename} arguments is permitted.

@item
@samp{@var{gcc_switches}} is a list of switches for
@command{gcc}. They will be passed on to all compiler invocations made by
@command{gnatmetric} to generate the ASIS trees. Here you can provide
@option{-I} switches to form the source search path,
and use the @option{-gnatec} switch to set the configuration file,
use the @option{-gnat05} switch if sources should be compiled in
Ada 2005 mode etc.
@end itemize

@node Switches for gnatmetric
@section Switches for @command{gnatmetric}

@noindent
The following subsections describe the various switches accepted by
@command{gnatmetric}, organized by category.

@menu
* Output Files Control::
* Disable Metrics For Local Units::
* Specifying a set of metrics to compute::
* Other gnatmetric Switches::
@ignore
* Generate project-wide metrics::
@end ignore
@end menu

@node Output Files Control
@subsection Output File Control
@cindex Output file control in @command{gnatmetric}

@noindent
@command{gnatmetric} has two output formats. It can generate a
textual (human-readable) form, and also XML. By default only textual
output is generated.

When generating the output in textual form, @command{gnatmetric} creates
for each Ada source file a corresponding text file
containing the computed metrics, except for the case when the set of metrics
specified by gnatmetric parameters consists only of metrics that are computed
for the whole set of analyzed sources, but not for each Ada source.
By default, the name of the file containing metric information for a source
is obtained by appending the @file{.metrix} suffix to the
name of the input source file. If not otherwise specified and no project file
is specified as @command{gnatmetric} option this file is placed in the same
directory as where the source file is located. If @command{gnatmetric} has a
project  file as its parameter, it places all the generated files in the
object directory of the project (or in the project source directory if the
project does not define an objects directory), if @option{--subdirs} option
is specified, the files are placed in the subrirectory of this directory
specified by this option.

All the output information generated in XML format is placed in a single
file. By default the name of this file is @file{metrix.xml}.
If not otherwise specified and if no project file is specified
as @command{gnatmetric} option  this file is placed in the
current directory.

Some of the computed metrics are summed over the units passed to
@command{gnatmetric}; for example, the total number of lines of code.
By default this information is sent to @file{stdout}, but a file
can be specified with the @option{-og} switch.

The following switches control the @command{gnatmetric} output:

@table @option
@cindex @option{-x} (@command{gnatmetric})
@item -x
Generate the XML output

@cindex @option{-xs} (@command{gnatmetric})
@item -xs
Generate the XML output and the XML schema file that describes the structure
of the XML metric report, this schema is assigned to the XML file. The schema
file has the same name as the XML output file with @file{.xml} suffix replaced
with @file{.xsd}

@cindex @option{-nt} (@command{gnatmetric})
@item -nt
Do not generate the output in text form (implies @option{-x})

@cindex @option{-d} (@command{gnatmetric})
@item -d @var{output_dir}
Put text files with detailed metrics into @var{output_dir}

@cindex @option{-o} (@command{gnatmetric})
@item -o @var{file_suffix}
Use @var{file_suffix}, instead of @file{.metrix}
in the name of the output file.

@cindex @option{-og} (@command{gnatmetric})
@item -og @var{file_name}
Put global metrics into @var{file_name}

@cindex @option{-ox} (@command{gnatmetric})
@item -ox @var{file_name}
Put the XML output into @var{file_name} (also implies @option{-x})

@cindex @option{-sfn} (@command{gnatmetric})
@item -sfn
Use ``short'' source file names in the output.  (The @command{gnatmetric}
output includes the name(s) of the Ada source file(s) from which the metrics
are computed.  By default each name includes the absolute path. The
@option{-sfn} switch causes @command{gnatmetric}
to exclude all directory information from the file names that are output.)

@end table

@node Disable Metrics For Local Units
@subsection Disable Metrics For Local Units
@cindex Disable Metrics For Local Units in @command{gnatmetric}

@noindent
@command{gnatmetric} relies on the GNAT compilation model @minus{}
one compilation
unit per one source file. It computes line metrics for the whole source
file, and it also computes syntax
and complexity metrics for the file's outermost unit.

By default, @command{gnatmetric} will also compute all metrics for certain
kinds of locally declared program units:

@itemize @bullet
@item
subprogram (and generic subprogram) bodies;

@item
package (and generic package) specs and bodies;

@item
task object and type specifications and bodies;

@item
protected object and type specifications and bodies.
@end itemize

@noindent
These kinds of entities will be referred to as
@emph{eligible local program units}, or simply @emph{eligible local units},
@cindex Eligible local unit (for @command{gnatmetric})
in the discussion below.

Note that a subprogram declaration, generic instantiation,
or renaming declaration only receives metrics
computation when it appear as the outermost entity
in a source file.

Suppression of metrics computation for eligible local units can be
obtained via the following switch:

@table @option
@cindex @option{-nolocal} (@command{gnatmetric})
@item -nolocal
Do not compute detailed metrics for eligible local program units

@end table

@node Specifying a set of metrics to compute
@subsection Specifying a set of metrics to compute

@noindent
By default all the metrics are computed and reported. The switches
described in this subsection allow you to control, on an individual
basis, whether metrics are computed and
reported. If at least one positive metric
switch is specified (that is, a switch that defines that a given
metric or set of metrics is to be computed), then only
explicitly specified metrics are reported.

@menu
* Line Metrics Control::
* Syntax Metrics Control::
* Complexity Metrics Control::
* Coupling Metrics Control::
@end menu

@node Line Metrics Control
@subsubsection Line Metrics Control
@cindex Line metrics control in @command{gnatmetric}

@noindent
For any (legal) source file, and for each of its
eligible local program units, @command{gnatmetric} computes the following
metrics:

@itemize @bullet
@item
the total number of lines;

@item
the total number of code lines (i.e., non-blank lines that are not comments)

@item
the number of comment lines

@item
the number of code lines containing end-of-line comments;

@item
the comment percentage: the ratio between the number of lines that contain
comments and the number of all non-blank lines, expressed as a percentage;

@item
the number of empty lines and lines containing only space characters and/or
format effectors (blank lines)

@item
the average number of code lines in subprogram bodies, task bodies, entry
bodies and statement sequences in package bodies (this metric is only computed
across the whole set of the analyzed units)

@end itemize

@noindent
@command{gnatmetric} sums the values of the line metrics for all the
files being processed and then generates the cumulative results. The tool
also computes for all the files being processed the average number of code
lines in bodies.

You can use the following switches to select the specific line metrics
to be computed and reported.

@table @option
@cindex @option{--lines@var{x}} (@command{gnatmetric})

@cindex @option{--no-lines@var{x}}

@item --lines-all
Report all the line metrics

@item --no-lines-all
Do not report any of line metrics

@item --lines
Report the number of all lines

@item --no-lines
Do not report the number of all lines

@item --lines-code
Report the number of code lines

@item --no-lines-code
Do not report the number of code lines

@item --lines-comment
Report the number of comment lines

@item --no-lines-comment
Do not report the number of comment lines

@item --lines-eol-comment
Report the number of code lines containing
end-of-line comments

@item --no-lines-eol-comment
Do not report the number of code lines containing
end-of-line comments

@item --lines-ratio
Report the comment percentage in the program text

@item --no-lines-ratio
Do not report the comment percentage in the program text

@item --lines-blank
Report the number of blank lines

@item --no-lines-blank
Do not report the number of blank lines

@item --lines-average
Report the average number of code lines in subprogram bodies, task bodies,
entry bodies and statement sequences in package bodies. The metric is computed
and reported for the whole set of processed Ada sources only.

@item --no-lines-average
Do not report the average number of code lines in subprogram bodies,
task bodies, entry bodies and statement sequences in package bodies.

@end table

@node Syntax Metrics Control
@subsubsection Syntax Metrics Control
@cindex Syntax metrics control in @command{gnatmetric}

@noindent
@command{gnatmetric} computes various syntactic metrics for the
outermost unit and for each eligible local unit:

@table @emph
@item LSLOC (``Logical Source Lines Of Code'')
The total number of declarations and the total number of statements. Note
that the definition of declarations is the one given in the reference
manual:

@noindent
``Each of the following is defined to be a declaration: any basic_declaration;
an enumeration_literal_specification; a discriminant_specification;
a component_declaration; a loop_parameter_specification; a
parameter_specification; a subprogram_body; an entry_declaration;
an entry_index_specification; a choice_parameter_specification;
a generic_formal_parameter_declaration.''

This means for example that each enumeration literal adds one to the count,
as well as each subprogram parameter.

Thus the results from this metric will be significantly greater than might
be expected from a naive view of counting semicolons.

@item Maximal static nesting level of inner program units
According to
@cite{Ada Reference Manual}, 10.1(1), ``A program unit is either a
package, a task unit, a protected unit, a
protected entry, a generic unit, or an explicitly declared subprogram other
than an enumeration literal.''

@item Maximal nesting level of composite syntactic constructs
This corresponds to the notion of the
maximum nesting level in the GNAT built-in style checks
(@pxref{Style Checking})
@end table

@noindent
For the outermost unit in the file, @command{gnatmetric} additionally computes
the following metrics:

@table @emph
@item Public subprograms
This metric is computed for package specs. It is the
number of subprograms and generic subprograms declared in the visible
part (including the visible part of nested packages, protected objects, and
protected types).

@item All subprograms
This metric is computed for bodies and subunits. The
metric is equal to a total number of subprogram bodies in the compilation
unit.
Neither generic instantiations nor renamings-as-a-body nor body stubs
are counted. Any subprogram body is counted, independently of its nesting
level and enclosing constructs. Generic bodies and bodies of protected
subprograms are counted in the same way as ``usual'' subprogram bodies.

@item Public types
This metric is computed for package specs and
generic package declarations. It is the total number of types
that can be referenced from outside this compilation unit, plus the
number of types from all the visible parts of all the visible generic
packages. Generic formal types are not counted.  Only types, not subtypes,
are included.

@noindent
Along with the total number of public types, the following
types are counted and reported separately:

@itemize @bullet
@item
Abstract types

@item
Root tagged types (abstract, non-abstract, private, non-private). Type
extensions are @emph{not} counted

@item
Private types (including private extensions)

@item
Task types

@item
Protected types

@end itemize

@item All types
This metric is computed for any compilation unit. It is equal to the total
number of the declarations of different types given in the compilation unit.
The private and the corresponding full type declaration are counted as one
type declaration. Incomplete type declarations and generic formal types
are not counted.
No distinction is made among different kinds of types (abstract,
private etc.); the total number of types is computed and reported.

@end table

@noindent
By default, all the syntax metrics are computed and reported. You can use the
following switches to select specific syntax metrics.

@table @option

@cindex @option{--syntax@var{x}} (@command{gnatmetric})

@cindex @option{--no-syntax@var{x}} (@command{gnatmetric})

@item --syntax-all
Report all the syntax metrics

@item --no-syntax-all
Do not report any of syntax metrics

@item --declarations
Report the total number of declarations

@item --no-declarations
Do not report the total number of declarations

@item --statements
Report the total number of statements

@item --no-statements
Do not report the total number of statements

@item --public-subprograms
Report the number of public subprograms in a compilation unit

@item --no-public-subprograms
Do not report the number of public subprograms in a compilation unit

@item --all-subprograms
Report the number of all the subprograms in a compilation unit

@item --no-all-subprograms
Do not report the number of all the subprograms in a compilation unit

@item --public-types
Report the number of public types in a compilation unit

@item --no-public-types
Do not report the number of public types in a compilation unit

@item --all-types
Report the number of all the types in a compilation unit

@item --no-all-types
Do not report the number of all the types in a compilation unit

@item --unit-nesting
Report the maximal program unit nesting level

@item --no-unit-nesting
Do not report the maximal program unit nesting level

@item --construct-nesting
Report the maximal construct nesting level

@item --no-construct-nesting
Do not report the maximal construct nesting level

@end table

@node Complexity Metrics Control
@subsubsection Complexity Metrics Control
@cindex Complexity metrics control in @command{gnatmetric}

@noindent
For a program unit that is an executable body (a subprogram body (including
generic bodies), task body, entry body or a package body containing
its own statement sequence) @command{gnatmetric} computes the following
complexity metrics:

@itemize @bullet
@item
McCabe cyclomatic complexity;

@item
McCabe essential complexity;

@item
maximal loop nesting level;

@item
extra exit points (for subprograms);
@end itemize

@noindent
The McCabe cyclomatic complexity metric is defined
in @url{http://www.mccabe.com/pdf/mccabe-nist235r.pdf}

According to McCabe, both control statements and short-circuit control forms
should be taken into account when computing cyclomatic complexity.
For Ada 2012 we have also take into account conditional expressions
and quantified expressions. For each body, we compute three metric values:

@itemize @bullet
@item
the complexity introduced by control
statements only, without taking into account short-circuit forms
(referred as @code{statement complexity} in @command{gnatmetric} output),

@item
the complexity introduced by short-circuit control forms only
(referred as @code{expression complexity} in @command{gnatmetric} output), and

@item
the total
cyclomatic complexity, which is the sum of these two values
(referred as @code{cyclomatic complexity} in @command{gnatmetric} output).
@end itemize

@noindent

The cyclomatic complexity is also computed for Ada 2012 expression functions.
An expression function cannot have statements as its components, so only one
metric value is computed as a cyclomatic complexity of an expression function.

The origin of cyclomatic complexity metric is the need to estimate the number
of independent paths in the control flow graph that in turn gives the number
of tests needed to satisfy paths coverage testing completeness criterion.
Considered from the testing point of view, a static Ada @code{loop} (that is,
the @code{loop} statement having static subtype in loop parameter
specification) does not add to cyclomatic complexity. By providing
@option{--no-static-loop} option a user
may specify that such loops should not be counted when computing the
cyclomatic complexity metric

The Ada essential complexity metric is a McCabe cyclomatic complexity metric
counted for the code that is reduced by excluding all the pure structural Ada
control statements. An compound statement is considered as a non-structural
if it contains a @code{raise} or @code{return} statement as it subcomponent,
or if it contains a @code{goto} statement that transfers the control outside
the operator. A selective accept statement with @code{terminate} alternative
is considered as non-structural statement. When computing this metric,
@code{exit} statements are treated in the same way as @code{goto}
statements unless @option{-ne} option is specified.

The Ada essential complexity metric defined here is intended to quantify
the extent to which the software is unstructured. It is adapted from
the McCabe essential complexity metric defined in
@url{http://www.mccabe.com/pdf/mccabe-nist235r.pdf} but is modified to be more
suitable for typical Ada usage. For example, short circuit forms
are not penalized as unstructured in the Ada essential complexity metric.

When computing cyclomatic and essential complexity, @command{gnatmetric} skips
the code in the exception handlers and in all the nested program units. The
code of assertions and predicates (that is, subprogram preconditions and
postconditions, subtype predicates and type invariants) is also skipped.

By default, all the complexity metrics are computed and reported.
For more fine-grained control you can use
the following switches:

@table @option
@cindex @option{-complexity@var{x}} (@command{gnatmetric})

@cindex @option{--no-complexity@var{x}}

@item --complexity-all
Report all the complexity metrics

@item --no-complexity-all
Do not report any of complexity metrics

@item --complexity-cyclomatic
Report the McCabe Cyclomatic Complexity

@item --no-complexity-cyclomatic
Do not report the McCabe Cyclomatic Complexity

@item --complexity-essential
Report the Essential Complexity

@item --no-complexity-essential
Do not report the Essential Complexity

@item --loop-nesting
Report maximal loop nesting level

@item --no-loop-nesting
Do not report maximal loop nesting level

@item --complexity-average
Report the average McCabe Cyclomatic Complexity for all the subprogram bodies,
task bodies, entry bodies and statement sequences in package bodies.
The metric is computed and reported for whole set of processed Ada sources
only.

@item --no-complexity-average
Do not report the average McCabe Cyclomatic Complexity for all the subprogram
bodies, task bodies, entry bodies and statement sequences in package bodies

@cindex @option{-ne} (@command{gnatmetric})
@item -ne
Do not consider @code{exit} statements as @code{goto}s when
computing Essential Complexity

@cindex @option{--no-static-loop} (@command{gnatmetric})
@item --no-static-loop
Do not consider static loops when computing cyclomatic complexity

@item --extra-exit-points
Report the extra exit points for subprogram bodies. As an exit point, this
metric counts @code{return} statements and raise statements in case when the
raised exception is not handled in the same body. In case of a function this
metric subtracts 1 from the number of exit points, because a function body
must contain at least one @code{return} statement.

@item --no-extra-exit-points
Do not report the extra exit points for subprogram bodies
@end table


@node Coupling Metrics Control
@subsubsection Coupling Metrics Control
@cindex Coupling metrics control in @command{gnatmetric}

@noindent
@cindex Coupling metrics (in @command{gnatmetric})
Coupling metrics measure the dependencies between a given entity and other
entities in the program. This information is useful since high coupling
may signal potential issues with maintainability as the program evolves.

@command{gnatmetric} computes the following coupling metrics:

@itemize @bullet

@item
@emph{object-oriented coupling}, for classes in traditional object-oriented
sense;

@item
@emph{unit coupling}, for all the program units making up a program;

@item
@emph{control coupling}, reflecting dependencies between a unit and
other units that contain subprograms.
@end itemize

@noindent
Two kinds of coupling metrics are computed:

@itemize @bullet
@item fan-out coupling (``efferent coupling''):
@cindex fan-out coupling
@cindex efferent coupling
the number of entities the given entity depends upon. This metric
reflects how the given entity depends on the changes in the
``external world''.

@item fan-in coupling (``afferent'' coupling):
@cindex fan-in coupling
@cindex afferent coupling
the number of entities that depend on a given entity.
This metric reflects how the ``external world'' depends on the changes in a
given entity.
@end itemize

@noindent
Object-oriented coupling metrics measure the dependencies
between a given class (or a group of classes) and the other classes in the
program. In this subsection the term ``class'' is used in its traditional
object-oriented programming sense (an instantiable module that contains data
and/or method members). A @emph{category} (of classes) is a group of closely
related classes that are reused and/or modified together.

A class @code{K}'s fan-out coupling is the number of classes
that @code{K} depends upon.
A category's fan-out coupling is the number of classes outside the
category that the classes inside the category depend upon.

A class @code{K}'s fan-in coupling is the number of classes
that depend upon @code{K}.
A category's fan-in coupling is the number of classes outside the
category that depend on classes belonging to the category.

Ada's object-oriented paradigm separates the instantiable entity
(type) from the module (package), so the definition of the coupling
metrics for Ada maps the class and class category notions
onto Ada constructs.

For the coupling metrics, several kinds of modules that define a tagged type
or an interface type  -- library packages, library generic packages, and
library generic package instantiations -- are considered to be classes.
A category consists of a library package (or
a library generic package) that defines a tagged or an interface type,
together with all its descendant (generic) packages that define tagged
or interface types. Thus a
category is an Ada hierarchy of library-level program units. Class
coupling in Ada is referred to as ``tagged coupling'', and category coupling
is referred to as ``hierarchy coupling''.

For any package serving as a class, its body and subunits (if any) are
considered together with its spec when computing dependencies, and coupling
metrics are reported for spec units only. Dependencies between classes
mean Ada semantic dependencies. For object-oriented coupling
metrics, only dependencies on units treated as classes are
considered.

Similarly, for unit and control coupling an entity is considered to be the
conceptual construct consisting of the entity's specification, body, and
any subunits (transitively).
@command{gnatmetric} computes
the dependencies of all these units as a whole, but
metrics are only reported for spec
units (or for a subprogram body unit in case if there is no
separate spec for the given subprogram).

For unit coupling, dependencies are computed between all kinds of program
units. For control coupling, the dependencies of a given unit are limited to
those units that define subprograms. Thus control fan-out coupling is reported
for all units, but control fan-in coupling is only reported for units
that define subprograms.

The following simple example illustrates the difference between unit coupling
and control coupling metrics:

@smallexample @c ada
@group
@b{package} Lib_1 @b{is}
    @b{function} F_1 (I : Integer) @b{return} Integer;
@b{end} Lib_1;
@end group

@group
@b{package} Lib_2 @b{is}
    @b{type} T_2 @b{is} @b{new} Integer;
@b{end} Lib_2;
@end group

@group
@b{package} @b{body} Lib_1 @b{is}
    @b{function} F_1 (I : Integer) @b{return} Integer @b{is}
    @b{begin}
       @b{return} I + 1;
    @b{end} F_1;
@b{end} Lib_1;
@end group

@group
@b{with} Lib_2; @b{use} Lib_2;
@b{package} Pack @b{is}
    Var : T_2;
    @b{function} Fun (I : Integer) @b{return} Integer;
@b{end} Pack;
@end group

@group
@b{with} Lib_1; @b{use} Lib_1;
@b{package} @b{body} Pack @b{is}
    @b{function} Fun (I : Integer) @b{return} Integer @b{is}
    @b{begin}
       @b{return} F_1 (I);
    @b{end} Fun;
@b{end} Pack;
@end group
@end smallexample

@noindent
If we apply @command{gnatmetric} with the @option{--coupling-all} option to
these units, the result will be:

@smallexample
@group
Coupling metrics:
=================
    Unit Lib_1 (C:\customers\662\L406-007\lib_1.ads)
       control fan-out coupling  : 0
       control fan-in coupling   : 1
       unit fan-out coupling     : 0
       unit fan-in coupling      : 1
@end group

@group
    Unit Pack (C:\customers\662\L406-007\pack.ads)
       control fan-out coupling  : 1
       control fan-in coupling   : 0
       unit fan-out coupling     : 2
       unit fan-in coupling      : 0
@end group

@group
    Unit Lib_2 (C:\customers\662\L406-007\lib_2.ads)
       control fan-out coupling  : 0
       unit fan-out coupling     : 0
       unit fan-in coupling      : 1
@end group
@end smallexample

@noindent
The result does not contain values for object-oriented
coupling because none of the argument units contains a tagged type and
therefore none of these units can be treated as a class.

The @code{Pack} package (spec and body) depends on two
units -- @code{Lib_1} @code{and Lib_2} -- and so its unit fan-out coupling
is 2. Since nothing depends on it, its unit fan-in coupling is 0, as
is its control fan-in coupling. Only one of the units @code{Pack} depends
upon defines a subprogram, so its control fan-out coupling is 1.

@code{Lib_2} depends on nothing, so its fan-out metrics are 0. It does
not define any subprograms, so it has no control fan-in metric.
One unit (@code{Pack}) depends on it , so its unit fan-in coupling is 1.

@code{Lib_1} is similar to @code{Lib_2}, but it does define a subprogram.
Its control fan-in coupling is 1 (because there is one unit
depending on it).

When computing coupling metrics, @command{gnatmetric} counts only
dependencies between units that are arguments of the @command{gnatmetric}
invocation. Coupling metrics are program-wide (or project-wide) metrics, so
you should invoke @command{gnatmetric} for
the complete set of sources comprising your program. This can be done
by invoking @command{gnatmetric} with the corresponding project file
and with the @option{-U} option.

By default, all the coupling metrics are disabled. You can use the following
switches to specify the coupling metrics to be computed and reported:

@table @option

@cindex @option{--tagged-coupling@var{x}} (@command{gnatmetric})
@cindex @option{--hierarchy-coupling@var{x}} (@command{gnatmetric})
@cindex @option{--unit-coupling@var{x}} (@command{gnatmetric})
@cindex @option{--control-coupling@var{x}} (@command{gnatmetric})


@item --coupling-all
Report all the coupling metrics

@item --tagged-coupling-out
Report tagged (class) fan-out coupling

@item --tagged-coupling-in
Report tagged (class) fan-in coupling

@item --hierarchy-coupling-out
Report hierarchy (category) fan-out coupling

@item --hierarchy-coupling-in
Report hierarchy (category) fan-in coupling

@item --unit-coupling-out
Report unit fan-out coupling

@item --unit-coupling-in
Report unit fan-in coupling

@item --control-coupling-out
Report control fan-out coupling

@item --control-coupling-in
Report control fan-in coupling
@end table

@node Other gnatmetric Switches
@subsection Other @code{gnatmetric} Switches

@noindent
Additional @command{gnatmetric} switches are as follows:

@table @option
@item --version
@cindex @option{--version} @command{gnatmetric}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatmetric}
Display usage, then exit disregarding all other options.

@item -P @var{file}
@cindex @option{-P} @command{gnatmetric}
Indicates the name of the project file that describes the set of sources
to be processed. The exact set of argument sources depends on other options
specified, see below.

@item -U
@cindex @option{-U} @command{gnatmetric}
If a project file is specified and no argument source is explicitly
specified (either directly or by means of @option{-files} option), process
all the units of the closure of the argument project. Otherwise this option
has no effect.

@item -U @var{main_unit}
If a project file is specified and no argument source is explicitly
specified (either directly or by means of @option{-files} option), process
the closure of units rooted at @var{main_unit}. Otherwise this option
has no effect.

@item -X@var{name}=@var{value}
@cindex @option{-X} @command{gnatmetric}
Indicates that external variable @var{name} in the argument project
has the value @var{value}. Has no effect if no project is specified as
tool argument.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gnatmetric})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item --subdirs=@var{dir}
@cindex @option{--subdirs=@var{dir}} @command{gnatmetric}
Use the specified subdirectory of the project objects file (or of the
project file directory if the project does not specify an object directory)
for tool output files. Has no effect if no project is specified as
tool argument r if @option{--no_objects_dir} is specified.

@item --no_objects_dir
@cindex @option{--no_objects_dir} @command{gnatmetric}
Place all the result files into the current directory instead of
project objects directory. This corresponds to the @command{gnatcheck}
behavior when it is called with the project file from the
GNAT driver. Has no effect if no project is specified.

@item -files @var{filename}
@cindex @option{-files} (@code{gnatmetric})
Take the argument source files from the specified file. This file should be an
ordinary text file containing file names separated by spaces or
line breaks. You can use this switch more than once in the same call to
@command{gnatmetric}. You also can combine this switch with
an explicit list of files.

@item -j@var{n}
@cindex @option{-j} (@command{gnatmetric})
Use @var{n} processes to carry out the tree creations (internal representations
of the argument sources). On a multiprocessor machine this speeds up processing
of big sets of argument sources. If @var{n} is 0, then the maximum number of
parallel tree creations is the number of core processors on the platform.

@cindex @option{-t} (@command{gnatmetric})
@item -t
Print out execution time.

@item -v
@cindex @option{-v} (@command{gnatmetric})
Verbose mode;
@command{gnatmetric} generates version information and then
a trace of sources being processed.

@item -q
@cindex @option{-q} (@command{gnatmetric})
Quiet mode.
@end table

@noindent
If a project file is specified and no argument source is explicitly
specified (either directly or by means of @option{-files} option), and no
@option{-U} is specified, then the set of processed sources is
all the immediate units of the argument project.


@ignore
@node Generate project-wide metrics
@subsection Generate project-wide metrics

In order to compute metrics on all units of a given project, you can use
the @command{gnat} driver along with the @option{-P} option:
@smallexample
   gnat metric -Pproj
@end smallexample

@noindent
If the project @code{proj} depends upon other projects, you can compute
the metrics on the project closure using the @option{-U} option:
@smallexample
   gnat metric -Pproj -U
@end smallexample

@noindent
Finally, if not all the units are relevant to a particular main
program in the project closure, you can generate metrics for the set
of units needed to create a given main program (unit closure) using
the @option{-U} option followed by the name of the main unit:
@smallexample
   gnat metric -Pproj -U main
@end smallexample
@end ignore
@end ifclear


@c ***********************************
@node File Name Krunching with gnatkr
@chapter File Name Krunching with @code{gnatkr}
@findex gnatkr

@noindent
This chapter discusses the method used by the compiler to shorten
the default file names chosen for Ada units so that they do not
exceed the maximum length permitted. It also describes the
@code{gnatkr} utility that can be used to determine the result of
applying this shortening.
@menu
* About gnatkr::
* Using gnatkr::
* Krunching Method::
* Examples of gnatkr Usage::
@end menu

@node About gnatkr
@section About @code{gnatkr}

@noindent
The default file naming rule in GNAT
is that the file name must be derived from
the unit name. The exact default rule is as follows:
@itemize @bullet
@item
Take the unit name and replace all dots by hyphens.
@item
If such a replacement occurs in the
second character position of a name, and the first character is
@samp{a}, @samp{g}, @samp{s}, or @samp{i},
then replace the dot by the character
@samp{~} (tilde)
instead of a minus.
@end itemize
The reason for this exception is to avoid clashes
with the standard names for children of System, Ada, Interfaces,
and GNAT, which use the prefixes
@samp{s-}, @samp{a-}, @samp{i-}, and @samp{g-},
respectively.

The @option{-gnatk@var{nn}}
switch of the compiler activates a ``krunching''
circuit that limits file names to nn characters (where nn is a decimal
integer). For example, using OpenVMS,
where the maximum file name length is
39, the value of nn is usually set to 39, but if you want to generate
a set of files that would be usable if ported to a system with some
different maximum file length, then a different value can be specified.
The default value of 39 for OpenVMS need not be specified.

The @code{gnatkr} utility can be used to determine the krunched name for
a given file, when krunched to a specified maximum length.

@node Using gnatkr
@section Using @code{gnatkr}

@noindent
The @code{gnatkr} command has the form

@smallexample
@c $ gnatkr @var{name} @ovar{length}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatkr @var{name} @r{[}@var{length}@r{]}
@end smallexample


@noindent
@var{name} is the uncrunched file name, derived from the name of the unit
in the standard manner described in the previous section (i.e., in particular
all dots are replaced by hyphens). The file name may or may not have an
extension (defined as a suffix of the form period followed by arbitrary
characters other than period). If an extension is present then it will
be preserved in the output. For example, when krunching @file{hellofile.ads}
to eight characters, the result will be hellofil.ads.

Note: for compatibility with previous versions of @code{gnatkr} dots may
appear in the name instead of hyphens, but the last dot will always be
taken as the start of an extension. So if @code{gnatkr} is given an argument
such as @file{Hello.World.adb} it will be treated exactly as if the first
period had been a hyphen, and for example krunching to eight characters
gives the result @file{hellworl.adb}.

Note that the result is always all lower case (except on OpenVMS where it is
all upper case). Characters of the other case are folded as required.

@var{length} represents the length of the krunched name. The default
when no argument is given is 8 characters. A length of zero stands for
unlimited, in other words do not chop except for system files where the
implied crunching length is always eight characters.

@noindent
The output is the krunched name. The output has an extension only if the
original argument was a file name with an extension.

@node Krunching Method
@section Krunching Method

@noindent
The initial file name is determined by the name of the unit that the file
contains. The name is formed by taking the full expanded name of the
unit and replacing the separating dots with hyphens and
using lowercase
for all letters, except that a hyphen in the second character position is
replaced by a tilde if the first character is
@samp{a}, @samp{i}, @samp{g}, or @samp{s}.
The extension is @code{.ads} for a
spec and @code{.adb} for a body.
Krunching does not affect the extension, but the file name is shortened to
the specified length by following these rules:

@itemize @bullet
@item
The name is divided into segments separated by hyphens, tildes or
underscores and all hyphens, tildes, and underscores are
eliminated. If this leaves the name short enough, we are done.

@item
If the name is too long, the longest segment is located (left-most
if there are two of equal length), and shortened by dropping
its last character. This is repeated until the name is short enough.

As an example, consider the krunching of @*@file{our-strings-wide_fixed.adb}
to fit the name into 8 characters as required by some operating systems.

@smallexample
our-strings-wide_fixed 22
our strings wide fixed 19
our string  wide fixed 18
our strin   wide fixed 17
our stri    wide fixed 16
our stri    wide fixe  15
our str     wide fixe  14
our str     wid  fixe  13
our str     wid  fix   12
ou  str     wid  fix   11
ou  st      wid  fix   10
ou  st      wi   fix   9
ou  st      wi   fi    8
Final file name: oustwifi.adb
@end smallexample

@item
The file names for all predefined units are always krunched to eight
characters. The krunching of these predefined units uses the following
special prefix replacements:

@table @file
@item ada-
replaced by @file{a-}

@item gnat-
replaced by @file{g-}

@item interfaces-
replaced by @file{i-}

@item system-
replaced by @file{s-}
@end table

These system files have a hyphen in the second character position. That
is why normal user files replace such a character with a
tilde, to
avoid confusion with system file names.

As an example of this special rule, consider
@*@file{ada-strings-wide_fixed.adb}, which gets krunched as follows:

@smallexample
ada-strings-wide_fixed 22
a-  strings wide fixed 18
a-  string  wide fixed 17
a-  strin   wide fixed 16
a-  stri    wide fixed 15
a-  stri    wide fixe  14
a-  str     wide fixe  13
a-  str     wid  fixe  12
a-  str     wid  fix   11
a-  st      wid  fix   10
a-  st      wi   fix   9
a-  st      wi   fi    8
Final file name: a-stwifi.adb
@end smallexample
@end itemize

Of course no file shortening algorithm can guarantee uniqueness over all
possible unit names, and if file name krunching is used then it is your
responsibility to ensure that no name clashes occur. The utility
program @code{gnatkr} is supplied for conveniently determining the
krunched name of a file.

@node Examples of gnatkr Usage
@section Examples of @code{gnatkr} Usage

@smallexample
@iftex
@leftskip=0cm
@end iftex
$ gnatkr very_long_unit_name.ads      --> velounna.ads
$ gnatkr grandparent-parent-child.ads --> grparchi.ads
$ gnatkr Grandparent.Parent.Child.ads --> grparchi.ads
$ gnatkr grandparent-parent-child     --> grparchi
$ gnatkr very_long_unit_name.ads/count=6 --> vlunna.ads
$ gnatkr very_long_unit_name.ads/count=0 --> very_long_unit_name.ads
@end smallexample

@node Preprocessing with gnatprep
@chapter Preprocessing with @code{gnatprep}
@findex gnatprep

@noindent
This chapter discusses how to use GNAT's @code{gnatprep} utility for simple
preprocessing.
Although designed for use with GNAT, @code{gnatprep} does not depend on any
special GNAT features.
For further discussion of conditional compilation in general, see
@ref{Conditional Compilation}.

@menu
* Preprocessing Symbols::
* Using gnatprep::
* Switches for gnatprep::
* Form of Definitions File::
* Form of Input Text for gnatprep::
@end menu

@node Preprocessing Symbols
@section Preprocessing Symbols

@noindent
Preprocessing symbols are defined in definition files and referred to in
sources to be preprocessed. A Preprocessing symbol is an identifier, following
normal Ada (case-insensitive) rules for its syntax, with the restriction that
all characters need to be in the ASCII set (no accented letters).

@node Using gnatprep
@section Using @code{gnatprep}

@noindent
To call @code{gnatprep} use

@smallexample
@c $ gnatprep @ovar{switches} @var{infile} @var{outfile} @ovar{deffile}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatprep @r{[}@var{switches}@r{]} @var{infile} @var{outfile} @r{[}@var{deffile}@r{]}
@end smallexample

@noindent
where
@table @var
@item switches
is an optional sequence of switches as described in the next section.

@item infile
is the full name of the input file, which is an Ada source
file containing preprocessor directives.

@item outfile
is the full name of the output file, which is an Ada source
in standard Ada form. When used with GNAT, this file name will
normally have an ads or adb suffix.

@item deffile
is the full name of a text file containing definitions of
preprocessing symbols to be referenced by the preprocessor. This argument is
optional, and can be replaced by the use of the @option{-D} switch.

@end table

@node Switches for gnatprep
@section Switches for @code{gnatprep}

@table @option
@c !sort!

@item -b
@cindex @option{-b} (@command{gnatprep})
Causes both preprocessor lines and the lines deleted by
preprocessing to be replaced by blank lines in the output source file,
preserving line numbers in the output file.

@item -c
@cindex @option{-c} (@command{gnatprep})
Causes both preprocessor lines and the lines deleted
by preprocessing to be retained in the output source as comments marked
with the special string @code{"--! "}. This option will result in line numbers
being preserved in the output file.

@item -C
@cindex @option{-C} (@command{gnatprep})
Causes comments to be scanned. Normally comments are ignored by gnatprep.
If this option is specified, then comments are scanned and any $symbol
substitutions performed as in program text. This is particularly useful
when structured comments are used (e.g., when writing programs in the
SPARK dialect of Ada). Note that this switch is not available when
doing integrated preprocessing (it would be useless in this context
since comments are ignored by the compiler in any case).

@item -Dsymbol=value
@cindex @option{-D} (@command{gnatprep})
Defines a new preprocessing symbol, associated with value. If no value is given
on the command line, then symbol is considered to be @code{True}. This switch
can be used in place of a definition file.


@item -r
@cindex @option{-r} (@command{gnatprep})
Causes a @code{Source_Reference} pragma to be generated that
references the original input file, so that error messages will use
the file name of this original file. The use of this switch implies
that preprocessor lines are not to be removed from the file, so its
use will force @option{-b} mode if
@option{-c}
has not been specified explicitly.

Note that if the file to be preprocessed contains multiple units, then
it will be necessary to @code{gnatchop} the output file from
@code{gnatprep}. If a @code{Source_Reference} pragma is present
in the preprocessed file, it will be respected by
@code{gnatchop -r}
so that the final chopped files will correctly refer to the original
input source file for @code{gnatprep}.

@item -s
@cindex @option{-s} (@command{gnatprep})
Causes a sorted list of symbol names and values to be
listed on the standard output file.

@item -u
@cindex @option{-u} (@command{gnatprep})
Causes undefined symbols to be treated as having the value FALSE in the context
of a preprocessor test. In the absence of this option, an undefined symbol in
a @code{#if} or @code{#elsif} test will be treated as an error.

@end table

@noindent
Note: if neither @option{-b} nor @option{-c} is present,
then preprocessor lines and
deleted lines are completely removed from the output, unless -r is
specified, in which case -b is assumed.

@node Form of Definitions File
@section Form of Definitions File

@noindent
The definitions file contains lines of the form

@smallexample
symbol := value
@end smallexample

@noindent
where symbol is a preprocessing symbol, and value is one of the following:

@itemize @bullet
@item
Empty, corresponding to a null substitution
@item
A string literal using normal Ada syntax
@item
Any sequence of characters from the set
(letters, digits, period, underline).
@end itemize

@noindent
Comment lines may also appear in the definitions file, starting with
the usual @code{--},
and comments may be added to the definitions lines.

@node Form of Input Text for gnatprep
@section Form of Input Text for @code{gnatprep}

@noindent
The input text may contain preprocessor conditional inclusion lines,
as well as general symbol substitution sequences.

The preprocessor conditional inclusion commands have the form

@smallexample
@group
@cartouche
#if @i{expression} @r{[}then@r{]}
   lines
#elsif @i{expression} @r{[}then@r{]}
   lines
#elsif @i{expression} @r{[}then@r{]}
   lines
@dots{}
#else
   lines
#end if;
@end cartouche
@end group
@end smallexample

@noindent
In this example, @i{expression} is defined by the following grammar:
@smallexample
@i{expression} ::=  <symbol>
@i{expression} ::=  <symbol> = "<value>"
@i{expression} ::=  <symbol> = <symbol>
@i{expression} ::=  <symbol> = <integer>
@i{expression} ::=  <symbol> > <integer>
@i{expression} ::=  <symbol> >= <integer>
@i{expression} ::=  <symbol> < <integer>
@i{expression} ::=  <symbol> <= <integer>
@i{expression} ::=  <symbol> 'Defined
@i{expression} ::=  not @i{expression}
@i{expression} ::=  @i{expression} and @i{expression}
@i{expression} ::=  @i{expression} or @i{expression}
@i{expression} ::=  @i{expression} and then @i{expression}
@i{expression} ::=  @i{expression} or else @i{expression}
@i{expression} ::=  ( @i{expression} )
@end smallexample

The following restriction exists: it is not allowed to have "and" or "or"
following "not" in the same expression without parentheses. For example, this
is not allowed:

@smallexample
   not X or Y
@end smallexample

This should be one of the following:

@smallexample
   (not X) or Y
   not (X or Y)
@end smallexample

@noindent
For the first test (@i{expression} ::= <symbol>) the symbol must have
either the value true or false, that is to say the right-hand of the
symbol definition must be one of the (case-insensitive) literals
@code{True} or @code{False}. If the value is true, then the
corresponding lines are included, and if the value is false, they are
excluded.

When comparing a symbol to an integer, the integer is any non negative
literal integer as defined in the Ada Reference Manual, such as 3, 16#FF# or
2#11#. The symbol value must also be a non negative integer. Integer values
in the range 0 .. 2**31-1 are supported.

The test (@i{expression} ::= <symbol> @code{'Defined}) is true only if
the symbol has been defined in the definition file or by a @option{-D}
switch on the command line. Otherwise, the test is false.

The equality tests are case insensitive, as are all the preprocessor lines.

If the symbol referenced is not defined in the symbol definitions file,
then the effect depends on whether or not switch @option{-u}
is specified. If so, then the symbol is treated as if it had the value
false and the test fails. If this switch is not specified, then
it is an error to reference an undefined symbol. It is also an error to
reference a symbol that is defined with a value other than @code{True}
or @code{False}.

The use of the @code{not} operator inverts the sense of this logical test.
The @code{not} operator cannot be combined with the @code{or} or @code{and}
operators, without parentheses. For example, "if not X or Y then" is not
allowed, but "if (not X) or Y then" and "if not (X or Y) then" are.

The @code{then} keyword is optional as shown

The @code{#} must be the first non-blank character on a line, but
otherwise the format is free form. Spaces or tabs may appear between
the @code{#} and the keyword. The keywords and the symbols are case
insensitive as in normal Ada code. Comments may be used on a
preprocessor line, but other than that, no other tokens may appear on a
preprocessor line. Any number of @code{elsif} clauses can be present,
including none at all. The @code{else} is optional, as in Ada.

The @code{#} marking the start of a preprocessor line must be the first
non-blank character on the line, i.e., it must be preceded only by
spaces or horizontal tabs.

Symbol substitution outside of preprocessor lines is obtained by using
the sequence

@smallexample
$symbol
@end smallexample

@noindent
anywhere within a source line, except in a comment or within a
string literal. The identifier
following the @code{$} must match one of the symbols defined in the symbol
definition file, and the result is to substitute the value of the
symbol in place of @code{$symbol} in the output file.

Note that although the substitution of strings within a string literal
is not possible, it is possible to have a symbol whose defined value is
a string literal. So instead of setting XYZ to @code{hello} and writing:

@smallexample
Header : String := "$XYZ";
@end smallexample

@noindent
you should set XYZ to @code{"hello"} and write:

@smallexample
Header : String := $XYZ;
@end smallexample

@noindent
and then the substitution will occur as desired.

@node The GNAT Library Browser gnatls
@chapter The GNAT Library Browser @code{gnatls}
@findex gnatls
@cindex Library browser

@noindent
@code{gnatls} is a tool that outputs information about compiled
units. It gives the relationship between objects, unit names and source
files. It can also be used to check the source dependencies of a unit
as well as various characteristics.

Note: to invoke @code{gnatls} with a project file, use the @code{gnat}
driver (see @ref{The GNAT Driver and Project Files}).

@menu
* Running gnatls::
* Switches for gnatls::
* Examples of gnatls Usage::
@end menu

@node Running gnatls
@section Running @code{gnatls}

@noindent
The @code{gnatls} command has the form

@smallexample
$ gnatls switches @var{object_or_ali_file}
@end smallexample

@noindent
The main argument is the list of object or @file{ali} files
(@pxref{The Ada Library Information Files})
for which information is requested.

In normal mode, without additional option, @code{gnatls} produces a
four-column listing. Each line represents information for a specific
object. The first column gives the full path of the object, the second
column gives the name of the principal unit in this object, the third
column gives the status of the source and the fourth column gives the
full path of the source representing this unit.
Here is a simple example of use:

@smallexample
$ gnatls *.o
./demo1.o            demo1            DIF demo1.adb
./demo2.o            demo2             OK demo2.adb
./hello.o            h1                OK hello.adb
./instr-child.o      instr.child      MOK instr-child.adb
./instr.o            instr             OK instr.adb
./tef.o              tef              DIF tef.adb
./text_io_example.o  text_io_example   OK text_io_example.adb
./tgef.o             tgef             DIF tgef.adb
@end smallexample

@noindent
The first line can be interpreted as follows: the main unit which is
contained in
object file @file{demo1.o} is demo1, whose main source is in
@file{demo1.adb}. Furthermore, the version of the source used for the
compilation of demo1 has been modified (DIF). Each source file has a status
qualifier which can be:

@table @code
@item OK (unchanged)
The version of the source file used for the compilation of the
specified unit corresponds exactly to the actual source file.

@item MOK (slightly modified)
The version of the source file used for the compilation of the
specified unit differs from the actual source file but not enough to
require recompilation. If you use gnatmake with the qualifier
@option{-m (minimal recompilation)}, a file marked
MOK will not be recompiled.

@item DIF (modified)
No version of the source found on the path corresponds to the source
used to build this object.

@item ??? (file not found)
No source file was found for this unit.

@item HID (hidden,  unchanged version not first on PATH)
The version of the source that corresponds exactly to the source used
for compilation has been found on the path but it is hidden by another
version of the same source that has been modified.

@end table

@node Switches for gnatls
@section Switches for @code{gnatls}

@noindent
@code{gnatls} recognizes the following switches:

@table @option
@c !sort!
@cindex @option{--version} @command{gnatls}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatls}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item -a
@cindex @option{-a} (@code{gnatls})
Consider all units, including those of the predefined Ada library.
Especially useful with @option{-d}.

@item -d
@cindex @option{-d} (@code{gnatls})
List sources from which specified units depend on.

@item -h
@cindex @option{-h} (@code{gnatls})
Output the list of options.

@item -o
@cindex @option{-o} (@code{gnatls})
Only output information about object files.

@item -s
@cindex @option{-s} (@code{gnatls})
Only output information about source files.

@item -u
@cindex @option{-u} (@code{gnatls})
Only output information about compilation units.

@item -files=@var{file}
@cindex @option{-files} (@code{gnatls})
Take as arguments the files listed in text file @var{file}.
Text file @var{file} may contain empty lines that are ignored.
Each nonempty line should contain the name of an existing file.
Several such switches may be specified simultaneously.

@item -aO@var{dir}
@itemx -aI@var{dir}
@itemx -I@var{dir}
@itemx  -I-
@itemx -nostdinc
@cindex @option{-aO} (@code{gnatls})
@cindex @option{-aI} (@code{gnatls})
@cindex @option{-I} (@code{gnatls})
@cindex @option{-I-} (@code{gnatls})
Source path manipulation. Same meaning as the equivalent @command{gnatmake}
flags (@pxref{Switches for gnatmake}).

@item -aP@var{dir}
@cindex @option{-aP} (@code{gnatls})
Add @var{dir} at the beginning of the project search dir.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@code{gnatls})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item -v
@cindex @option{-v} (@code{gnatls})
Verbose mode. Output the complete source, object and project paths. Do not use
the default column layout but instead use long format giving as much as
information possible on each requested units, including special
characteristics such as:

@table @code
@item  Preelaborable
The unit is preelaborable in the Ada sense.

@item No_Elab_Code
No elaboration code has been produced by the compiler for this unit.

@item Pure
The unit is pure in the Ada sense.

@item Elaborate_Body
The unit contains a pragma Elaborate_Body.

@item Remote_Types
The unit contains a pragma Remote_Types.

@item Shared_Passive
The unit contains a pragma Shared_Passive.

@item Predefined
This unit is part of the predefined environment and cannot be modified
by the user.

@item Remote_Call_Interface
The unit contains a pragma Remote_Call_Interface.

@end table

@end table

@node Examples of gnatls Usage
@section Example of @code{gnatls} Usage

@noindent
Example of using the verbose switch. Note how the source and
object paths are affected by the -I switch.

@smallexample
$ gnatls -v -I.. demo1.o

GNATLS 5.03w (20041123-34)
Copyright 1997-2004 Free Software Foundation, Inc.

Source Search Path:
   <Current_Directory>
   ../
   /home/comar/local/adainclude/

Object Search Path:
   <Current_Directory>
   ../
   /home/comar/local/lib/gcc-lib/x86-linux/3.4.3/adalib/

Project Search Path:
   <Current_Directory>
   /home/comar/local/lib/gnat/

./demo1.o
   Unit =>
     Name   => demo1
     Kind   => subprogram body
     Flags  => No_Elab_Code
     Source => demo1.adb    modified
@end smallexample

@noindent
The following is an example of use of the dependency list.
Note the use of the -s switch
which gives a straight list of source files. This can be useful for
building specialized scripts.

@smallexample
$ gnatls -d demo2.o
./demo2.o   demo2        OK demo2.adb
                         OK gen_list.ads
                         OK gen_list.adb
                         OK instr.ads
                         OK instr-child.ads

$ gnatls -d -s -a demo1.o
demo1.adb
/home/comar/local/adainclude/ada.ads
/home/comar/local/adainclude/a-finali.ads
/home/comar/local/adainclude/a-filico.ads
/home/comar/local/adainclude/a-stream.ads
/home/comar/local/adainclude/a-tags.ads
gen_list.ads
gen_list.adb
/home/comar/local/adainclude/gnat.ads
/home/comar/local/adainclude/g-io.ads
instr.ads
/home/comar/local/adainclude/system.ads
/home/comar/local/adainclude/s-exctab.ads
/home/comar/local/adainclude/s-finimp.ads
/home/comar/local/adainclude/s-finroo.ads
/home/comar/local/adainclude/s-secsta.ads
/home/comar/local/adainclude/s-stalib.ads
/home/comar/local/adainclude/s-stoele.ads
/home/comar/local/adainclude/s-stratt.ads
/home/comar/local/adainclude/s-tasoli.ads
/home/comar/local/adainclude/s-unstyp.ads
/home/comar/local/adainclude/unchconv.ads
@end smallexample


@node Cleaning Up with gnatclean
@chapter Cleaning Up with @code{gnatclean}
@findex gnatclean
@cindex Cleaning tool

@noindent
@code{gnatclean} is a tool that allows the deletion of files produced by the
compiler, binder and linker, including ALI files, object files, tree files,
expanded source files, library files, interface copy source files, binder
generated files and executable files.

@menu
* Running gnatclean::
* Switches for gnatclean::
@c * Examples of gnatclean Usage::
@end menu

@node Running gnatclean
@section Running @code{gnatclean}

@noindent
The @code{gnatclean} command has the form:

@smallexample
$ gnatclean switches @var{names}
@end smallexample

@noindent
@var{names} is a list of source file names. Suffixes @code{.ads} and
@code{adb} may be omitted. If a project file is specified using switch
@code{-P}, then @var{names} may be completely omitted.

@noindent
In normal mode, @code{gnatclean} delete the files produced by the compiler and,
if switch @code{-c} is not specified, by the binder and
the linker. In informative-only mode, specified by switch
@code{-n}, the list of files that would have been deleted in
normal mode is listed, but no file is actually deleted.

@node Switches for gnatclean
@section Switches for @code{gnatclean}

@noindent
@code{gnatclean} recognizes the following switches:

@table @option
@c !sort!
@cindex @option{--version} @command{gnatclean}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatclean}
If @option{--version} was not used, display usage, then exit disregarding
all other options.

@item --subdirs=subdir
Actual object directory of each project file is the subdirectory subdir of the
object directory specified or defaulted in the project file.

@item --unchecked-shared-lib-imports
By default, shared library projects are not allowed to import static library
projects. When this switch is used on the command line, this restriction is
relaxed.

@item -c
@cindex @option{-c} (@code{gnatclean})
Only attempt to delete the files produced by the compiler, not those produced
by the binder or the linker. The files that are not to be deleted are library
files, interface copy files, binder generated files and executable files.

@item -D @var{dir}
@cindex @option{-D} (@code{gnatclean})
Indicate that ALI and object files should normally be found in directory
@var{dir}.

@item -F
@cindex @option{-F} (@code{gnatclean})
When using project files, if some errors or warnings are detected during
parsing and verbose mode is not in effect (no use of switch
-v), then error lines start with the full path name of the project
file, rather than its simple file name.

@item -h
@cindex @option{-h} (@code{gnatclean})
Output a message explaining the usage of @code{gnatclean}.

@item -n
@cindex @option{-n} (@code{gnatclean})
Informative-only mode. Do not delete any files. Output the list of the files
that would have been deleted if this switch was not specified.

@item -P@var{project}
@cindex @option{-P} (@code{gnatclean})
Use project file @var{project}. Only one such switch can be used.
When cleaning a project file, the files produced by the compilation of the
immediate sources or inherited sources of the project files are to be
deleted. This is not depending on the presence or not of executable names
on the command line.

@item -q
@cindex @option{-q} (@code{gnatclean})
Quiet output. If there are no errors, do not output anything, except in
verbose mode (switch -v) or in informative-only mode
(switch -n).

@item -r
@cindex @option{-r} (@code{gnatclean})
When a project file is specified (using switch -P),
clean all imported and extended project files, recursively. If this switch
is not specified, only the files related to the main project file are to be
deleted. This switch has no effect if no project file is specified.

@item -v
@cindex @option{-v} (@code{gnatclean})
Verbose mode.

@item -vP@emph{x}
@cindex @option{-vP} (@code{gnatclean})
Indicates the verbosity of the parsing of GNAT project files.
@xref{Switches Related to Project Files}.

@item -X@var{name=value}
@cindex @option{-X} (@code{gnatclean})
Indicates that external variable @var{name} has the value @var{value}.
The Project Manager will use this value for occurrences of
@code{external(name)} when parsing the project file.
@xref{Switches Related to Project Files}.

@item -aO@var{dir}
@cindex @option{-aO} (@code{gnatclean})
When searching for ALI and object files, look in directory
@var{dir}.

@item -I@var{dir}
@cindex @option{-I} (@code{gnatclean})
Equivalent to @option{-aO@var{dir}}.

@item -I-
@cindex @option{-I-} (@code{gnatclean})
@cindex Source files, suppressing search
Do not look for ALI or object files in the directory
where @code{gnatclean} was invoked.

@end table

@c @node Examples of gnatclean Usage
@c @section Examples of @code{gnatclean} Usage

@node GNAT and Libraries
@chapter GNAT and Libraries
@cindex Library, building, installing, using

@noindent
This chapter describes how to build and use libraries with GNAT, and also shows
how to recompile the GNAT run-time library. You should be familiar with the
Project Manager facility (@pxref{GNAT Project Manager}) before reading this
chapter.

@menu
* Introduction to Libraries in GNAT::
* General Ada Libraries::
* Stand-alone Ada Libraries::
* Rebuilding the GNAT Run-Time Library::
@end menu

@node Introduction to Libraries in GNAT
@section Introduction to Libraries in GNAT

@noindent
A library is, conceptually, a collection of objects which does not have its
own main thread of execution, but rather provides certain services to the
applications that use it. A library can be either statically linked with the
application, in which case its code is directly included in the application,
or, on platforms that support it, be dynamically linked, in which case
its code is shared by all applications making use of this library.

GNAT supports both types of libraries.
In the static case, the compiled code can be provided in different ways. The
simplest approach is to provide directly the set of objects resulting from
compilation of the library source files. Alternatively, you can group the
objects into an archive using whatever commands are provided by the operating
system. For the latter case, the objects are grouped into a shared library.

In the GNAT environment, a library has three types of components:
@itemize @bullet
@item
Source files.
@item
@file{ALI} files.
@xref{The Ada Library Information Files}.
@item
Object files, an archive or a shared library.
@end itemize

@noindent
A GNAT library may expose all its source files, which is useful for
documentation purposes. Alternatively, it may expose only the units needed by
an external user to make use of the library. That is to say, the specs
reflecting the library services along with all the units needed to compile
those specs, which can include generic bodies or any body implementing an
inlined routine. In the case of @emph{stand-alone libraries} those exposed
units are called @emph{interface units} (@pxref{Stand-alone Ada Libraries}).

All compilation units comprising an application, including those in a library,
need to be elaborated in an order partially defined by Ada's semantics. GNAT
computes the elaboration order from the @file{ALI} files and this is why they
constitute a mandatory part of GNAT libraries.
@emph{Stand-alone libraries} are the exception to this rule because a specific
library elaboration routine is produced independently of the application(s)
using the library.

@node General Ada Libraries
@section General Ada Libraries

@menu
* Building a library::
* Installing a library::
* Using a library::
@end menu

@node Building a library
@subsection Building a library

@noindent
The easiest way to build a library is to use the Project Manager,
which supports a special type of project called a @emph{Library Project}
(@pxref{Library Projects}).

A project is considered a library project, when two project-level attributes
are defined in it: @code{Library_Name} and @code{Library_Dir}. In order to
control different aspects of library configuration, additional optional
project-level attributes can be specified:
@table @code
@item Library_Kind
This attribute controls whether the library is to be static or dynamic

@item Library_Version
This attribute specifies the library version; this value is used
during dynamic linking of shared libraries to determine if the currently
installed versions of the binaries are compatible.

@item Library_Options
@item Library_GCC
These attributes specify additional low-level options to be used during
library generation, and redefine the actual application used to generate
library.
@end table

@noindent
The GNAT Project Manager takes full care of the library maintenance task,
including recompilation of the source files for which objects do not exist
or are not up to date, assembly of the library archive, and installation of
the library (i.e., copying associated source, object and @file{ALI} files
to the specified location).

Here is a simple library project file:
@smallexample @c ada
project My_Lib @b{is}
   @b{for} Source_Dirs @b{use} ("src1", "src2");
   @b{for} Object_Dir @b{use} "obj";
   @b{for} Library_Name @b{use} "mylib";
   @b{for} Library_Dir @b{use} "lib";
   @b{for} Library_Kind @b{use} "dynamic";
@b{end} My_lib;
@end smallexample

@noindent
and the compilation command to build and install the library:

@smallexample @c ada
  $ gnatmake -Pmy_lib
@end smallexample

@noindent
It is not entirely trivial to perform manually all the steps required to
produce a library. We recommend that you use the GNAT Project Manager
for this task. In special cases where this is not desired, the necessary
steps are discussed below.

There are various possibilities for compiling the units that make up the
library: for example with a Makefile (@pxref{Using the GNU make Utility}) or
with a conventional script. For simple libraries, it is also possible to create
a dummy main program which depends upon all the packages that comprise the
interface of the library. This dummy main program can then be given to
@command{gnatmake}, which will ensure that all necessary objects are built.

After this task is accomplished, you should follow the standard procedure
of the underlying operating system to produce the static or shared library.

Here is an example of such a dummy program:
@smallexample @c ada
@group
@b{with} My_Lib.Service1;
@b{with} My_Lib.Service2;
@b{with} My_Lib.Service3;
@b{procedure} My_Lib_Dummy @b{is}
@b{begin}
   @b{null};
@b{end};
@end group
@end smallexample

@noindent
Here are the generic commands that will build an archive or a shared library.

@smallexample
# compiling the library
$ gnatmake -c my_lib_dummy.adb

# we don't need the dummy object itself
$ rm my_lib_dummy.o my_lib_dummy.ali

# create an archive with the remaining objects
$ ar rc libmy_lib.a *.o
# some systems may require "ranlib" to be run as well

# or create a shared library
$ gcc -shared -o libmy_lib.so *.o
# some systems may require the code to have been compiled with -fPIC

# remove the object files that are now in the library
$ rm *.o

# Make the ALI files read-only so that gnatmake will not try to
# regenerate the objects that are in the library
$ chmod -w *.ali
@end smallexample

@noindent
Please note that the library must have a name of the form @file{lib@var{xxx}.a}
or @file{lib@var{xxx}.so} (or @file{lib@var{xxx}.dll} on Windows) in order to
be accessed by the directive @option{-l@var{xxx}} at link time.

@node Installing a library
@subsection Installing a library
@cindex @code{ADA_PROJECT_PATH}
@cindex @code{GPR_PROJECT_PATH}

@noindent
If you use project files, library installation is part of the library build
process (@pxref{Installing a library with project files}).

When project files are not an option, it is also possible, but not recommended,
to install the library so that the sources needed to use the library are on the
Ada source path and the ALI files & libraries be on the Ada Object path (see
@ref{Search Paths and the Run-Time Library (RTL)}. Alternatively, the system
administrator can place general-purpose libraries in the default compiler
paths, by specifying the libraries' location in the configuration files
@file{ada_source_path} and @file{ada_object_path}. These configuration files
must be located in the GNAT installation tree at the same place as the gcc spec
file. The location of the gcc spec file can be determined as follows:
@smallexample
$ gcc -v
@end smallexample

@noindent
The configuration files mentioned above have a simple format: each line
must contain one unique directory name.
Those names are added to the corresponding path
in their order of appearance in the file. The names can be either absolute
or relative; in the latter case, they are relative to where theses files
are located.

The files @file{ada_source_path} and @file{ada_object_path} might not be
present in a
GNAT installation, in which case, GNAT will look for its run-time library in
the directories @file{adainclude} (for the sources) and @file{adalib} (for the
objects and @file{ALI} files). When the files exist, the compiler does not
look in @file{adainclude} and @file{adalib}, and thus the
@file{ada_source_path} file
must contain the location for the GNAT run-time sources (which can simply
be @file{adainclude}). In the same way, the @file{ada_object_path} file must
contain the location for the GNAT run-time objects (which can simply
be @file{adalib}).

You can also specify a new default path to the run-time library at compilation
time with the switch @option{--RTS=rts-path}. You can thus choose / change
the run-time library you want your program to be compiled with. This switch is
recognized by @command{gcc}, @command{gnatmake}, @command{gnatbind},
@command{gnatls}, @command{gnatfind} and @command{gnatxref}.

It is possible to install a library before or after the standard GNAT
library, by reordering the lines in the configuration files. In general, a
library must be installed before the GNAT library if it redefines
any part of it.

@node Using a library
@subsection Using a library

@noindent Once again, the project facility greatly simplifies the use of
libraries. In this context, using a library is just a matter of adding a
@code{with} clause in the user project. For instance, to make use of the
library @code{My_Lib} shown in examples in earlier sections, you can
write:

@smallexample @c projectfile
@b{with} "my_lib";
@b{project} My_Proj @b{is}
  @dots{}
@b{end} My_Proj;
@end smallexample

Even if you have a third-party, non-Ada library, you can still use GNAT's
Project Manager facility to provide a wrapper for it. For example, the
following project, when @code{with}ed by your main project, will link with the
third-party library @file{liba.a}:

@smallexample @c projectfile
@group
@b{project} Liba @b{is}
   @b{for} Externally_Built @b{use} "true";
   @b{for} Source_Files @b{use} ();
   @b{for} Library_Dir @b{use} "lib";
   @b{for} Library_Name @b{use} "a";
   @b{for} Library_Kind @b{use} "static";
@b{end} Liba;
@end group
@end smallexample
This is an alternative to the use of @code{pragma Linker_Options}. It is
especially interesting in the context of systems with several interdependent
static libraries where finding a proper linker order is not easy and best be
left to the tools having visibility over project dependence information.

@noindent
In order to use an Ada library manually, you need to make sure that this
library is on both your source and object path
(see @ref{Search Paths and the Run-Time Library (RTL)}
and @ref{Search Paths for gnatbind}). Furthermore, when the objects are grouped
in an archive or a shared library, you need to specify the desired
library at link time.

For example, you can use the library @file{mylib} installed in
@file{/dir/my_lib_src} and @file{/dir/my_lib_obj} with the following commands:

@smallexample
$ gnatmake -aI/dir/my_lib_src -aO/dir/my_lib_obj my_appl \
  -largs -lmy_lib
@end smallexample

@noindent
This can be expressed more simply:
@smallexample
$ gnatmake my_appl
@end smallexample
@noindent
when the following conditions are met:
@itemize @bullet
@item
@file{/dir/my_lib_src} has been added by the user to the environment
variable @env{ADA_INCLUDE_PATH}, or by the administrator to the file
@file{ada_source_path}
@item
@file{/dir/my_lib_obj} has been added by the user to the environment
variable @env{ADA_OBJECTS_PATH}, or by the administrator to the file
@file{ada_object_path}
@item
a pragma @code{Linker_Options} has been added to one of the sources.
For example:

@smallexample @c ada
@b{pragma} Linker_Options ("-lmy_lib");
@end smallexample
@end itemize

Note that you may also load a library dynamically at
run time given its filename, as illustrated in the GNAT @file{plugins} example
in the directory @file{share/examples/gnat/plugins} within the GNAT
install area.

@node Stand-alone Ada Libraries
@section Stand-alone Ada Libraries
@cindex Stand-alone library, building, using

@menu
* Introduction to Stand-alone Libraries::
* Building a Stand-alone Library::
* Creating a Stand-alone Library to be used in a non-Ada context::
* Restrictions in Stand-alone Libraries::
@end menu

@node Introduction to Stand-alone Libraries
@subsection Introduction to Stand-alone Libraries

@noindent
A Stand-alone Library (abbreviated ``SAL'') is a library that contains the
necessary code to
elaborate the Ada units that are included in the library. In contrast with
an ordinary library, which consists of all sources, objects and @file{ALI}
files of the
library, a SAL may specify a restricted subset of compilation units
to serve as a library interface. In this case, the fully
self-sufficient set of files will normally consist of an objects
archive, the sources of interface units' specs, and the @file{ALI}
files of interface units.
If an interface spec contains a generic unit or an inlined subprogram,
the body's
source must also be provided; if the units that must be provided in the source
form depend on other units, the source and @file{ALI} files of those must
also be provided.

The main purpose of a SAL is to minimize the recompilation overhead of client
applications when a new version of the library is installed. Specifically,
if the interface sources have not changed, client applications do not need to
be recompiled. If, furthermore, a SAL is provided in the shared form and its
version, controlled by @code{Library_Version} attribute, is not changed,
then the clients do not need to be relinked.

SALs also allow the library providers to minimize the amount of library source
text exposed to the clients.  Such ``information hiding'' might be useful or
necessary for various reasons.

Stand-alone libraries are also well suited to be used in an executable whose
main routine is not written in Ada.

@node Building a Stand-alone Library
@subsection Building a Stand-alone Library

@noindent
GNAT's Project facility provides a simple way of building and installing
stand-alone libraries; see @ref{Stand-alone Library Projects}.
To be a Stand-alone Library Project, in addition to the two attributes
that make a project a Library Project (@code{Library_Name} and
@code{Library_Dir}; see @ref{Library Projects}), the attribute
@code{Library_Interface} must be defined.  For example:

@smallexample @c projectfile
@group
   @b{for} Library_Dir @b{use} "lib_dir";
   @b{for} Library_Name @b{use} "dummy";
   @b{for} Library_Interface @b{use} ("int1", "int1.child");
@end group
@end smallexample

@noindent
Attribute @code{Library_Interface} has a non-empty string list value,
each string in the list designating a unit contained in an immediate source
of the project file.

When a Stand-alone Library is built, first the binder is invoked to build
a package whose name depends on the library name
(@file{b~dummy.ads/b} in the example above).
This binder-generated package includes initialization and
finalization procedures whose
names depend on the library name (@code{dummyinit} and @code{dummyfinal}
in the example
above). The object corresponding to this package is included in the library.

You must ensure timely (e.g., prior to any use of interfaces in the SAL)
calling of these procedures if a static SAL is built, or if a shared SAL
is built
with the project-level attribute @code{Library_Auto_Init} set to
@code{"false"}.

For a Stand-Alone Library, only the @file{ALI} files of the Interface Units
(those that are listed in attribute @code{Library_Interface}) are copied to
the Library Directory. As a consequence, only the Interface Units may be
imported from Ada units outside of the library. If other units are imported,
the binding phase will fail.

@noindent
It is also possible to build an encapsulated library where not only
the code to elaborate and finalize the library is embedded but also
ensuring that the library is linked only against static
libraries. So an encapsulated library only depends on system
libraries, all other code, including the GNAT runtime, is embedded. To
build an encapsulated library the attribute
@code{Library_Standalone} must be set to @code{encapsulated}:

@smallexample @c projectfile
@group
   @b{for} Library_Dir @b{use} "lib_dir";
   @b{for} Library_Name @b{use} "dummy";
   @b{for} Library_Kind @b{use} "dynamic";
   @b{for} Library_Interface @b{use} ("int1", "int1.child");
   @b{for} Library_Standalone @b{use} "encapsulated";
@end group
@end smallexample

@noindent
The default value for this attribute is @code{standard} in which case
a stand-alone library is built.

The attribute @code{Library_Src_Dir} may be specified for a
Stand-Alone Library. @code{Library_Src_Dir} is a simple attribute that has a
single string value. Its value must be the path (absolute or relative to the
project directory) of an existing directory. This directory cannot be the
object directory or one of the source directories, but it can be the same as
the library directory. The sources of the Interface
Units of the library that are needed by an Ada client of the library will be
copied to the designated directory, called the Interface Copy directory.
These sources include the specs of the Interface Units, but they may also
include bodies and subunits, when pragmas @code{Inline} or @code{Inline_Always}
are used, or when there is a generic unit in the spec. Before the sources
are copied to the Interface Copy directory, an attempt is made to delete all
files in the Interface Copy directory.

Building stand-alone libraries by hand is somewhat tedious, but for those
occasions when it is necessary here are the steps that you need to perform:
@itemize @bullet
@item
Compile all library sources.

@item
Invoke the binder with the switch @option{-n} (No Ada main program),
with all the @file{ALI} files of the interfaces, and
with the switch @option{-L} to give specific names to the @code{init}
and @code{final} procedures.  For example:
@smallexample
  gnatbind -n int1.ali int2.ali -Lsal1
@end smallexample

@item
Compile the binder generated file:
@smallexample
  gcc -c b~int2.adb
@end smallexample

@item
Link the dynamic library with all the necessary object files,
indicating to the linker the names of the @code{init} (and possibly
@code{final}) procedures for automatic initialization (and finalization).
The built library should be placed in a directory different from
the object directory.

@item
Copy the @code{ALI} files of the interface to the library directory,
add in this copy an indication that it is an interface to a SAL
(i.e., add a word @option{SL} on the line in the @file{ALI} file that starts
with letter ``P'') and make the modified copy of the @file{ALI} file
read-only.
@end itemize

@noindent
Using SALs is not different from using other libraries
(see @ref{Using a library}).

@node Creating a Stand-alone Library to be used in a non-Ada context
@subsection Creating a Stand-alone Library to be used in a non-Ada context

@noindent
It is easy to adapt the SAL build procedure discussed above for use of a SAL in
a non-Ada context.

The only extra step required is to ensure that library interface subprograms
are compatible with the main program, by means of @code{pragma Export}
or @code{pragma Convention}.

Here is an example of simple library interface for use with C main program:

@smallexample @c ada
@b{package} My_Package @b{is}

   @b{procedure} Do_Something;
   @b{pragma} Export (C, Do_Something, "do_something");

   @b{procedure} Do_Something_Else;
   @b{pragma} Export (C, Do_Something_Else, "do_something_else");

@b{end} My_Package;
@end smallexample

@noindent
On the foreign language side, you must provide a ``foreign'' view of the
library interface; remember that it should contain elaboration routines in
addition to interface subprograms.

The example below shows the content of @code{mylib_interface.h} (note
that there is no rule for the naming of this file, any name can be used)
@smallexample
/* the library elaboration procedure */
extern void mylibinit (void);

/* the library finalization procedure */
extern void mylibfinal (void);

/* the interface exported by the library */
extern void do_something (void);
extern void do_something_else (void);
@end smallexample

@noindent
Libraries built as explained above can be used from any program, provided
that the elaboration procedures (named @code{mylibinit} in the previous
example) are called before the library services are used. Any number of
libraries can be used simultaneously, as long as the elaboration
procedure of each library is called.

Below is an example of a C program that uses the @code{mylib} library.

@smallexample
#include "mylib_interface.h"

int
main (void)
@{
   /* First, elaborate the library before using it */
   mylibinit ();

   /* Main program, using the library exported entities */
   do_something ();
   do_something_else ();

   /* Library finalization at the end of the program */
   mylibfinal ();
   return 0;
@}
@end smallexample

@noindent
Note that invoking any library finalization procedure generated by
@code{gnatbind} shuts down the Ada run-time environment.
Consequently, the
finalization of all Ada libraries must be performed at the end of the program.
No call to these libraries or to the Ada run-time library should be made
after the finalization phase.

@node Restrictions in Stand-alone Libraries
@subsection Restrictions in Stand-alone Libraries

@noindent
The pragmas listed below should be used with caution inside libraries,
as they can create incompatibilities with other Ada libraries:
@itemize @bullet
@item pragma @code{Locking_Policy}
@item pragma @code{Partition_Elaboration_Policy}
@item pragma @code{Queuing_Policy}
@item pragma @code{Task_Dispatching_Policy}
@item pragma @code{Unreserve_All_Interrupts}
@end itemize

@noindent
When using a library that contains such pragmas, the user must make sure
that all libraries use the same pragmas with the same values. Otherwise,
@code{Program_Error} will
be raised during the elaboration of the conflicting
libraries. The usage of these pragmas and its consequences for the user
should therefore be well documented.

Similarly, the traceback in the exception occurrence mechanism should be
enabled or disabled in a consistent manner across all libraries.
Otherwise, Program_Error will be raised during the elaboration of the
conflicting libraries.

If the @code{Version} or @code{Body_Version}
attributes are used inside a library, then you need to
perform a @code{gnatbind} step that specifies all @file{ALI} files in all
libraries, so that version identifiers can be properly computed.
In practice these attributes are rarely used, so this is unlikely
to be a consideration.

@node  Rebuilding the GNAT Run-Time Library
@section Rebuilding the GNAT Run-Time Library
@cindex GNAT Run-Time Library, rebuilding
@cindex Building the GNAT Run-Time Library
@cindex Rebuilding the GNAT Run-Time Library
@cindex Run-Time Library, rebuilding

@noindent
It may be useful to recompile the GNAT library in various contexts, the
most important one being the use of partition-wide configuration pragmas
such as @code{Normalize_Scalars}. A special Makefile called
@code{Makefile.adalib} is provided to that effect and can be found in
the directory containing the GNAT library. The location of this
directory depends on the way the GNAT environment has been installed and can
be determined by means of the command:

@smallexample
$ gnatls -v
@end smallexample

@noindent
The last entry in the object search path usually contains the
gnat library. This Makefile contains its own documentation and in
particular the set of instructions needed to rebuild a new library and
to use it.

@node Using the GNU make Utility
@chapter Using the GNU @code{make} Utility
@findex make

@noindent
This chapter offers some examples of makefiles that solve specific
problems. It does not explain how to write a makefile (@pxref{Top,, GNU
make, make, GNU @code{make}}), nor does it try to replace the
@command{gnatmake} utility (@pxref{The GNAT Make Program gnatmake}).

All the examples in this section are specific to the GNU version of
make. Although @command{make} is a standard utility, and the basic language
is the same, these examples use some advanced features found only in
@code{GNU make}.

@menu
* Using gnatmake in a Makefile::
* Automatically Creating a List of Directories::
* Generating the Command Line Switches::
* Overcoming Command Line Length Limits::
@end menu

@node Using gnatmake in a Makefile
@section Using gnatmake in a Makefile
@findex makefile
@cindex GNU make

@noindent
Complex project organizations can be handled in a very powerful way by
using GNU make combined with gnatmake. For instance, here is a Makefile
which allows you to build each subsystem of a big project into a separate
shared library. Such a makefile allows you to significantly reduce the link
time of very big applications while maintaining full coherence at
each step of the build process.

The list of dependencies are handled automatically by
@command{gnatmake}. The Makefile is simply used to call gnatmake in each of
the appropriate directories.

Note that you should also read the example on how to automatically
create the list of directories
(@pxref{Automatically Creating a List of Directories})
which might help you in case your project has a lot of subdirectories.

@smallexample
@iftex
@leftskip=0cm
@font@heightrm=cmr8
@heightrm
@end iftex
## This Makefile is intended to be used with the following directory
## configuration:
##  - The sources are split into a series of csc (computer software components)
##    Each of these csc is put in its own directory.
##    Their name are referenced by the directory names.
##    They will be compiled into shared library (although this would also work
##    with static libraries
##  - The main program (and possibly other packages that do not belong to any
##    csc is put in the top level directory (where the Makefile is).
##       toplevel_dir __ first_csc  (sources) __ lib (will contain the library)
##                    \_ second_csc (sources) __ lib (will contain the library)
##                    \_ @dots{}
## Although this Makefile is build for shared library, it is easy to modify
## to build partial link objects instead (modify the lines with -shared and
## gnatlink below)
##
## With this makefile, you can change any file in the system or add any new
## file, and everything will be recompiled correctly (only the relevant shared
## objects will be recompiled, and the main program will be re-linked).

# The list of computer software component for your project. This might be
# generated automatically.
CSC_LIST=aa bb cc

# Name of the main program (no extension)
MAIN=main

# If we need to build objects with -fPIC, uncomment the following line
#NEED_FPIC=-fPIC

# The following variable should give the directory containing libgnat.so
# You can get this directory through 'gnatls -v'. This is usually the last
# directory in the Object_Path.
GLIB=@dots{}

# The directories for the libraries
# (This macro expands the list of CSC to the list of shared libraries, you
# could simply use the expanded form:
# LIB_DIR=aa/lib/libaa.so bb/lib/libbb.so cc/lib/libcc.so
LIB_DIR=$@{foreach dir,$@{CSC_LIST@},$@{dir@}/lib/lib$@{dir@}.so@}

$@{MAIN@}: objects $@{LIB_DIR@}
    gnatbind $@{MAIN@} $@{CSC_LIST:%=-aO%/lib@} -shared
    gnatlink $@{MAIN@} $@{CSC_LIST:%=-l%@}

objects::
    # recompile the sources
    gnatmake -c -i $@{MAIN@}.adb $@{NEED_FPIC@} $@{CSC_LIST:%=-I%@}

# Note: In a future version of GNAT, the following commands will be simplified
# by a new tool, gnatmlib
$@{LIB_DIR@}:
    mkdir -p $@{dir $@@ @}
    cd $@{dir $@@ @} && gcc -shared -o $@{notdir $@@ @} ../*.o -L$@{GLIB@} -lgnat
    cd $@{dir $@@ @} && cp -f ../*.ali .

# The dependencies for the modules
# Note that we have to force the expansion of *.o, since in some cases
# make won't be able to do it itself.
aa/lib/libaa.so: $@{wildcard aa/*.o@}
bb/lib/libbb.so: $@{wildcard bb/*.o@}
cc/lib/libcc.so: $@{wildcard cc/*.o@}

# Make sure all of the shared libraries are in the path before starting the
# program
run::
    LD_LIBRARY_PATH=`pwd`/aa/lib:`pwd`/bb/lib:`pwd`/cc/lib ./$@{MAIN@}

clean::
    $@{RM@} -rf $@{CSC_LIST:%=%/lib@}
    $@{RM@} $@{CSC_LIST:%=%/*.ali@}
    $@{RM@} $@{CSC_LIST:%=%/*.o@}
    $@{RM@} *.o *.ali $@{MAIN@}
@end smallexample

@node Automatically Creating a List of Directories
@section Automatically Creating a List of Directories

@noindent
In most makefiles, you will have to specify a list of directories, and
store it in a variable. For small projects, it is often easier to
specify each of them by hand, since you then have full control over what
is the proper order for these directories, which ones should be
included.

However, in larger projects, which might involve hundreds of
subdirectories, it might be more convenient to generate this list
automatically.

The example below presents two methods. The first one, although less
general, gives you more control over the list. It involves wildcard
characters, that are automatically expanded by @command{make}. Its
shortcoming is that you need to explicitly specify some of the
organization of your project, such as for instance the directory tree
depth, whether some directories are found in a separate tree, @enddots{}

The second method is the most general one. It requires an external
program, called @command{find}, which is standard on all Unix systems. All
the directories found under a given root directory will be added to the
list.

@smallexample
@iftex
@leftskip=0cm
@font@heightrm=cmr8
@heightrm
@end iftex
# The examples below are based on the following directory hierarchy:
# All the directories can contain any number of files
# ROOT_DIRECTORY ->  a  ->  aa  ->  aaa
#                       ->  ab
#                       ->  ac
#                ->  b  ->  ba  ->  baa
#                       ->  bb
#                       ->  bc
# This Makefile creates a variable called DIRS, that can be reused any time
# you need this list (see the other examples in this section)

# The root of your project's directory hierarchy
ROOT_DIRECTORY=.

####
# First method: specify explicitly the list of directories
# This allows you to specify any subset of all the directories you need.
####

DIRS := a/aa/ a/ab/ b/ba/

####
# Second method: use wildcards
# Note that the argument(s) to wildcard below should end with a '/'.
# Since wildcards also return file names, we have to filter them out
# to avoid duplicate directory names.
# We thus use make's @code{dir} and @code{sort} functions.
# It sets DIRs to the following value (note that the directories aaa and baa
# are not given, unless you change the arguments to wildcard).
# DIRS= ./a/a/ ./b/ ./a/aa/ ./a/ab/ ./a/ac/ ./b/ba/ ./b/bb/ ./b/bc/
####

DIRS := $@{sort $@{dir $@{wildcard $@{ROOT_DIRECTORY@}/*/
                    $@{ROOT_DIRECTORY@}/*/*/@}@}@}

####
# Third method: use an external program
# This command is much faster if run on local disks, avoiding NFS slowdowns.
# This is the most complete command: it sets DIRs to the following value:
# DIRS= ./a ./a/aa ./a/aa/aaa ./a/ab ./a/ac ./b ./b/ba ./b/ba/baa ./b/bb ./b/bc
####

DIRS := $@{shell find $@{ROOT_DIRECTORY@} -type d -print@}

@end smallexample

@node Generating the Command Line Switches
@section Generating the Command Line Switches

@noindent
Once you have created the list of directories as explained in the
previous section (@pxref{Automatically Creating a List of Directories}),
you can easily generate the command line arguments to pass to gnatmake.

For the sake of completeness, this example assumes that the source path
is not the same as the object path, and that you have two separate lists
of directories.

@smallexample
# see "Automatically creating a list of directories" to create
# these variables
SOURCE_DIRS=
OBJECT_DIRS=

GNATMAKE_SWITCHES := $@{patsubst %,-aI%,$@{SOURCE_DIRS@}@}
GNATMAKE_SWITCHES += $@{patsubst %,-aO%,$@{OBJECT_DIRS@}@}

all:
        gnatmake $@{GNATMAKE_SWITCHES@} main_unit
@end smallexample

@node Overcoming Command Line Length Limits
@section Overcoming Command Line Length Limits

@noindent
One problem that might be encountered on big projects is that many
operating systems limit the length of the command line. It is thus hard to give
gnatmake the list of source and object directories.

This example shows how you can set up environment variables, which will
make @command{gnatmake} behave exactly as if the directories had been
specified on the command line, but have a much higher length limit (or
even none on most systems).

It assumes that you have created a list of directories in your Makefile,
using one of the methods presented in
@ref{Automatically Creating a List of Directories}.
For the sake of completeness, we assume that the object
path (where the ALI files are found) is different from the sources patch.

Note a small trick in the Makefile below: for efficiency reasons, we
create two temporary variables (SOURCE_LIST and OBJECT_LIST), that are
expanded immediately by @code{make}. This way we overcome the standard
make behavior which is to expand the variables only when they are
actually used.

On Windows, if you are using the standard Windows command shell, you must
replace colons with semicolons in the assignments to these variables.

@smallexample
@iftex
@leftskip=0cm
@font@heightrm=cmr8
@heightrm
@end iftex
# In this example, we create both ADA_INCLUDE_PATH and ADA_OBJECTS_PATH.
# This is the same thing as putting the -I arguments on the command line.
# (the equivalent of using -aI on the command line would be to define
#  only ADA_INCLUDE_PATH, the equivalent of -aO is ADA_OBJECTS_PATH).
# You can of course have different values for these variables.
#
# Note also that we need to keep the previous values of these variables, since
# they might have been set before running 'make' to specify where the GNAT
# library is installed.

# see "Automatically creating a list of directories" to create these
# variables
SOURCE_DIRS=
OBJECT_DIRS=

empty:=
space:=$@{empty@} $@{empty@}
SOURCE_LIST := $@{subst $@{space@},:,$@{SOURCE_DIRS@}@}
OBJECT_LIST := $@{subst $@{space@},:,$@{OBJECT_DIRS@}@}
ADA_INCLUDE_PATH += $@{SOURCE_LIST@}
ADA_OBJECTS_PATH += $@{OBJECT_LIST@}
export ADA_INCLUDE_PATH
export ADA_OBJECTS_PATH

all:
        gnatmake main_unit
@end smallexample

@node Memory Management Issues
@chapter Memory Management Issues

@noindent
This chapter describes some useful memory pools provided in the GNAT library
and in particular the GNAT Debug Pool facility, which can be used to detect
incorrect uses of access values (including ``dangling references'').
@ifclear FSFEDITION
It also describes the @command{gnatmem} tool, which can be used to track down
``memory leaks''.
@end ifclear

@menu
* Some Useful Memory Pools::
* The GNAT Debug Pool Facility::
@ifclear FSFEDITION
* The gnatmem Tool::
@end ifclear
@end menu

@node Some Useful Memory Pools
@section Some Useful Memory Pools
@findex Memory Pool
@cindex storage, pool

@noindent
The @code{System.Pool_Global} package offers the Unbounded_No_Reclaim_Pool
storage pool. Allocations use the standard system call @code{malloc} while
deallocations use the standard system call @code{free}. No reclamation is
performed when the pool goes out of scope. For performance reasons, the
standard default Ada allocators/deallocators do not use any explicit storage
pools but if they did, they could use this storage pool without any change in
behavior. That is why this storage pool is used  when the user
manages to make the default implicit allocator explicit as in this example:
@smallexample @c ada
   @b{type} T1 @b{is} @b{access} Something;
    --@i{ no Storage pool is defined for T2}
   @b{type} T2 @b{is} @b{access} Something_Else;
   @b{for} T2'Storage_Pool @b{use} T1'Storage_Pool;
   --@i{ the above is equivalent to}
   @b{for} T2'Storage_Pool @b{use} System.Pool_Global.Global_Pool_Object;
@end smallexample

@noindent
The @code{System.Pool_Local} package offers the Unbounded_Reclaim_Pool storage
pool. The allocation strategy is similar to @code{Pool_Local}'s
except that the all
storage allocated with this pool is reclaimed when the pool object goes out of
scope. This pool provides a explicit mechanism similar to the implicit one
provided by several Ada 83 compilers for allocations performed through a local
access type and whose purpose was to reclaim memory when exiting the
scope of a given local access. As an example, the following program does not
leak memory even though it does not perform explicit deallocation:

@smallexample @c ada
@b{with} System.Pool_Local;
@b{procedure} Pooloc1 @b{is}
   @b{procedure} Internal @b{is}
      @b{type} A @b{is} @b{access} Integer;
      X : System.Pool_Local.Unbounded_Reclaim_Pool;
      @b{for} A'Storage_Pool @b{use} X;
      v : A;
   @b{begin}
      @b{for} I @b{in}  1 .. 50 @b{loop}
         v := @b{new} Integer;
      @b{end} @b{loop};
   @b{end} Internal;
@b{begin}
   @b{for} I @b{in}  1 .. 100 @b{loop}
      Internal;
   @b{end} @b{loop};
@b{end} Pooloc1;
@end smallexample

@noindent
The @code{System.Pool_Size} package implements the Stack_Bounded_Pool used when
@code{Storage_Size} is specified for an access type.
The whole storage for the pool is
allocated at once, usually on the stack at the point where the access type is
elaborated. It is automatically reclaimed when exiting the scope where the
access type is defined. This package is not intended to be used directly by the
user and it is implicitly used for each such declaration:

@smallexample @c ada
   @b{type} T1 @b{is} @b{access} Something;
   @b{for} T1'Storage_Size @b{use} 10_000;
@end smallexample

@node The GNAT Debug Pool Facility
@section The GNAT Debug Pool Facility
@findex Debug Pool
@cindex storage, pool, memory corruption

@noindent
The use of unchecked deallocation and unchecked conversion can easily
lead to incorrect memory references. The problems generated by such
references are usually difficult to tackle because the symptoms can be
very remote from the origin of the problem. In such cases, it is
very helpful to detect the problem as early as possible. This is the
purpose of the Storage Pool provided by @code{GNAT.Debug_Pools}.

In order to use the GNAT specific debugging pool, the user must
associate a debug pool object with each of the access types that may be
related to suspected memory problems. See Ada Reference Manual 13.11.
@smallexample @c ada
@b{type} Ptr @b{is} @b{access} Some_Type;
Pool : GNAT.Debug_Pools.Debug_Pool;
@b{for} Ptr'Storage_Pool @b{use} Pool;
@end smallexample

@noindent
@code{GNAT.Debug_Pools} is derived from a GNAT-specific kind of
pool: the @code{Checked_Pool}. Such pools, like standard Ada storage pools,
allow the user to redefine allocation and deallocation strategies. They
also provide a checkpoint for each dereference, through the use of
the primitive operation @code{Dereference} which is implicitly called at
each dereference of an access value.

Once an access type has been associated with a debug pool, operations on
values of the type may raise four distinct exceptions,
which correspond to four potential kinds of memory corruption:
@itemize @bullet
@item
@code{GNAT.Debug_Pools.Accessing_Not_Allocated_Storage}
@item
@code{GNAT.Debug_Pools.Accessing_Deallocated_Storage}
@item
@code{GNAT.Debug_Pools.Freeing_Not_Allocated_Storage}
@item
@code{GNAT.Debug_Pools.Freeing_Deallocated_Storage }
@end itemize

@noindent
For types associated with a Debug_Pool, dynamic allocation is performed using
the standard GNAT allocation routine. References to all allocated chunks of
memory are kept in an internal dictionary. Several deallocation strategies are
provided, whereupon the user can choose to release the memory to the system,
keep it allocated for further invalid access checks, or fill it with an easily
recognizable pattern for debug sessions. The memory pattern is the old IBM
hexadecimal convention: @code{16#DEADBEEF#}.

See the documentation in the file g-debpoo.ads for more information on the
various strategies.

Upon each dereference, a check is made that the access value denotes a
properly allocated memory location. Here is a complete example of use of
@code{Debug_Pools}, that includes typical instances of  memory corruption:
@smallexample @c ada
@iftex
@leftskip=0cm
@end iftex
@b{with} Gnat.Io; @b{use} Gnat.Io;
@b{with} Unchecked_Deallocation;
@b{with} Unchecked_Conversion;
@b{with} GNAT.Debug_Pools;
@b{with} System.Storage_Elements;
@b{with} Ada.Exceptions; @b{use} Ada.Exceptions;
@b{procedure} Debug_Pool_Test @b{is}

   @b{type} T @b{is} @b{access} Integer;
   @b{type} U @b{is} @b{access} @b{all} T;

   P : GNAT.Debug_Pools.Debug_Pool;
   @b{for} T'Storage_Pool @b{use} P;

   @b{procedure} Free @b{is} @b{new} Unchecked_Deallocation (Integer, T);
   @b{function} UC @b{is} @b{new} Unchecked_Conversion (U, T);
   A, B : @b{aliased} T;

   @b{procedure} Info @b{is} @b{new} GNAT.Debug_Pools.Print_Info(Put_Line);

@b{begin}
   Info (P);
   A := @b{new} Integer;
   B := @b{new} Integer;
   B := A;
   Info (P);
   Free (A);
   @b{begin}
      Put_Line (Integer'Image(B.@b{all}));
   @b{exception}
      @b{when} E : @b{others} => Put_Line ("raised: " & Exception_Name (E));
   @b{end};
   @b{begin}
      Free (B);
   @b{exception}
      @b{when} E : @b{others} => Put_Line ("raised: " & Exception_Name (E));
   @b{end};
   B := UC(A'Access);
   @b{begin}
      Put_Line (Integer'Image(B.@b{all}));
   @b{exception}
      @b{when} E : @b{others} => Put_Line ("raised: " & Exception_Name (E));
   @b{end};
   @b{begin}
      Free (B);
   @b{exception}
      @b{when} E : @b{others} => Put_Line ("raised: " & Exception_Name (E));
   @b{end};
   Info (P);
@b{end} Debug_Pool_Test;
@end smallexample

@noindent
The debug pool mechanism provides the following precise diagnostics on the
execution of this erroneous program:
@smallexample
Debug Pool info:
  Total allocated bytes :  0
  Total deallocated bytes :  0
  Current Water Mark:  0
  High Water Mark:  0

Debug Pool info:
  Total allocated bytes :  8
  Total deallocated bytes :  0
  Current Water Mark:  8
  High Water Mark:  8

raised: GNAT.DEBUG_POOLS.ACCESSING_DEALLOCATED_STORAGE
raised: GNAT.DEBUG_POOLS.FREEING_DEALLOCATED_STORAGE
raised: GNAT.DEBUG_POOLS.ACCESSING_NOT_ALLOCATED_STORAGE
raised: GNAT.DEBUG_POOLS.FREEING_NOT_ALLOCATED_STORAGE
Debug Pool info:
  Total allocated bytes :  8
  Total deallocated bytes :  4
  Current Water Mark:  4
  High Water Mark:  8
@end smallexample

@ifclear FSFEDITION
@node The gnatmem Tool
@section The @command{gnatmem} Tool
@findex gnatmem

@noindent
The @code{gnatmem} utility monitors dynamic allocation and
deallocation activity in a program, and displays information about
incorrect deallocations and possible sources of memory leaks.
It is designed to work in association with a static runtime library
only and in this context provides three types of information:
@itemize @bullet
@item
General information concerning memory management, such as the total
number of allocations and deallocations, the amount of allocated
memory and the high water mark, i.e.@: the largest amount of allocated
memory in the course of program execution.

@item
Backtraces for all incorrect deallocations, that is to say deallocations
which do not correspond to a valid allocation.

@item
Information on each allocation that is potentially the origin of a memory
leak.
@end itemize

@menu
* Running gnatmem::
* Switches for gnatmem::
* Example of gnatmem Usage::
@end menu

@node Running gnatmem
@subsection Running @code{gnatmem}

@noindent
@code{gnatmem} makes use of the output created by the special version of
allocation and deallocation routines that record call information. This allows
it to obtain accurate dynamic memory usage history at a minimal cost to the
execution speed. Note however, that @code{gnatmem} is not supported on all
platforms (currently, it is supported on AIX, HP-UX, GNU/Linux, Solaris and
Windows NT/2000/XP (x86).

@noindent
The @code{gnatmem} command has the form

@smallexample
@c    $ gnatmem @ovar{switches} user_program
@c Expanding @ovar macro inline (explanation in macro def comments)
      $ gnatmem @r{[}@var{switches}@r{]} @var{user_program}
@end smallexample

@noindent
The program must have been linked with the instrumented version of the
allocation and deallocation routines. This is done by linking with the
@file{libgmem.a} library. For correct symbolic backtrace information,
the user program should be compiled with debugging options
(see @ref{Switches for gcc}). For example to build @file{my_program}:

@smallexample
$ gnatmake -g my_program -largs -lgmem
@end smallexample

@noindent
As library @file{libgmem.a} contains an alternate body for package
@code{System.Memory}, @file{s-memory.adb} should not be compiled and linked
when an executable is linked with library @file{libgmem.a}. It is then not
recommended to use @command{gnatmake} with switch @option{-a}.

@noindent
When @file{my_program} is executed, the file @file{gmem.out} is produced.
This file contains information about all allocations and deallocations
performed by the program. It is produced by the instrumented allocations and
deallocations routines and will be used by @code{gnatmem}.

In order to produce symbolic backtrace information for allocations and
deallocations performed by the GNAT run-time library, you need to use a
version of that library that has been compiled with the @option{-g} switch
(see @ref{Rebuilding the GNAT Run-Time Library}).

Gnatmem must be supplied with the @file{gmem.out} file and the executable to
examine. If the location of @file{gmem.out} file was not explicitly supplied by
@option{-i} switch, gnatmem will assume that this file can be found in the
current directory. For example, after you have executed @file{my_program},
@file{gmem.out} can be analyzed by @code{gnatmem} using the command:

@smallexample
$ gnatmem my_program
@end smallexample

@noindent
This will produce the output with the following format:

*************** debut cc
@smallexample
$ gnatmem my_program

Global information
------------------
   Total number of allocations        :  45
   Total number of deallocations      :   6
   Final Water Mark (non freed mem)   :  11.29 Kilobytes
   High Water Mark                    :  11.40 Kilobytes

.
.
.
Allocation Root # 2
-------------------
 Number of non freed allocations    :  11
 Final Water Mark (non freed mem)   :   1.16 Kilobytes
 High Water Mark                    :   1.27 Kilobytes
 Backtrace                          :
   my_program.adb:23 my_program.alloc
.
.
.
@end smallexample

The first block of output gives general information. In this case, the
Ada construct ``@code{@b{new}}'' was executed 45 times, and only 6 calls to an
Unchecked_Deallocation routine occurred.

@noindent
Subsequent paragraphs display  information on all allocation roots.
An allocation root is a specific point in the execution of the program
that generates some dynamic allocation, such as a ``@code{@b{new}}''
construct. This root is represented by an execution backtrace (or subprogram
call stack). By default the backtrace depth for allocations roots is 1, so
that a root corresponds exactly to a source location. The backtrace can
be made deeper, to make the root more specific.

@node Switches for gnatmem
@subsection Switches for @code{gnatmem}

@noindent
@code{gnatmem} recognizes the following switches:

@table @option

@item -q
@cindex @option{-q} (@code{gnatmem})
Quiet. Gives the minimum output needed to identify the origin of the
memory leaks. Omits statistical information.

@item @var{N}
@cindex @var{N} (@code{gnatmem})
N is an integer literal (usually between 1 and 10) which controls the
depth of the backtraces defining allocation root. The default value for
N is 1. The deeper the backtrace, the more precise the localization of
the root. Note that the total number of roots can depend on this
parameter. This parameter must be specified @emph{before} the name of the
executable to be analyzed, to avoid ambiguity.

@item -b n
@cindex @option{-b} (@code{gnatmem})
This switch has the same effect as just depth parameter.

@item -i @var{file}
@cindex @option{-i} (@code{gnatmem})
Do the @code{gnatmem} processing starting from @file{file}, rather than
@file{gmem.out} in the current directory.

@item -m n
@cindex @option{-m} (@code{gnatmem})
This switch causes @code{gnatmem} to mask the allocation roots that have less
than n leaks.  The default value is 1. Specifying the value of 0 will allow
examination of even the roots that did not result in leaks.

@item -s order
@cindex @option{-s} (@code{gnatmem})
This switch causes @code{gnatmem} to sort the allocation roots according to the
specified order of sort criteria, each identified by a single letter. The
currently supported criteria are @code{n, h, w} standing respectively for
number of unfreed allocations, high watermark, and final watermark
corresponding to a specific root. The default order is @code{nwh}.

@item -t
@cindex @option{-t} (@code{gnatmem})
This switch causes memory allocated size to be always output in bytes.
Default @code{gnatmem} behavior is to show memory sizes less then 1 kilobyte
in bytes, from 1 kilobyte till 1 megabyte in kilobytes and the rest in
megabytes.

@end table

@node Example of gnatmem Usage
@subsection Example of @code{gnatmem} Usage

@noindent
The following example shows the use of @code{gnatmem}
on a simple memory-leaking program.
Suppose that we have the following Ada program:

@smallexample @c ada
@group
@cartouche
@b{with} Unchecked_Deallocation;
@b{procedure} Test_Gm @b{is}

   @b{type} T @b{is} @b{array} (1..1000) @b{of} Integer;
   @b{type} Ptr @b{is} @b{access} T;
   @b{procedure} Free @b{is} @b{new} Unchecked_Deallocation (T, Ptr);
   A : Ptr;

   @b{procedure} My_Alloc @b{is}
   @b{begin}
      A := @b{new} T;
   @b{end} My_Alloc;

   @b{procedure} My_DeAlloc @b{is}
      B : Ptr := A;
   @b{begin}
      Free (B);
   @b{end} My_DeAlloc;

@b{begin}
   My_Alloc;
   @b{for} I @b{in} 1 .. 5 @b{loop}
      @b{for} J @b{in} I .. 5 @b{loop}
         My_Alloc;
      @b{end} @b{loop};
      My_Dealloc;
   @b{end} @b{loop};
@b{end};
@end cartouche
@end group
@end smallexample

@noindent
The program needs to be compiled with debugging option and linked with
@code{gmem} library:

@smallexample
$ gnatmake -g test_gm -largs -lgmem
@end smallexample

@noindent
Then we execute the program as usual:

@smallexample
$ test_gm
@end smallexample

@noindent
Then @code{gnatmem} is invoked simply with
@smallexample
$ gnatmem test_gm
@end smallexample

@noindent
which produces the following output (result may vary on different platforms):

@smallexample
Global information
------------------
   Total number of allocations        :  18
   Total number of deallocations      :   5
   Final Water Mark (non freed mem)   :  53.00 Kilobytes
   High Water Mark                    :  56.90 Kilobytes

Allocation Root # 1
-------------------
 Number of non freed allocations    :  11
 Final Water Mark (non freed mem)   :  42.97 Kilobytes
 High Water Mark                    :  46.88 Kilobytes
 Backtrace                          :
   test_gm.adb:11 test_gm.my_alloc

Allocation Root # 2
-------------------
 Number of non freed allocations    :   1
 Final Water Mark (non freed mem)   :  10.02 Kilobytes
 High Water Mark                    :  10.02 Kilobytes
 Backtrace                          :
   s-secsta.adb:81 system.secondary_stack.ss_init

Allocation Root # 3
-------------------
 Number of non freed allocations    :   1
 Final Water Mark (non freed mem)   :  12 Bytes
 High Water Mark                    :  12 Bytes
 Backtrace                          :
   s-secsta.adb:181 system.secondary_stack.ss_init
@end smallexample

@noindent
Note that the GNAT run time contains itself a certain number of
allocations that have no  corresponding deallocation,
as shown here for root #2 and root
#3. This is a normal behavior when the number of non-freed allocations
is one, it allocates dynamic data structures that the run time needs for
the complete lifetime of the program. Note also that there is only one
allocation root in the user program with a single line back trace:
test_gm.adb:11 test_gm.my_alloc, whereas a careful analysis of the
program shows that 'My_Alloc' is called at 2 different points in the
source (line 21 and line 24). If those two allocation roots need to be
distinguished, the backtrace depth parameter can be used:

@smallexample
$ gnatmem 3 test_gm
@end smallexample

@noindent
which will give the following output:

@smallexample
Global information
------------------
   Total number of allocations        :  18
   Total number of deallocations      :   5
   Final Water Mark (non freed mem)   :  53.00 Kilobytes
   High Water Mark                    :  56.90 Kilobytes

Allocation Root # 1
-------------------
 Number of non freed allocations    :  10
 Final Water Mark (non freed mem)   :  39.06 Kilobytes
 High Water Mark                    :  42.97 Kilobytes
 Backtrace                          :
   test_gm.adb:11 test_gm.my_alloc
   test_gm.adb:24 test_gm
   b_test_gm.c:52 main

Allocation Root # 2
-------------------
 Number of non freed allocations    :   1
 Final Water Mark (non freed mem)   :  10.02 Kilobytes
 High Water Mark                    :  10.02 Kilobytes
 Backtrace                          :
   s-secsta.adb:81  system.secondary_stack.ss_init
   s-secsta.adb:283 <system__secondary_stack___elabb>
   b_test_gm.c:33   adainit

Allocation Root # 3
-------------------
 Number of non freed allocations    :   1
 Final Water Mark (non freed mem)   :   3.91 Kilobytes
 High Water Mark                    :   3.91 Kilobytes
 Backtrace                          :
   test_gm.adb:11 test_gm.my_alloc
   test_gm.adb:21 test_gm
   b_test_gm.c:52 main

Allocation Root # 4
-------------------
 Number of non freed allocations    :   1
 Final Water Mark (non freed mem)   :  12 Bytes
 High Water Mark                    :  12 Bytes
 Backtrace                          :
   s-secsta.adb:181 system.secondary_stack.ss_init
   s-secsta.adb:283 <system__secondary_stack___elabb>
   b_test_gm.c:33   adainit
@end smallexample

@noindent
The allocation root #1 of the first example has been split in 2 roots #1
and #3 thanks to the more precise associated backtrace.
@end ifclear

@node Stack Related Facilities
@chapter Stack Related Facilities

@noindent
This chapter describes some useful tools associated with stack
checking and analysis. In
particular, it deals with dynamic and static stack usage measurements.

@menu
* Stack Overflow Checking::
* Static Stack Usage Analysis::
* Dynamic Stack Usage Analysis::
@end menu

@node Stack Overflow Checking
@section Stack Overflow Checking
@cindex Stack Overflow Checking
@cindex -fstack-check

@noindent
For most operating systems, @command{gcc} does not perform stack overflow
checking by default. This means that if the main environment task or
some other task exceeds the available stack space, then unpredictable
behavior will occur. Most native systems offer some level of protection by
adding a guard page at the end of each task stack. This mechanism is usually
not enough for dealing properly with stack overflow situations because
a large local variable could ``jump'' above the guard page.
Furthermore, when the
guard page is hit, there may not be any space left on the stack for executing
the exception propagation code. Enabling stack checking avoids
such situations.

To activate stack checking, compile all units with the gcc option
@option{-fstack-check}. For example:

@smallexample
gcc -c -fstack-check package1.adb
@end smallexample

@noindent
Units compiled with this option will generate extra instructions to check
that any use of the stack (for procedure calls or for declaring local
variables in declare blocks) does not exceed the available stack space.
If the space is exceeded, then a @code{Storage_Error} exception is raised.

For declared tasks, the stack size is controlled by the size
given in an applicable @code{Storage_Size} pragma or by the value specified
at bind time with @option{-d} (@pxref{Switches for gnatbind}) or is set to
the default size as defined in the GNAT runtime otherwise.

For the environment task, the stack size depends on
system defaults and is unknown to the compiler. Stack checking
may still work correctly if a fixed
size stack is allocated, but this cannot be guaranteed.
To ensure that a clean exception is signalled for stack
overflow, set the environment variable
@env{GNAT_STACK_LIMIT} to indicate the maximum
stack area that can be used, as in:
@cindex GNAT_STACK_LIMIT

@smallexample
SET GNAT_STACK_LIMIT 1600
@end smallexample

@noindent
The limit is given in kilobytes, so the above declaration would
set the stack limit of the environment task to 1.6 megabytes.
Note that the only purpose of this usage is to limit the amount
of stack used by the environment task. If it is necessary to
increase the amount of stack for the environment task, then this
is an operating systems issue, and must be addressed with the
appropriate operating systems commands.

@node Static Stack Usage Analysis
@section Static Stack Usage Analysis
@cindex Static Stack Usage Analysis
@cindex -fstack-usage

@noindent
A unit compiled with @option{-fstack-usage} will generate an extra file
that specifies
the maximum amount of stack used, on a per-function basis.
The file has the same
basename as the target object file with a @file{.su} extension.
Each line of this file is made up of three fields:

@itemize
@item
The name of the function.
@item
A number of bytes.
@item
One or more qualifiers: @code{static}, @code{dynamic}, @code{bounded}.
@end itemize

The second field corresponds to the size of the known part of the function
frame.

The qualifier @code{static} means that the function frame size
is purely static.
It usually means that all local variables have a static size.
In this case, the second field is a reliable measure of the function stack
utilization.

The qualifier @code{dynamic} means that the function frame size is not static.
It happens mainly when some local variables have a dynamic size. When this
qualifier appears alone, the second field is not a reliable measure
of the function stack analysis. When it is qualified with  @code{bounded}, it
means that the second field is a reliable maximum of the function stack
utilization.

A unit compiled with @option{-Wstack-usage} will issue a warning for each
subprogram whose stack usage might be larger than the specified amount of
bytes.  The wording is in keeping with the qualifier documented above.

@node Dynamic Stack Usage Analysis
@section Dynamic Stack Usage Analysis

@noindent
It is possible to measure the maximum amount of stack used by a task, by
adding a switch to @command{gnatbind}, as:

@smallexample
$ gnatbind -u0 file
@end smallexample

@noindent
With this option, at each task termination, its stack usage is  output on
@file{stderr}.
It is not always convenient to output the stack usage when the program
is still running. Hence, it is possible to delay this output until program
termination. for a given number of tasks specified as the argument of the
@option{-u} option. For instance:

@smallexample
$ gnatbind -u100 file
@end smallexample

@noindent
will buffer the stack usage information of the first 100 tasks to terminate and
output this info at program termination. Results are displayed in four
columns:

@noindent
Index | Task Name | Stack Size | Stack Usage

@noindent
where:

@table @emph
@item Index
is a number associated with each task.

@item Task Name
is the name of the task analyzed.

@item Stack Size
is the maximum size for the stack.

@item Stack Usage
is the measure done by the stack analyzer. In order to prevent overflow, the stack
is not entirely analyzed, and it's not possible to know exactly how
much has actually been used.

@end table

@noindent
The environment task stack, e.g., the stack that contains the main unit, is
only processed when the environment variable GNAT_STACK_LIMIT is set.

@noindent
The package @code{GNAT.Task_Stack_Usage} provides facilities to get
stack usage reports at run-time. See its body for the details.

@ifclear FSFEDITION
@c *********************************
@c *            GNATCHECK          *
@c *********************************
@node Verifying Properties with gnatcheck
@chapter Verifying Properties with @command{gnatcheck}
@findex gnatcheck
@cindex @command{gnatcheck}

@noindent
The @command{gnatcheck} tool is an ASIS-based utility that checks properties
of Ada source files according to a given set of semantic rules.
@cindex ASIS

In order to check compliance with a given rule, @command{gnatcheck} has to
semantically analyze the Ada sources.
Therefore, checks can only be performed on
legal Ada units. Moreover, when a unit depends semantically upon units located
outside the current directory, the source search path has to be provided when
calling @command{gnatcheck}, either through a specified project file or
through @command{gnatcheck} switches.

For full details, refer to @cite{GNATcheck Reference Manual} document.
@end ifclear

@ifclear FSFEDITION
@c *********************************
@node Creating Sample Bodies with gnatstub
@chapter Creating Sample Bodies with @command{gnatstub}
@findex gnatstub

@noindent
@command{gnatstub} creates empty but compilable bodies
for library unit declarations and empty but compilable
subunit for body stubs.

To create a body or a subunit, @command{gnatstub} invokes the Ada
compiler and generates and uses the ASIS tree for the input source;
thus the input must be legal Ada code, and the tool should have all the
information needed to compile the input source. To provide this information,
you may specify as a tool parameter the project file the input source belongs to
(or you may call @command{gnatstub}
through the @command{gnat} driver (see @ref{The GNAT Driver and
Project Files}). Another possibility is to specify the source search
path and needed configuration files in @option{-cargs} section of @command{gnatstub}
call, see the description of the @command{gnatstub} switches below.

If the @command{gnatstub} argument source contains preprocessing directives
then the needed options should be provided to run preprocessor as a part of
the @command{gnatstub} call, and the generated body stub will correspond to
the preprocessed source.

By default, all the program unit bodies generated by @code{gnatstub}
raise the predefined @code{Program_Error} exception, which will catch
accidental calls of generated stubs. This behavior can be changed with
option @option{--no-exception} (see below).

@menu
* Running gnatstub::
* Switches for gnatstub::
@end menu

@node Running gnatstub
@section Running @command{gnatstub}

@noindent
@command{gnatstub} has a command-line interface of the form:

@smallexample
@c $ gnatstub @ovar{switches} @var{filename}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatstub @r{[}@var{switches}@r{]} @var{filename} @r{[}-cargs @var{gcc_switches}@r{]}
@end smallexample

@noindent
where
@table @var
@item filename
is the name of the source file that contains a library unit declaration
for which a body must be created or a library unit body for which subunits
must be created for the body stubs declared in this body.
The file name may contain the path information.
If the name does not follow GNAT file naming conventions and a set
of seitches does not contain a project file that defines naming
conventions, the name of the body file must
be provided
explicitly as the value of the @option{-o@var{body-name}} option.
If the file name follows the GNAT file naming
conventions and the name of the body file is not provided,
@command{gnatstub}
takes the naming conventions for the generated source from the
project file provided as a parameter of @option{-P} switch if any,
or creates the name file to generate using the standard GNAT
naming conventions.

@item @samp{@var{gcc_switches}} is a list of switches for
@command{gcc}. They will be passed on to all compiler invocations made by
@command{gnatstub} to generate the ASIS trees. Here you can provide
@option{-I} switches to form the source search path,
use the @option{-gnatec} switch to set the configuration file,
use the @option{-gnat05} switch if sources should be compiled in
Ada 2005 mode etc.

@item switches
is an optional sequence of switches as described in the next section
@end table

@node Switches for gnatstub
@section Switches for @command{gnatstub}

@table @option
@c !sort!

@item --version
@cindex @option{--version} @command{gnatstub}
Display Copyright and version, then exit disregarding all other options.

@item --help
@cindex @option{--help} @command{gnatstub}
Display usage, then exit disregarding all other options.

@item -P @var{file}
@cindex @option{-P} @command{gnatstub}
Indicates the name of the project file that describes the set of sources
to be processed.

@item -X@var{name}=@var{value}
@cindex @option{-X} @command{gnatstub}
Indicates that external variable @var{name} in the argument project
has the value @var{value}. Has no effect if no project is specified as
tool argument.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gnatstub})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).

@item --subunits
@cindex @option{--subunits} (@command{gnatstub})
Generate subunits for body stubs. If this switch is specified,
@command{gnatstub} expects a library unit body as an agrument file,
otherwise a library unit declaration is expected. If a body stub
already has a corresponding subunit, @command{gnatstub} does not
generate anything for it.

@item -f
@cindex @option{-f} (@command{gnatstub})
If the destination directory already contains a file with the name of the
body file
for the argument spec file, replace it with the generated body stub.
This switch cannot be used together with @option{--subunits}.

@item -hs
@cindex @option{-hs} (@command{gnatstub})
Put the comment header (i.e., all the comments preceding the
compilation unit) from the source of the library unit declaration
into the body stub.

@item -hg
@cindex @option{-hg} (@command{gnatstub})
Put a sample comment header into the body stub.

@item --header-file=@var{filename}
@cindex @option{--header-file} (@command{gnatstub})
Use the content of the file as the comment header for a generated body stub.

@item -IDIR
@cindex @option{-IDIR} (@command{gnatstub})
@itemx -I-
@cindex @option{-I-} (@command{gnatstub})
These switches have  the same meaning as in calls to
@command{gcc}.
They define  the source search path in the call to
@command{gcc} issued
by @command{gnatstub} to compile an argument source file.

@item -gnatec@var{PATH}
@cindex @option{-gnatec} (@command{gnatstub})
This switch has the same meaning as in calls to @command{gcc}.
It defines the additional configuration file to be passed to the call to
@command{gcc} issued
by @command{gnatstub} to compile an argument source file.

@item -gnatyM@var{n}
@cindex @option{-gnatyM} (@command{gnatstub})
(@var{n} is a non-negative integer). Set the maximum line length that is
allowed in a source file. The default is 79. The maximum value that can be
specified is 32767. Note that in the special case of configuration
pragma files, the maximum is always 32767 regardless of whether or
not this switch appears.

@item -gnaty@var{n}
@cindex @option{-gnaty} (@command{gnatstub})
(@var{n} is a non-negative integer from 1 to 9). Set the indentation level in
the generated body sample to @var{n}.
The default indentation is 3.

@item -gnatyo
@cindex @option{-gnatyo} (@command{gnatstub})
Order local bodies alphabetically. (By default local bodies are ordered
in the same way as the corresponding local specs in the argument spec file.)

@item -i@var{n}
@cindex @option{-i} (@command{gnatstub})
Same as @option{-gnaty@var{n}}

@item -k
@cindex @option{-k} (@command{gnatstub})
Do not remove the tree file (i.e., the snapshot of the compiler internal
structures used by @command{gnatstub}) after creating the body stub.

@item -l@var{n}
@cindex @option{-l} (@command{gnatstub})
Same as @option{-gnatyM@var{n}}

@item --no-exception
@cindex @option{--no-exception} (@command{gnatstub})
Avoid raising PROGRAM_ERROR in the generated bodies of program unit stubs.
This is not always possible for function stubs.

@item --no-local-header
@cindex @option{--no-local-header} (@command{gnatstub})
Do not place local comment header with unit name before body stub for a
unit.

@item -o @var{body-name}
@cindex @option{-o} (@command{gnatstub})
Body file name.  This should be set if the argument file name does not
follow
the GNAT file naming
conventions. If this switch is omitted the default name for the body will be
obtained
from the argument file name according to the GNAT file naming conventions.

@item --dir=@var{dir-name}
@cindex @option{--dir} (@command{gnatstub})
The path to the directory to place the generated files into.
If this switch is not set, the generated library unit body is
placed in the current directory, and generated sununits -
in the directory where the argument body is located.

@item -W@var{e}
@cindex @option{-W} (@command{gnatstub})
Specify the wide character encoding method for the output body file.
@var{e} is one of the following:

@itemize @bullet

@item h
Hex encoding

@item u
Upper half encoding

@item s
Shift/JIS encoding

@item e
EUC encoding

@item 8
UTF-8 encoding

@item b
Brackets encoding (default value)
@end itemize

@item -q
@cindex @option{-q} (@command{gnatstub})
Quiet mode: do not generate a confirmation when a body is
successfully created, and do not generate a message when a body is not
required for an
argument unit.

@item -r
@cindex @option{-r} (@command{gnatstub})
Reuse the tree file (if it exists) instead of creating it.  Instead of
creating the tree file for the library unit declaration, @command{gnatstub}
tries to find it in the current directory and use it for creating
a body. If the tree file is not found, no body is created. This option
also implies @option{-k}, whether or not
the latter is set explicitly.

@item -t
@cindex @option{-t} (@command{gnatstub})
Overwrite the existing tree file.  If the current directory already
contains the file which, according to the GNAT file naming rules should
be considered as a tree file for the argument source file,
@command{gnatstub}
will refuse to create the tree file needed to create a sample body
unless this option is set.

@item -v
@cindex @option{-v} (@command{gnatstub})
Verbose mode: generate version information.

@end table
@end ifclear

@ifclear FSFEDITION
@c *********************************
@node Creating Unit Tests with gnattest
@chapter Creating Unit Tests with @command{gnattest}
@findex gnattest

@noindent
@command{gnattest} is an ASIS-based utility that creates unit-test skeletons
as well as a test driver infrastructure (harness). @command{gnattest} creates
a skeleton for each visible subprogram in the packages under consideration when
they do not exist already.

The user can choose to generate a single test driver
that will run all individual tests, or separate test drivers for each test. The
second option allows much greater flexibility in test execution environment,
allows to benefit from parallel tests execution to increase performance, and
provides stubbing support.

@command{gnattest} also has a mode of operation where it acts as the test
aggregator when multiple test executables must be run, in particular when
the separate test drivers were generated. In this mode it handles individual
tests execution and upon completion reports the summary results of the test
run.

In order to process source files from a project, @command{gnattest} has to
semantically analyze the sources. Therefore, test skeletons can only be
generated for legal Ada units. If a unit is dependent on other units,
those units should be among the source files of the project or of other projects
imported by this one.

Generated skeletons and harnesses are based on the AUnit testing framework.
AUnit is an Ada adaptation of the xxxUnit testing frameworks, similar to JUnit
for Java or CppUnit for C++. While it is advised that gnattest users read
the AUnit manual, deep knowledge of AUnit is not necessary for using gnattest.
For correct operation of @command{gnattest}, AUnit should be installed and
aunit.gpr must be on the project path. This happens automatically when Aunit
is installed at its default location.

@menu
* Running gnattest::
* Switches for gnattest in framework generation mode::
* Switches for gnattest in tests execution mode::
* Project Attributes for gnattest::
* Simple Example::
* Setting Up and Tearing Down the Testing Environment::
* Regenerating Tests::
* Default Test Behavior::
* Testing Primitive Operations of Tagged Types::
* Testing Inheritance::
* Tagged Types Substitutability Testing::
* Testing with Contracts::
* Additional Tests::
* Individual Test Drivers::
* Stubbing::
* Putting Tests under Version Control::
* Support for other platforms/run-times::
* Current Limitations::
@end menu

@node Running gnattest
@section Running @command{gnattest}

@noindent
@b{In the framework generation mode}, @command{gnattest} has a command-line
interface of the form

@smallexample
@c $ gnattest @var{-Pprojname} @ovar{switches} @ovar{filename} @ovar{directory}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnattest @var{-Pprojname} @r{[}@var{--harness-dir=dirname}@r{]} @r{[}@var{switches}@r{]} @r{[}@var{filename}@r{]} @r{[}-cargs @var{gcc_switches}@r{]}
@end smallexample

@noindent
where
@table @var

@item -Pprojname
specifies the project defining the location of source files. When no
file names are provided on the command line, all sources in the project
are used as input. This switch is required.

@item filename
is the name of the source file containing the library unit package declaration
for which a test package will be created. The file name may be given with a
path.

@item @samp{@var{gcc_switches}}
is a list of switches for
@command{gcc}. These switches will be passed on to all compiler invocations
made by @command{gnattest} to generate a set of ASIS trees. Here you can provide
@option{-I} switches to form the source search path,
use the @option{-gnatec} switch to set the configuration file,
use the @option{-gnat05} switch if sources should be compiled in
Ada 2005 mode, etc.

@item switches
is an optional sequence of switches as described in the next section.

@end table

@command{gnattest} results can be found in two different places.

@itemize @bullet
@item automatic harness:
the harness code, which is located by default in "gnattest/harness" directory
that is created in the object directory of corresponding project file. All of
this code is generated completely automatically and can be destroyed and
regenerated at will. It is not recommended to modify this code manually, since
it could easily be overridden by mistake. The entry point in the harness code is
the project file named @command{test_driver.gpr}. Tests can be compiled and run
using a command such as:

@smallexample
gnatmake -P<harness-dir>/test_driver
test_runner
@end smallexample

Note that you might need to specify the necessary values of scenario variables
when you are not using the AUnit defaults.

@item actual unit test skeletons:
a test skeleton for each visible subprogram is created in a separate file, if it
doesn't exist already. By default, those separate test files are located in a
"gnattest/tests" directory that is created in the object directory of
corresponding project file. For example, if a source file my_unit.ads in
directory src contains a visible subprogram Proc, then the corresponding unit
test will be found in file src/tests/my_unit-test_data-tests.adb and will be
called Test_Proc_<code>. <code> is a signature encoding used to differentiate
test names in case of overloading.

Note that if the project already has both my_unit.ads and my_unit-test_data.ads,
this will cause a name conflict with the generated test package.
@end itemize


@noindent
@b{In the tests execution mode mode}, @command{gnattest} has a command-line
interface of the form

@smallexample
@c $ gnattest @var{-Pprojname} @ovar{switches} @ovar{filename} @ovar{directory}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnattest @var{test_drivers.list} @r{[}@var{switches}@r{]}
@end smallexample

@noindent
where
@table @var

@item test_drivers.list
is the name of the text file containing the list of executables to treat as
test drivers. This file is automatically generated by gnattest, but can be
hand-edited to add or remove tests. This switch is required.

@item switches
is an optional sequence of switches as described below.

@end table


@node Switches for gnattest in framework generation mode
@section Switches for @command{gnattest} in framework generation mode

@table @option
@c !sort!

@item -q
@cindex @option{-q} (@command{gnattest})
Quiet mode: suppresses noncritical output messages.

@item -v
@cindex @option{-v} (@command{gnattest})
Verbose mode: generates version information if specified by itself on the
command line.  If specified via GNATtest_Switches, produces output
about the execution of the tool.

@item -r
@cindex @option{-r} (@command{gnattest})
Recursively considers all sources from all projects.


@item -X@var{name=value}
@cindex @option{-X} (@command{gnattest})
Indicate that external variable @var{name} has the value @var{value}.

@item --RTS=@var{rts-path}
@cindex @option{--RTS} (@command{gnattest})
Specifies the default location of the runtime library. Same meaning as the
equivalent @command{gnatmake} flag (@pxref{Switches for gnatmake}).


@item --additional-tests=@var{projname}
@cindex @option{--additional-tests} (@command{gnattest})
Sources described in @var{projname} are considered potential additional
manual tests to be added to the test suite.

@item --harness-only
@cindex @option{--harness-only} (@command{gnattest})
When this option is given, @command{gnattest} creates a harness for all
sources, treating them as test packages.

@item --separate-drivers
@cindex @option{--separate-drivers} (@command{gnattest})
Generates a separate test driver for each test, rather than a single
executable incorporating all tests.

@item --stub
@cindex @option{--stub} (@command{gnattest})
Generates the testing framework that uses subsystem stubbing to isolate the
code under test.


@item --harness-dir=@var{dirname}
@cindex @option{--harness-dir} (@command{gnattest})
Specifies the directory that will hold the harness packages and project file
for the test driver. If the @var{dirname} is a relative path, it is considered
relative to the object directory of the project file.

@item --tests-dir=@var{dirname}
@cindex @option{--tests-dir} (@command{gnattest})
All test packages are placed in the @var{dirname} directory.
If the @var{dirname} is a relative path, it is considered relative to the object
directory of the project file. When all sources from all projects are taken
recursively from all projects, @var{dirname} directories are created for each
project in their object directories and test packages are placed accordingly.

@item --subdir=@var{dirname}
@cindex @option{--subdir} (@command{gnattest})
Test packages are placed in a subdirectory of the corresponding source
directory, with the name @var{dirname}. Thus, each set of unit tests is located
in a subdirectory of the code under test.  If the sources are in separate
directories, each source directory has a test subdirectory named @var{dirname}.

@item --tests-root=@var{dirname}
@cindex @option{--tests-root} (@command{gnattest})
The hierarchy of source directories, if any, is recreated in the @var{dirname}
directory, with test packages placed in directories corresponding to those
of the sources.
If the @var{dirname} is a relative path, it is considered relative to the object
directory of the project file. When projects are considered recursively,
directory hierarchies of tested sources are
recreated for each project in their object directories and test packages are
placed accordingly.

@item --stubs-dir=@var{dirname}
@cindex @option{--stubs-dir} (@command{gnattest})
The hierarchy of directories containing stubbed units is recreated in
the @var{dirname} directory, with stubs placed in directories corresponding to
projects they are derived from.
If the @var{dirname} is a relative path, it is considered relative to the object
directory of the project file. When projects are considered recursively,
directory hierarchies of stubs are
recreated for each project in their object directories and test packages are
placed accordingly.


@item --validate-type-extensions
@cindex @option{--validate-type-extensions} (@command{gnattest})
Enables substitution check: run all tests from all parents in order
to check substitutability in accordance with LSP.

@item --skeleton-default=@var{val}
@cindex @option{--skeleton-default} (@command{gnattest})
Specifies the default behavior of generated skeletons. @var{val} can be either
"fail" or "pass", "fail" being the default.

@item --passed-tests=@var{val}
@cindex @option{--passed-tests} (@command{gnattest})
Specifies whether or not passed tests should be shown. @var{val} can be either
"show" or "hide", "show" being the default.

@item --exit-status=@var{val}
@cindex @option{--exit-status} (@command{gnattest})
Specifies whether or not generated test driver should return failure exit
status if at least one test fails or crashes. @var{val} can be either
"on" or "off", "off" being the default.

@item --omit-sloc
@cindex @option{--omit-sloc} (@command{gnattest})
Suppresses comment line containing file name and line number of corresponding
subprograms in test skeletons.


@item --separates
@cindex @option{--separates} (@command{gnattest})
Bodies of all test routines are generated as separates. Note that this mode is
kept for compatibility reasons only and it is not advised to use it due to
possible problems with hash in names of test skeletons when using an
inconsistent casing. Separate test skeletons can be incorporated to monolith
test package with improved hash being used by using @option{--transition}
switch.


@item --transition
@cindex @option{--transition} (@command{gnattest})
This allows transition from separate test routines to monolith test packages.
All matching test routines are overwritten with contents of corresponding
separates. Note that if separate test routines had any manually added with
clauses they will be moved to the test package body as is and have to be moved
by hand.

@item --test-duration
@cindex @option{--test-duration} (@command{gnattest})
Adds time measurements for each test in generated test driver.

@end table

@option{--tests_root}, @option{--subdir} and @option{--tests-dir} switches are
mutually exclusive.


@node Switches for gnattest in tests execution mode
@section Switches for @command{gnattest} in tests execution mode

@table @option
@c !sort!

@item --passed-tests=@var{val}
@cindex @option{--passed-tests} (@command{gnattest})
Specifies whether or not passed tests should be shown. @var{val} can be either
"show" or "hide", "show" being the default.

@item --queues=@var{n}, -j@var{n}
@cindex @option{--queues} (@command{gnattest})
@cindex @option{-j} (@command{gnattest})
Runs @var{n} tests in parallel (default is 1).

@end table


@node Project Attributes for gnattest
@section Project Attributes for @command{gnattest}

@noindent

Most of the command-line options can also be passed to the tool by adding
special attributes to the project file. Those attributes should be put in
package gnattest. Here is the list of attributes:

@itemize @bullet

@item Tests_Root
is used to select the same output mode as with the --tests-root option.
This attribute cannot be used together with Subdir or Tests_Dir.

@item Subdir
is used to select the same output mode as with the --subdir option.
This attribute cannot be used together with Tests_Root or Tests_Dir.

@item Tests_Dir
is used to select the same output mode as with the --tests-dir option.
This attribute cannot be used together with Subdir or Tests_Root.

@item Harness_Dir
is used to specify the directory in which to place harness packages and project
file for the test driver, otherwise specified by --harness-dir.

@item Additional_Tests
is used to specify the project file, otherwise given by
--additional-tests switch.

@item Skeletons_Default
is used to specify the default behaviour of test skeletons, otherwise
specified by --skeleton-default option. The value of this attribute
should be either "pass" or "fail".

@end itemize

Each of those attributes can be overridden from the command line if needed.
Other @command{gnattest} switches can also be passed via the project
file as an attribute list called GNATtest_Switches.

@node Simple Example
@section Simple Example

@noindent

Let's take a very simple example using the first @command{gnattest} example
located in:

@smallexample
<install_prefix>/share/examples/gnattest/simple
@end smallexample

This project contains a simple package containing one subprogram. By running gnattest:

@smallexample
$ gnattest --harness-dir=driver -Psimple.gpr
@end smallexample

a test driver is created in directory "driver". It can be compiled and run:

@smallexample
$ cd obj/driver
$ gnatmake -Ptest_driver
$ test_runner
@end smallexample

One failed test with diagnosis "test not implemented" is reported.
Since no special output option was specified, the test package Simple.Tests
is located in:

@smallexample
<install_prefix>/share/examples/gnattest/simple/obj/gnattest/tests
@end smallexample

For each package containing visible subprograms, a child test package is
generated. It contains one test routine per tested subprogram. Each
declaration of a test subprogram has a comment specifying which tested
subprogram it corresponds to. Bodies of test routines are placed in test package
bodies and are surrounded by special comment sections. Those comment sections
should not be removed or modified in order for gnattest to be able to regenerate
test packages and keep already written tests in place.
The test routine Test_Inc_5eaee3 located at simple-test_data-tests.adb contains
a single statement: a call to procedure Assert. It has two arguments:
the Boolean expression we want to check and the diagnosis message to display if
the condition is false.

That is where actual testing code should be written after a proper setup.
An actual check can be performed by replacing the Assert call with:

@smallexample @c ada
Assert (Inc (1) = 2, "wrong incrementation");
@end smallexample

After recompiling and running the test driver, one successfully passed test
is reported.

@node Setting Up and Tearing Down the Testing Environment
@section Setting Up and Tearing Down the Testing Environment

@noindent

Besides test routines themselves, each test package has a parent package
Test_Data that has two procedures: Set_Up and Tear_Down. This package is never
overwritten by the tool. Set_Up is called before each test routine of the
package and Tear_Down is called after each test routine. Those two procedures
can be used to perform necessary initialization and finalization,
memory allocation, etc. Test type declared in Test_Data package is parent type
for the test type of test package and can have user-defined components whose
values can be set by Set_Up routine and used in test routines afterwards.

@node Regenerating Tests
@section Regenerating Tests

@noindent

Bodies of test routines and test_data packages are never overridden after they
have been created once. As long as the name of the subprogram, full expanded Ada
names, and the order of its parameters is the same, and comment sections are
intact the old test routine will fit in its place and no test skeleton will be
generated for the subprogram.

This can be demonstrated with the previous example. By uncommenting declaration
and body of function Dec in simple.ads and simple.adb, running
@command{gnattest} on the project, and then running the test driver:

@smallexample
gnattest --harness-dir=driver -Psimple.gpr
cd obj/driver
gnatmake -Ptest_driver
test_runner
@end smallexample

the old test is not replaced with a stub, nor is it lost, but a new test
skeleton is created for function Dec.

The only way of regenerating tests skeletons is to remove the previously created
tests together with corresponding comment sections.

@node Default Test Behavior
@section Default Test Behavior

@noindent

The generated test driver can treat unimplemented tests in two ways:
either count them all as failed (this is useful to see which tests are still
left to implement) or as passed (to sort out unimplemented ones from those
actually failing).

The test driver accepts a switch to specify this behavior:
--skeleton-default=val, where val is either "pass" or "fail" (exactly as for
@command{gnattest}).

The default behavior of the test driver is set with the same switch
as passed to gnattest when generating the test driver.

Passing it to the driver generated on the first example:

@smallexample
test_runner --skeleton-default=pass
@end smallexample

makes both tests pass, even the unimplemented one.

@node Testing Primitive Operations of Tagged Types
@section Testing Primitive Operations of Tagged Types

@noindent

Creation of test skeletons for primitive operations of tagged types entails
a number of features. Test routines for all primitives of a given tagged type
are placed in a separate child package named according to the tagged type. For
example, if you have tagged type T in package P, all tests for primitives
of T will be in P.T_Test_Data.T_Tests.

Consider running gnattest on the second example (note: actual tests for this
example already exist, so there's no need to worry if the tool reports that
no new stubs were generated):

@smallexample
cd <install_prefix>/share/examples/gnattest/tagged_rec
gnattest --harness-dir=driver -Ptagged_rec.gpr
@end smallexample

Taking a closer look at the test type declared in the test package
Speed1.Controller_Test_Data is necessary. It is declared in:

@smallexample
<install_prefix>/share/examples/gnattest/tagged_rec/obj/gnattest/tests
@end smallexample

Test types are direct or indirect descendants of
AUnit.Test_Fixtures.Test_Fixture type. In the case of nonprimitive tested
subprograms, the user doesn't need to be concerned with them. However,
when generating test packages for primitive operations, there are some things
the user needs to know.

Type Test_Controller has components that allow assignment of various
derivations of type Controller. And if you look at the specification of
package Speed2.Auto_Controller, you will see that Test_Auto_Controller
actually derives from Test_Controller rather than AUnit type Test_Fixture.
Thus, test types mirror the hierarchy of tested types.

The Set_Up procedure of Test_Data package corresponding to a test package
of primitive operations of type T assigns to Fixture a reference to an
object of that exact type T. Notice, however, that if the tagged type has
discriminants, the Set_Up only has a commented template for setting
up the fixture, since filling the discriminant with actual value is up
to the user.

The knowledge of the structure of test types allows additional testing
without additional effort. Those possibilities are described below.

@node Testing Inheritance
@section Testing Inheritance

@noindent

Since the test type hierarchy mimics the hierarchy of tested types, the
inheritance of tests takes place. An example of such inheritance can be
seen by running the test driver generated for the second example. As previously
mentioned, actual tests are already written for this example.

@smallexample
cd obj/driver
gnatmake -Ptest_driver
test_runner
@end smallexample

There are 6 passed tests while there are only 5 testable subprograms. The test
routine for function Speed has been inherited and run against objects of the
derived type.

@node Tagged Types Substitutability Testing
@section Tagged Types Substitutability Testing

@noindent

Tagged Types Substitutability Testing is a way of verifying the global type
consistency by testing. Global type consistency is a principle stating that if
S is a subtype of T (in Ada, S is a derived type of tagged type T),
then objects of type T may be replaced with objects of type S (that is,
objects of type S may be substituted for objects of type T), without
altering any of the desirable properties of the program. When the properties
of the program are expressed in the form of subprogram preconditions and
postconditions (let's call them pre and post), the principle is formulated as
relations between the pre and post of primitive operations and the pre and post
of their derived operations. The pre of a derived operation should not be
stronger than the original pre, and the post of the derived operation should
not be weaker than the original post. Those relations ensure that verifying if
a dispatching call is safe can be done just by using the pre and post of the
root operation.

Verifying global type consistency by testing consists of running all the unit
tests associated with the primitives of a given tagged type with objects of its
derived types.

In the example used in the previous section, there was clearly a violation of
type consistency. The overriding primitive Adjust_Speed in package Speed2
removes the functionality of the overridden primitive and thus doesn't respect
the consistency principle.
Gnattest has a special option to run overridden parent tests against objects
of the type which have overriding primitives:

@smallexample
gnattest --harness-dir=driver --validate-type-extensions -Ptagged_rec.gpr
cd obj/driver
gnatmake -Ptest_driver
test_runner
@end smallexample

While all the tests pass by themselves, the parent test for Adjust_Speed fails
against objects of the derived type.

Non-overridden tests are already inherited for derived test types, so the
--validate-type-extensions enables the application of overriden tests to objects
of derived types.

@node Testing with Contracts
@section Testing with Contracts

@noindent

@command{gnattest} supports pragmas Precondition, Postcondition, and Test_Case,
as well as corresponding aspects.
Test routines are generated, one per each Test_Case associated with a tested
subprogram. Those test routines have special wrappers for tested functions
that have composition of pre- and postcondition of the subprogram with
"requires" and "ensures" of the Test_Case (depending on the mode, pre and post
either count for Nominal mode or do not count for Robustness mode).

The third example demonstrates how this works:

@smallexample
cd <install_prefix>/share/examples/gnattest/contracts
gnattest --harness-dir=driver -Pcontracts.gpr
@end smallexample

Putting actual checks within the range of the contract does not cause any
error reports. For example, for the test routine which corresponds to
test case 1:

@smallexample @c ada
Assert (Sqrt (9.0) = 3.0, "wrong sqrt");
@end smallexample

and for the test routine corresponding to test case 2:

@smallexample @c ada
Assert (Sqrt (-5.0) = -1.0, "wrong error indication");
@end smallexample

are acceptable:

@smallexample
cd obj/driver
gnatmake -Ptest_driver
test_runner
@end smallexample

However, by changing 9.0 to 25.0 and 3.0 to 5.0, for example, you can get
a precondition violation for test case one. Also, by using any otherwise
correct but positive pair of numbers in the second test routine, you can also
get a precondition violation. Postconditions are checked and reported
the same way.

@node Additional Tests
@section Additional Tests

@noindent
@command{gnattest} can add user-written tests to the main suite of the test
driver. @command{gnattest} traverses the given packages and searches for test
routines. All procedures with a single in out parameter of a type which is
derived from AUnit.Test_Fixtures.Test_Fixture and that are declared in package
specifications are added to the suites and are then executed by the test driver.
(Set_Up and Tear_Down are filtered out.)

An example illustrates two ways of creating test harnesses for user-written
tests. Directory additional_tests contains an AUnit-based test driver written
by hand.

@smallexample
<install_prefix>/share/examples/gnattest/additional_tests/
@end smallexample

To create a test driver for already-written tests, use the --harness-only
option:

@smallexample
gnattest -Padditional/harness/harness.gpr --harness-dir=harness_only \
  --harness-only
gnatmake -Pharness_only/test_driver.gpr
harness_only/test_runner
@end smallexample

Additional tests can also be executed together with generated tests:

@smallexample
gnattest -Psimple.gpr --additional-tests=additional/harness/harness.gpr \
  --harness-dir=mixing
gnatmake -Pmixing/test_driver.gpr
mixing/test_runner
@end smallexample


@node Individual Test Drivers
@section Individual Test Drivers

@noindent
By default, @command{gnattest} generates a monolithic test driver that
aggregates the individual tests into a single executable. It is also possible
to generate separate executables for each test, by passing the switch
@option{--separate-drivers}. This approach scales better for large testing
campaigns, especially involving target architectures with limited resources
typical for embedded development. It can also provide a major performance
benefit on multi-core systems by allowing simultaneous execution of multiple
tests.

@command{gnattest} can take charge of executing the individual tests; for this,
instead of passing a project file, a text file containing the list of
executables can be passed. Such a file is automatically generated by gnattest
under the name @option{test_drivers.list}, but it can be
hand-edited to add or remove tests, or replaced. The individual tests can
also be executed standalone, or from any user-defined scripted framework.


@node Stubbing
@section Stubbing

@noindent
Depending on the testing campaign, it is sometimes necessary to isolate the
part of the algorithm under test from its dependencies. This is accomplished
via @emph{stubbing}, i.e. replacing the subprograms that are called from the
subprogram under test by stand-in subprograms that match the profiles of the
original ones, but simply return predetermined values required by the test
scenario.

This mode of test harness generation is activated by the switch @option{--stub}.

The implementation approach chosen by @command{gnattest} is as follows.
For each package under consideration all the packages it is directly depending
on are stubbed, excluding the generic packages and package instantiations.
The stubs are shared for each package under test. The specs of packages to stub
remain intact, while their bodies are replaced, and hide the original bodies by
means of extending projects. Also, for each stubbed
package, a child package with setter routines for each subprogram declaration
is created. These setters are meant to be used to set the behaviour of
stubbed subprograms from within test cases.

Note that subprograms belonging to the same package as the subprogram under
test are not stubbed. This guarantees that the sources being tested are
exactly the sources used for production, which is an important property for
establishing the traceability between the testing campaign and production code.

Due to the nature of stubbing process, this mode implies the switch
@option{--separate-drivers}, i.e. an individual test driver (with the
corresponding hierarchy of extending projects) is generated for each test.

@quotation Note
Developing a stubs-based testing campaign requires
good understanding of the infrastructure created by @command{gnattest} for
this purpose. We recommend following the stubbing tutorials provided
under @file{<install_prefix>/share/examples/gnattest/stubbing*} before
attempting to use this powerful feature.
@end quotation

@node Putting Tests under Version Control
@section Putting Tests under Version Control

@noindent
As has been stated earlier, @command{gnattest} generates two different types
of code, test skeletons and harness. The harness is generated completely
automatically each time, does not require manual changes and therefore should
not be put under version control.
It makes sense to put under version control files containing test data packages,
both specs and bodies, and files containing bodies of test packages. Note that
test package specs are also generated automatically each time and should not be
put under version control.
Option @option{--omit-sloc} may be usefull when putting test packages under VCS.

@node Support for other platforms/run-times
@section Support for other platforms/run-times

@noindent
@command{gnattest} can be used to generate the test harness for platforms
and run-time libraries others than the default native target with the
default full run-time. For example, when using a limited run-time library
such as Zero FootPrint (ZFP), a simplified harness is generated.

Two variables are used to tell the underlying AUnit framework how to generate
the test harness: @code{PLATFORM}, which identifies the target, and
@code{RUNTIME}, used to determine the run-time library for which the harness
is generated. Corresponding prefix should also be used when calling
@command{gnattest} for non-native targets. For example, the following options
are used to generate the AUnit test harness for a PowerPC ELF target using
the ZFP run-time library:

@smallexample
powerpc-elf-gnattest -Psimple.gpr -XPLATFORM=powerpc-elf -XRUNTIME=zfp
@end smallexample

@node Current Limitations
@section Current Limitations

@noindent

The tool currently does not support following features:

@itemize @bullet
@item generic tests for nested generic packages and their instantiations
@item tests for protected subprograms and entries

@end itemize
@end ifclear


@c *********************************
@node Performing Dimensionality Analysis in GNAT
@chapter Performing Dimensionality Analysis in GNAT
@cindex Dimensionality analysis

@noindent
The GNAT compiler now supports dimensionality checking. The user can
specify physical units for objects, and the compiler will verify that uses
of these objects are compatible with their dimensions, in a fashion that is
familiar to engineering practice. The dimensions of algebraic expressions
(including powers with static exponents) are computed from their constituents.

This feature depends on Ada 2012 aspect specifications, and is available from
version 7.0.1 of GNAT onwards.
The GNAT-specific aspect @code{Dimension_System}
@cindex @code{Dimension_System} aspect
allows you to define a system of units; the aspect @code{Dimension}
@cindex @code{Dimension} aspect
then allows the user to declare dimensioned quantities within a given system.
(These aspects are described in the @i{Implementation Defined Aspects}
chapter of the @i{GNAT Reference Manual}).

The major advantage of this model is that it does not require the declaration of
multiple operators for all possible combinations of types: it is only necessary
to use the proper subtypes in object declarations.

The simplest way to impose dimensionality checking on a computation is to make
use of the package @code{System.Dim.Mks},
@cindex @code{System.Dim.Mks} package (GNAT library)
which is part of the GNAT library. This
package defines a floating-point type @code{MKS_Type},
@cindex @code{MKS_Type} type
for which a sequence of
dimension names are specified, together with their conventional abbreviations.
The following should be read together with the full specification of the
package, in file @file{s-dimmks.ads}.
@cindex @file{s-dimmks.ads} file

@smallexample @c ada
@group
   @b{type} Mks_Type @b{is} @b{new} Long_Long_Float
     @b{with}
      Dimension_System => (
        (Unit_Name => Meter,    Unit_Symbol => 'm',   Dim_Symbol => 'L'),
        (Unit_Name => Kilogram, Unit_Symbol => "kg",  Dim_Symbol => 'M'),
        (Unit_Name => Second,   Unit_Symbol => 's',   Dim_Symbol => 'T'),
        (Unit_Name => Ampere,   Unit_Symbol => 'A',   Dim_Symbol => 'I'),
        (Unit_Name => Kelvin,   Unit_Symbol => 'K',   Dim_Symbol => "Theta"),
        (Unit_Name => Mole,     Unit_Symbol => "mol", Dim_Symbol => 'N'),
        (Unit_Name => Candela,  Unit_Symbol => "cd",  Dim_Symbol => 'J'));
@end group
@end smallexample

@noindent
The package then defines a series of subtypes that correspond to these
conventional units. For example:

@smallexample @c ada
@group
   @b{subtype} Length @b{is} Mks_Type
     @b{with}
      Dimension => (Symbol => 'm', Meter  => 1, @b{others} => 0);
@end group
@end smallexample

@noindent
and similarly for @code{Mass}, @code{Time}, @code{Electric_Current},
@code{Thermodynamic_Temperature}, @code{Amount_Of_Substance}, and
@code{Luminous_Intensity} (the standard set of units of the SI system).

The package also defines conventional names for values of each unit, for
example:

@smallexample @c ada
@group
   m   : @b{constant} Length           := 1.0;
   kg  : @b{constant} Mass             := 1.0;
   s   : @b{constant} Time             := 1.0;
   A   : @b{constant} Electric_Current := 1.0;
@end group
@end smallexample

@noindent
as well as useful multiples of these units:

@smallexample @c ada
@group
   cm  : @b{constant} Length := 1.0E-02;
   g   : @b{constant} Mass   := 1.0E-03;
   min : @b{constant} Time   := 60.0;
   day : @b{constant} Time   := 60.0 * 24.0 * min;
  ...
@end group
@end smallexample

@noindent
Using this package, you can then define a derived unit by
providing the aspect that
specifies its dimensions within the MKS system, as well as the string to
be used for output of a value of that unit:

@smallexample @c ada
@group
  @b{subtype} Acceleration @b{is} Mks_Type
    @b{with} Dimension => ("m/sec^2",
                       Meter => 1,
                       Second => -2,
                       @b{others} => 0);
@end group
@end smallexample

@noindent
Here is a complete example of use:

@smallexample @c ada
@group
@b{with} System.Dim.MKS; @b{use} System.Dim.Mks;
@b{with} System.Dim.Mks_IO; @b{use} System.Dim.Mks_IO;
@b{with} Text_IO; @b{use} Text_IO;
@b{procedure} Free_Fall @b{is}
  @b{subtype} Acceleration @b{is} Mks_Type
    @b{with} Dimension => ("m/sec^2", 1, 0, -2, @b{others} => 0);
  G : @b{constant} acceleration := 9.81 * m / (s ** 2);
  T : Time := 10.0*s;
  Distance : Length;
@end group
@group
@b{begin}
  Put ("Gravitational constant: ");
  Put (G, Aft => 2, Exp => 0); Put_Line ("");
  Distance := 0.5 * G * T ** 2;
  Put ("distance travelled in 10 seconds of free fall ");
  Put (Distance, Aft => 2, Exp => 0);
  Put_Line ("");
@b{end} Free_Fall;
@end group
@end smallexample

@noindent
Execution of this program yields:
@smallexample
@group
Gravitational constant:  9.81 m/sec^2
distance travelled in 10 seconds of free fall 490.50 m
@end group
@end smallexample

@noindent
However, incorrect assignments such as:

@smallexample @c ada
@group
   Distance := 5.0;
   Distance := 5.0 * kg:
@end group
@end smallexample

@noindent
are rejected with the following diagnoses:

@smallexample
@group
   Distance := 5.0;
      >>> dimensions mismatch in assignment
      >>> left-hand side has dimension [L]
      >>> right-hand side is dimensionless
@end group

@group
   Distance := 5.0 * kg:
      >>> dimensions mismatch in assignment
      >>> left-hand side has dimension [L]
      >>> right-hand side has dimension [M]
@end group
@end smallexample

@noindent
The dimensions of an expression are properly displayed, even if there is
no explicit subtype for it. If we add to the program:

@smallexample @c ada
@group
      Put ("Final velocity: ");
      Put (G * T, Aft =>2, Exp =>0);
      Put_Line ("");
@end group
@end smallexample

@noindent
then the output includes:
@smallexample
     Final velocity: 98.10 m.s**(-1)
@end smallexample


@c *********************************
@node Generating Ada Bindings for C and C++ headers
@chapter Generating Ada Bindings for C and C++ headers
@findex binding

@noindent
GNAT now comes with a binding generator for C and C++ headers which is
intended to do 95% of the tedious work of generating Ada specs from C
or C++ header files.

Note that this capability is not intended to generate 100% correct Ada specs,
and will is some cases require manual adjustments, although it can often
be used out of the box in practice.

Some of the known limitations include:

@itemize @bullet
@item only very simple character constant macros are translated into Ada
constants. Function macros (macros with arguments) are partially translated
as comments, to be completed manually if needed.
@item some extensions (e.g. vector types) are not supported
@item pointers to pointers or complex structures are mapped to System.Address
@item identifiers with identical name (except casing) will generate compilation
      errors (e.g. @code{shm_get} vs @code{SHM_GET}).
@end itemize

The code generated is using the Ada 2005 syntax, which makes it
easier to interface with other languages than previous versions of Ada.

@menu
* Running the binding generator::
* Generating bindings for C++ headers::
* Switches::
@end menu

@node Running the binding generator
@section Running the binding generator

@noindent
The binding generator is part of the @command{gcc} compiler and can be
invoked via the @option{-fdump-ada-spec} switch, which will generate Ada
spec files for the header files specified on the command line, and all
header files needed by these files transitively. For example:

@smallexample
$ g++ -c -fdump-ada-spec -C /usr/include/time.h
$ gcc -c -gnat05 *.ads
@end smallexample

will generate, under GNU/Linux, the following files: @file{time_h.ads},
@file{bits_time_h.ads}, @file{stddef_h.ads}, @file{bits_types_h.ads} which
correspond to the files @file{/usr/include/time.h},
@file{/usr/include/bits/time.h}, etc@dots{}, and will then compile in Ada 2005
mode these Ada specs.

The @code{-C} switch tells @command{gcc} to extract comments from headers,
and will attempt to generate corresponding Ada comments.

If you want to generate a single Ada file and not the transitive closure, you
can use instead the @option{-fdump-ada-spec-slim} switch.

You can optionally specify a parent unit, of which all generated units will
be children, using @code{-fada-spec-parent=}@var{unit}.

Note that we recommend when possible to use the @command{g++} driver to
generate bindings, even for most C headers, since this will in general
generate better Ada specs. For generating bindings for C++ headers, it is
mandatory to use the @command{g++} command, or @command{gcc -x c++} which
is equivalent in this case. If @command{g++} cannot work on your C headers
because of incompatibilities between C and C++, then you can fallback to
@command{gcc} instead.

For an example of better bindings generated from the C++ front-end,
the name of the parameters (when available) are actually ignored by the C
front-end. Consider the following C header:

@smallexample
extern void foo (int variable);
@end smallexample

with the C front-end, @code{variable} is ignored, and the above is handled as:

@smallexample
extern void foo (int);
@end smallexample

generating a generic:

@smallexample
procedure foo (param1 : int);
@end smallexample

with the C++ front-end, the name is available, and we generate:

@smallexample
procedure foo (variable : int);
@end smallexample

In some cases, the generated bindings will be more complete or more meaningful
when defining some macros, which you can do via the @option{-D} switch. This
is for example the case with @file{Xlib.h} under GNU/Linux:

@smallexample
g++ -c -fdump-ada-spec -DXLIB_ILLEGAL_ACCESS -C /usr/include/X11/Xlib.h
@end smallexample

The above will generate more complete bindings than a straight call without
the @option{-DXLIB_ILLEGAL_ACCESS} switch.

In other cases, it is not possible to parse a header file in a stand-alone
manner, because other include files need to be included first. In this
case, the solution is to create a small header file including the needed
@code{#include} and possible @code{#define} directives. For example, to
generate Ada bindings for @file{readline/readline.h}, you need to first
include @file{stdio.h}, so you can create a file with the following two
lines in e.g. @file{readline1.h}:

@smallexample
#include <stdio.h>
#include <readline/readline.h>
@end smallexample

and then generate Ada bindings from this file:

@smallexample
$ g++ -c -fdump-ada-spec readline1.h
@end smallexample

@node Generating bindings for C++ headers
@section Generating bindings for C++ headers

@noindent
Generating bindings for C++ headers is done using the same options, always
with the @command{g++} compiler.

In this mode, C++ classes will be mapped to Ada tagged types, constructors
will be mapped using the @code{CPP_Constructor} pragma, and when possible,
multiple inheritance of abstract classes will be mapped to Ada interfaces
(@xref{Interfacing to C++,,,gnat_rm, GNAT Reference Manual}, for additional
information on interfacing to C++).

For example, given the following C++ header file:

@smallexample
@group
@cartouche
class Carnivore @{
public:
   virtual int Number_Of_Teeth () = 0;
@};

class Domestic @{
public:
   virtual void Set_Owner (char* Name) = 0;
@};

class Animal @{
public:
  int Age_Count;
  virtual void Set_Age (int New_Age);
@};

class Dog : Animal, Carnivore, Domestic @{
 public:
  int  Tooth_Count;
  char *Owner;

  virtual int  Number_Of_Teeth ();
  virtual void Set_Owner (char* Name);

  Dog();
@};
@end cartouche
@end group
@end smallexample

The corresponding Ada code is generated:

@smallexample @c ada
@group
@cartouche
  @b{package} Class_Carnivore @b{is}
    @b{type} Carnivore @b{is} @b{limited} interface;
    @b{pragma} Import (CPP, Carnivore);

    @b{function} Number_Of_Teeth (this : @b{access} Carnivore) @b{return} int @b{is} @b{abstract};
  @b{end};
  @b{use} Class_Carnivore;

  @b{package} Class_Domestic @b{is}
    @b{type} Domestic @b{is} @b{limited} interface;
    @b{pragma} Import (CPP, Domestic);

    @b{procedure} Set_Owner
      (this : @b{access} Domestic;
       Name : Interfaces.C.Strings.chars_ptr) @b{is} @b{abstract};
  @b{end};
  @b{use} Class_Domestic;

  @b{package} Class_Animal @b{is}
    @b{type} Animal @b{is} @b{tagged} @b{limited} @b{record}
      Age_Count : @b{aliased} int;
    @b{end} @b{record};
    @b{pragma} Import (CPP, Animal);

    @b{procedure} Set_Age (this : @b{access} Animal; New_Age : int);
    @b{pragma} Import (CPP, Set_Age, "_ZN6Animal7Set_AgeEi");
  @b{end};
  @b{use} Class_Animal;

  @b{package} Class_Dog @b{is}
    @b{type} Dog @b{is} @b{new} Animal @b{and} Carnivore @b{and} Domestic @b{with} @b{record}
      Tooth_Count : @b{aliased} int;
      Owner : Interfaces.C.Strings.chars_ptr;
    @b{end} @b{record};
    @b{pragma} Import (CPP, Dog);

    @b{function} Number_Of_Teeth (this : @b{access} Dog) @b{return} int;
    @b{pragma} Import (CPP, Number_Of_Teeth, "_ZN3Dog15Number_Of_TeethEv");

    @b{procedure} Set_Owner
      (this : @b{access} Dog; Name : Interfaces.C.Strings.chars_ptr);
    @b{pragma} Import (CPP, Set_Owner, "_ZN3Dog9Set_OwnerEPc");

    @b{function} New_Dog @b{return} Dog;
    @b{pragma} CPP_Constructor (New_Dog);
    @b{pragma} Import (CPP, New_Dog, "_ZN3DogC1Ev");
  @b{end};
  @b{use} Class_Dog;
@end cartouche
@end group
@end smallexample

@node Switches
@section Switches

@table @option
@item -fdump-ada-spec
@cindex @option{-fdump-ada-spec} (@command{gcc})
Generate Ada spec files for the given header files transitively (including
all header files that these headers depend upon).

@item -fdump-ada-spec-slim
@cindex @option{-fdump-ada-spec-slim} (@command{gcc})
Generate Ada spec files for the header files specified on the command line
only.

@item -fada-spec-parent=@var{unit}
@cindex -fada-spec-parent (@command{gcc})
Specifies that all files generated by @option{-fdump-ada-spec*} are
to be child units of the specified parent unit.

@item -C
@cindex @option{-C} (@command{gcc})
Extract comments from headers and generate Ada comments in the Ada spec files.
@end table

@node Other Utility Programs
@chapter Other Utility Programs

@noindent
This chapter discusses some other utility programs available in the Ada
environment.

@menu
* Using Other Utility Programs with GNAT::
* The External Symbol Naming Scheme of GNAT::
* Converting Ada Files to html with gnathtml::
* Installing gnathtml::
@end menu

@node Using Other Utility Programs with GNAT
@section Using Other Utility Programs with GNAT

@noindent
The object files generated by GNAT are in standard system format and in
particular the debugging information uses this format. This means
programs generated by GNAT can be used with existing utilities that
depend on these formats.

In general, any utility program that works with C will also often work with
Ada programs generated by GNAT. This includes software utilities such as
gprof (a profiling program), @code{gdb} (the FSF debugger), and utilities such
as Purify.

@node The External Symbol Naming Scheme of GNAT
@section The External Symbol Naming Scheme of GNAT

@noindent
In order to interpret the output from GNAT, when using tools that are
originally intended for use with other languages, it is useful to
understand the conventions used to generate link names from the Ada
entity names.

All link names are in all lowercase letters. With the exception of library
procedure names, the mechanism used is simply to use the full expanded
Ada name with dots replaced by double underscores. For example, suppose
we have the following package spec:

@smallexample @c ada
@group
@cartouche
@b{package} QRS @b{is}
   MN : Integer;
@b{end} QRS;
@end cartouche
@end group
@end smallexample

@noindent
The variable @code{MN} has a full expanded Ada name of @code{QRS.MN}, so
the corresponding link name is @code{qrs__mn}.
@findex Export
Of course if a @code{pragma Export} is used this may be overridden:

@smallexample @c ada
@group
@cartouche
@b{package} Exports @b{is}
   Var1 : Integer;
   @b{pragma} Export (Var1, C, External_Name => "var1_name");
   Var2 : Integer;
   @b{pragma} Export (Var2, C, Link_Name => "var2_link_name");
@b{end} Exports;
@end cartouche
@end group
@end smallexample

@noindent
In this case, the link name for @var{Var1} is whatever link name the
C compiler would assign for the C function @var{var1_name}. This typically
would be either @var{var1_name} or @var{_var1_name}, depending on operating
system conventions, but other possibilities exist. The link name for
@var{Var2} is @var{var2_link_name}, and this is not operating system
dependent.

@findex _main
One exception occurs for library level procedures. A potential ambiguity
arises between the required name @code{_main} for the C main program,
and the name we would otherwise assign to an Ada library level procedure
called @code{Main} (which might well not be the main program).

To avoid this ambiguity, we attach the prefix @code{_ada_} to such
names. So if we have a library level procedure such as

@smallexample @c ada
@group
@cartouche
@b{procedure} Hello (S : String);
@end cartouche
@end group
@end smallexample

@noindent
the external name of this procedure will be @var{_ada_hello}.


@node Converting Ada Files to html with gnathtml
@section Converting Ada Files to HTML with @code{gnathtml}

@noindent
This @code{Perl} script allows Ada source files to be browsed using
standard Web browsers. For installation procedure, see the section
@xref{Installing gnathtml}.

Ada reserved keywords are highlighted in a bold font and Ada comments in
a blue font. Unless your program was compiled with the gcc @option{-gnatx}
switch to suppress the generation of cross-referencing information, user
defined variables and types will appear in a different color; you will
be able to click on any identifier and go to its declaration.

The command line is as follow:
@smallexample
@c $ perl gnathtml.pl @ovar{switches} @var{ada-files}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ perl gnathtml.pl @r{[}@var{switches}@r{]} @var{ada-files}
@end smallexample

@noindent
You can pass it as many Ada files as you want. @code{gnathtml} will generate
an html file for every ada file, and a global file called @file{index.htm}.
This file is an index of every identifier defined in the files.

The available switches are the following ones:

@table @option
@item -83
@cindex @option{-83} (@code{gnathtml})
Only the Ada 83 subset of keywords will be highlighted.

@item -cc @var{color}
@cindex @option{-cc} (@code{gnathtml})
This option allows you to change the color used for comments. The default
value is green. The color argument can be any name accepted by html.

@item -d
@cindex @option{-d} (@code{gnathtml})
If the Ada files depend on some other files (for instance through
@code{with} clauses, the latter files will also be converted to html.
Only the files in the user project will be converted to html, not the files
in the run-time library itself.

@item -D
@cindex @option{-D} (@code{gnathtml})
This command is the same as @option{-d} above, but @command{gnathtml} will
also look for files in the run-time library, and generate html files for them.

@item -ext @var{extension}
@cindex @option{-ext} (@code{gnathtml})
This option allows you to change the extension of the generated HTML files.
If you do not specify an extension, it will default to @file{htm}.

@item -f
@cindex @option{-f} (@code{gnathtml})
By default, gnathtml will generate html links only for global entities
('with'ed units, global variables and types,@dots{}).  If you specify
@option{-f} on the command line, then links will be generated for local
entities too.

@item -l @var{number}
@cindex @option{-l} (@code{gnathtml})
If this switch is provided and @var{number} is not 0, then
@code{gnathtml} will number the html files every @var{number} line.

@item -I @var{dir}
@cindex @option{-I} (@code{gnathtml})
Specify a directory to search for library files (@file{.ALI} files) and
source files. You can provide several -I switches on the command line,
and the directories will be parsed in the order of the command line.

@item -o @var{dir}
@cindex @option{-o} (@code{gnathtml})
Specify the output directory for html files. By default, gnathtml will
saved the generated html files in a subdirectory named @file{html/}.

@item -p @var{file}
@cindex @option{-p} (@code{gnathtml})
If you are using Emacs and the most recent Emacs Ada mode, which provides
a full Integrated Development Environment for compiling, checking,
running and debugging applications, you may use @file{.gpr} files
to give the directories where Emacs can find sources and object files.

Using this switch, you can tell gnathtml to use these files.
This allows you to get an html version of your application, even if it
is spread over multiple directories.

@item -sc @var{color}
@cindex @option{-sc} (@code{gnathtml})
This switch allows you to change the color used for symbol
definitions.
The default value is red. The color argument can be any name accepted by html.

@item -t @var{file}
@cindex @option{-t} (@code{gnathtml})
This switch provides the name of a file. This file contains a list of
file names to be converted, and the effect is exactly as though they had
appeared explicitly on the command line. This
is the recommended way to work around the command line length limit on some
systems.

@end table

@node Installing gnathtml
@section Installing @code{gnathtml}

@noindent
@code{Perl} needs to be installed on your machine to run this script.
@code{Perl} is freely available for almost every architecture and
Operating System via the Internet.

On Unix systems, you  may want to modify  the  first line of  the script
@code{gnathtml},  to explicitly  tell  the Operating  system  where Perl
is. The syntax of this line is:
@smallexample
#!full_path_name_to_perl
@end smallexample

@noindent
Alternatively, you may run the script using the following command line:

@smallexample
@c $ perl gnathtml.pl @ovar{switches} @var{files}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ perl gnathtml.pl @r{[}@var{switches}@r{]} @var{files}
@end smallexample


@c ******************************
@node Code Coverage and Profiling
@chapter Code Coverage and Profiling
@cindex Code Coverage
@cindex Profiling

@noindent
This chapter describes how to use @code{gcov} - coverage testing tool - and
@code{gprof} - profiler tool - on your Ada programs.

@menu
* Code Coverage of Ada Programs with gcov::
* Profiling an Ada Program with gprof::
@end menu

@node Code Coverage of Ada Programs with gcov
@section Code Coverage of Ada Programs with gcov
@cindex gcov
@cindex -fprofile-arcs
@cindex -ftest-coverage
@cindex -coverage
@cindex Code Coverage

@noindent
@code{gcov} is a test coverage program: it analyzes the execution of a given
program on selected tests, to help you determine the portions of the program
that are still untested.

@code{gcov} is part of the GCC suite, and is described in detail in the GCC
User's Guide. You can refer to this documentation for a more complete
description.

This chapter provides a quick startup guide, and
details some Gnat-specific features.

@menu
* Quick startup guide::
* Gnat specifics::
@end menu

@node Quick startup guide
@subsection Quick startup guide

In order to perform coverage analysis of a program using @code{gcov}, 3
steps are needed:

@itemize @bullet
@item
Code instrumentation during the compilation process
@item
Execution of the instrumented program
@item
Execution of the @code{gcov} tool to generate the result.
@end itemize

The code instrumentation needed by gcov is created at the object level:
The source code is not modified in any way, because the instrumentation code is
inserted by gcc during the compilation process. To compile your code with code
coverage activated, you need to recompile your whole project using the
switches
@code{-fprofile-arcs} and @code{-ftest-coverage}, and link it using
@code{-fprofile-arcs}.

@smallexample
$ gnatmake -P my_project.gpr -f -cargs -fprofile-arcs -ftest-coverage \
   -largs -fprofile-arcs
@end smallexample

This compilation process will create @file{.gcno} files together with
the usual object files.

Once the program is compiled with coverage instrumentation, you can
run it as many times as needed - on portions of a test suite for
example. The first execution will produce @file{.gcda} files at the
same location as the @file{.gcno} files.  The following executions
will update those files, so that a cumulative result of the covered
portions of the program is generated.

Finally, you need to call the @code{gcov} tool. The different options of
@code{gcov} are available in the GCC User's Guide, section 'Invoking gcov'.

This will create annotated source files with a @file{.gcov} extension:
@file{my_main.adb} file will be analysed in @file{my_main.adb.gcov}.

@node Gnat specifics
@subsection Gnat specifics

Because Ada semantics, portions of the source code may be shared among
several object files. This is the case for example when generics are
involved, when inlining is active  or when declarations generate  initialisation
calls. In order to take
into account this shared code, you need to call @code{gcov} on all
source files of the tested program at once.

The list of source files might exceed the system's maximum command line
length. In order to bypass this limitation, a new mechanism has been
implemented in @code{gcov}: you can now list all your project's files into a
text file, and provide this file to gcov as a parameter,  preceded by a @@
(e.g. @samp{gcov @@mysrclist.txt}).

Note that on AIX compiling a static library with @code{-fprofile-arcs} is
not supported as there can be unresolved symbols during the final link.

@node Profiling an Ada Program with gprof
@section Profiling an Ada Program with gprof
@cindex gprof
@cindex -pg
@cindex Profiling

@noindent
This section is not meant to be an exhaustive documentation of @code{gprof}.
Full documentation for it can be found in the GNU Profiler User's Guide
documentation that is part of this GNAT distribution.

Profiling a program helps determine the parts of a program that are executed
most often, and are therefore the most time-consuming.

@code{gprof} is the standard GNU profiling tool; it has been enhanced to
better handle Ada programs and multitasking.
It is currently supported on the following platforms
@itemize @bullet
@item
linux x86/x86_64
@item
solaris sparc/sparc64/x86
@item
windows x86
@end itemize

@noindent
In order to profile a program using @code{gprof}, 3 steps are needed:

@itemize @bullet
@item
Code instrumentation, requiring a full recompilation of the project with the
proper switches.
@item
Execution of the program under the analysis conditions, i.e. with the desired
input.
@item
Analysis of the results using the @code{gprof} tool.
@end itemize

@noindent
The following sections detail the different steps, and indicate how
to interpret the results:
@menu
* Compilation for profiling::
* Program execution::
* Running gprof::
* Interpretation of profiling results::
@end menu

@node Compilation for profiling
@subsection Compilation for profiling
@cindex -pg
@cindex Profiling

In order to profile a program the first step is to tell the compiler
to generate the necessary profiling information. The compiler switch to be used
is @code{-pg}, which must be added to other compilation switches. This
switch needs to be specified both during compilation and link stages, and can
be specified once when using gnatmake:

@smallexample
gnatmake -f -pg -P my_project
@end smallexample

@noindent
Note that only the objects that were compiled with the @samp{-pg} switch will
be profiled; if you need to profile your whole project, use the @samp{-f}
gnatmake switch to force full recompilation.

@node Program execution
@subsection Program execution

@noindent
Once the program has been compiled for profiling, you can run it as usual.

The only constraint imposed by profiling is that the program must terminate
normally. An interrupted program (via a Ctrl-C, kill, etc.) will not be
properly analyzed.

Once the program completes execution, a data file called @file{gmon.out} is
generated in the directory where the program was launched from. If this file
already exists, it will be overwritten.

@node Running gprof
@subsection Running gprof

@noindent
The @code{gprof} tool is called as follow:

@smallexample
gprof my_prog gmon.out
@end smallexample

@noindent
or simpler:

@smallexample
gprof my_prog
@end smallexample

@noindent
The complete form of the gprof command line is the following:

@smallexample
gprof [switches] [executable [data-file]]
@end smallexample

@noindent
@code{gprof} supports numerous switch. The order of these
switch does not matter. The full list of options can be found in
the GNU Profiler User's Guide documentation that comes with this documentation.

The following is the subset of those switches that is most relevant:

@table @option

@item --demangle[=@var{style}]
@itemx --no-demangle
@cindex @option{--demangle} (@code{gprof})
These options control whether symbol names should be demangled when
printing output.  The default is to demangle C++ symbols.  The
@code{--no-demangle} option may be used to turn off demangling. Different
compilers have different mangling styles.  The optional demangling style
argument can be used to choose an appropriate demangling style for your
compiler, in particular Ada symbols generated by GNAT can be demangled using
@code{--demangle=gnat}.

@item -e @var{function_name}
@cindex @option{-e} (@code{gprof})
The @samp{-e @var{function}} option tells @code{gprof} not to print
information about the function @var{function_name} (and its
children@dots{}) in the call graph.  The function will still be listed
as a child of any functions that call it, but its index number will be
shown as @samp{[not printed]}.  More than one @samp{-e} option may be
given; only one @var{function_name} may be indicated with each @samp{-e}
option.

@item -E @var{function_name}
@cindex @option{-E} (@code{gprof})
The @code{-E @var{function}} option works like the @code{-e} option, but
execution time spent in the function (and children who were not called from
anywhere else), will not be used to compute the percentages-of-time for
the call graph.  More than one @samp{-E} option may be given; only one
@var{function_name} may be indicated with each @samp{-E} option.

@item -f @var{function_name}
@cindex @option{-f} (@code{gprof})
The @samp{-f @var{function}} option causes @code{gprof} to limit the
call graph to the function @var{function_name} and its children (and
their children@dots{}).  More than one @samp{-f} option may be given;
only one @var{function_name} may be indicated with each @samp{-f}
option.

@item -F @var{function_name}
@cindex @option{-F} (@code{gprof})
The @samp{-F @var{function}} option works like the @code{-f} option, but
only time spent in the function and its children (and their
children@dots{}) will be used to determine total-time and
percentages-of-time for the call graph.  More than one @samp{-F} option
may be given; only one @var{function_name} may be indicated with each
@samp{-F} option.  The @samp{-F} option overrides the @samp{-E} option.

@end table

@node Interpretation of profiling results
@subsection Interpretation of profiling results

@noindent

The results of the profiling analysis are represented by two arrays: the
'flat profile' and the 'call graph'. Full documentation of those outputs
can be found in the GNU Profiler User's Guide.

The flat profile shows the time spent in each function of the program, and how
many time it has been called. This allows you to locate easily the most
time-consuming functions.

The call graph shows, for each subprogram, the subprograms that call it,
and the subprograms that it calls. It also provides an estimate of the time
spent in each of those callers/called subprograms.

@c ******************************
@node Running and Debugging Ada Programs
@chapter Running and Debugging Ada Programs
@cindex Debugging

@noindent
This chapter discusses how to debug Ada programs.

An incorrect Ada program may be handled in three ways by the GNAT compiler:

@enumerate
@item
The illegality may be a violation of the static semantics of Ada. In
that case GNAT diagnoses the constructs in the program that are illegal.
It is then a straightforward matter for the user to modify those parts of
the program.

@item
The illegality may be a violation of the dynamic semantics of Ada. In
that case the program compiles and executes, but may generate incorrect
results, or may terminate abnormally with some exception.

@item
When presented with a program that contains convoluted errors, GNAT
itself may terminate abnormally without providing full diagnostics on
the incorrect user program.
@end enumerate

@menu
* The GNAT Debugger GDB::
* Running GDB::
* Introduction to GDB Commands::
* Using Ada Expressions::
* Calling User-Defined Subprograms::
* Using the Next Command in a Function::
* Ada Exceptions::
* Ada Tasks::
* Debugging Generic Units::
* Remote Debugging with gdbserver::
* GNAT Abnormal Termination or Failure to Terminate::
* Naming Conventions for GNAT Source Files::
* Getting Internal Debugging Information::
* Stack Traceback::
@end menu

@cindex Debugger
@findex gdb

@node The GNAT Debugger GDB
@section The GNAT Debugger GDB

@noindent
@code{GDB} is a general purpose, platform-independent debugger that
can be used to debug mixed-language programs compiled with @command{gcc},
and in particular is capable of debugging Ada programs compiled with
GNAT. The latest versions of @code{GDB} are Ada-aware and can handle
complex Ada data structures.

@xref{Top,, Debugging with GDB, gdb, Debugging with GDB},
for full details on the usage of @code{GDB}, including a section on
its usage on programs. This manual should be consulted for full
details. The section that follows is a brief introduction to the
philosophy and use of @code{GDB}.

When GNAT programs are compiled, the compiler optionally writes debugging
information into the generated object file, including information on
line numbers, and on declared types and variables. This information is
separate from the generated code. It makes the object files considerably
larger, but it does not add to the size of the actual executable that
will be loaded into memory, and has no impact on run-time performance. The
generation of debug information is triggered by the use of the
-g switch in the @command{gcc} or @command{gnatmake} command
used to carry out the compilations. It is important to emphasize that
the use of these options does not change the generated code.

The debugging information is written in standard system formats that
are used by many tools, including debuggers and profilers. The format
of the information is typically designed to describe C types and
semantics, but GNAT implements a translation scheme which allows full
details about Ada types and variables to be encoded into these
standard C formats. Details of this encoding scheme may be found in
the file exp_dbug.ads in the GNAT source distribution. However, the
details of this encoding are, in general, of no interest to a user,
since @code{GDB} automatically performs the necessary decoding.

When a program is bound and linked, the debugging information is
collected from the object files, and stored in the executable image of
the program. Again, this process significantly increases the size of
the generated executable file, but it does not increase the size of
the executable program itself. Furthermore, if this program is run in
the normal manner, it runs exactly as if the debug information were
not present, and takes no more actual memory.

However, if the program is run under control of @code{GDB}, the
debugger is activated.  The image of the program is loaded, at which
point it is ready to run.  If a run command is given, then the program
will run exactly as it would have if @code{GDB} were not present. This
is a crucial part of the @code{GDB} design philosophy.  @code{GDB} is
entirely non-intrusive until a breakpoint is encountered.  If no
breakpoint is ever hit, the program will run exactly as it would if no
debugger were present. When a breakpoint is hit, @code{GDB} accesses
the debugging information and can respond to user commands to inspect
variables, and more generally to report on the state of execution.

@c **************
@node Running GDB
@section Running GDB

@noindent
This section describes how to initiate the debugger.
@c The above sentence is really just filler, but it was otherwise
@c clumsy to get the first paragraph nonindented given the conditional
@c nature of the description

The debugger can be launched from a @code{GPS} menu or
directly from the command line. The description below covers the latter use.
All the commands shown can be used in the @code{GPS} debug console window,
but there are usually more GUI-based ways to achieve the same effect.

The command to run @code{GDB} is

@smallexample
$ gdb program
@end smallexample

@noindent
where @code{program} is the name of the executable file. This
activates the debugger and results in a prompt for debugger commands.
The simplest command is simply @code{run}, which causes the program to run
exactly as if the debugger were not present. The following section
describes some of the additional commands that can be given to @code{GDB}.

@c *******************************
@node Introduction to GDB Commands
@section Introduction to GDB Commands

@noindent
@code{GDB} contains a large repertoire of commands.  @xref{Top,,
Debugging with GDB, gdb, Debugging with GDB},
for extensive documentation on the use
of these commands, together with examples of their use. Furthermore,
the command @command{help} invoked from within GDB activates a simple help
facility which summarizes the available commands and their options.
In this section we summarize a few of the most commonly
used commands to give an idea of what @code{GDB} is about. You should create
a simple program with debugging information and experiment with the use of
these @code{GDB} commands on the program as you read through the
following section.

@table @code
@item set args @var{arguments}
The @var{arguments} list above is a list of arguments to be passed to
the program on a subsequent run command, just as though the arguments
had been entered on a normal invocation of the program. The @code{set args}
command is not needed if the program does not require arguments.

@item run
The @code{run} command causes execution of the program to start from
the beginning. If the program is already running, that is to say if
you are currently positioned at a breakpoint, then a prompt will ask
for confirmation that you want to abandon the current execution and
restart.

@item breakpoint @var{location}
The breakpoint command sets a breakpoint, that is to say a point at which
execution will halt and @code{GDB} will await further
commands. @var{location} is
either a line number within a file, given in the format @code{file:linenumber},
or it is the name of a subprogram. If you request that a breakpoint be set on
a subprogram that is overloaded, a prompt will ask you to specify on which of
those subprograms you want to breakpoint. You can also
specify that all of them should be breakpointed. If the program is run
and execution encounters the breakpoint, then the program
stops and @code{GDB} signals that the breakpoint was encountered by
printing the line of code before which the program is halted.

@item catch exception @var{name}
This command causes the program execution to stop whenever exception
@var{name} is raised.  If @var{name} is omitted, then the execution is
suspended when any exception is raised.

@item print @var{expression}
This will print the value of the given expression. Most simple
Ada expression formats are properly handled by @code{GDB}, so the expression
can contain function calls, variables, operators, and attribute references.

@item continue
Continues execution following a breakpoint, until the next breakpoint or the
termination of the program.

@item step
Executes a single line after a breakpoint. If the next statement
is a subprogram call, execution continues into (the first statement of)
the called subprogram.

@item next
Executes a single line. If this line is a subprogram call, executes and
returns from the call.

@item list
Lists a few lines around the current source location. In practice, it
is usually more convenient to have a separate edit window open with the
relevant source file displayed. Successive applications of this command
print subsequent lines. The command can be given an argument which is a
line number, in which case it displays a few lines around the specified one.

@item backtrace
Displays a backtrace of the call chain. This command is typically
used after a breakpoint has occurred, to examine the sequence of calls that
leads to the current breakpoint. The display includes one line for each
activation record (frame) corresponding to an active subprogram.

@item up
At a breakpoint, @code{GDB} can display the values of variables local
to the current frame. The command @code{up} can be used to
examine the contents of other active frames, by moving the focus up
the stack, that is to say from callee to caller, one frame at a time.

@item down
Moves the focus of @code{GDB} down from the frame currently being
examined to the frame of its callee (the reverse of the previous command),

@item frame @var{n}
Inspect the frame with the given number. The value 0 denotes the frame
of the current breakpoint, that is to say the top of the call stack.

@item kill
Kills the child process in which the program is running under GDB.
This may be useful for several purposes:
@itemize @bullet
@item
It allows you to recompile and relink your program, since on many systems
you cannot regenerate an executable file while it is running in a process.
@item
You can run your program outside the debugger, on systems that do not
permit executing a program outside GDB while breakpoints are set
within GDB.
@item
It allows you to debug a core dump rather than a running process.
@end itemize
@end table

@noindent
The above list is a very short introduction to the commands that
@code{GDB} provides. Important additional capabilities, including conditional
breakpoints, the ability to execute command sequences on a breakpoint,
the ability to debug at the machine instruction level and many other
features are described in detail in @ref{Top,, Debugging with GDB, gdb,
Debugging with GDB}.  Note that most commands can be abbreviated
(for example, c for continue, bt for backtrace).

@node Using Ada Expressions
@section Using Ada Expressions
@cindex Ada expressions

@noindent
@code{GDB} supports a fairly large subset of Ada expression syntax, with some
extensions. The philosophy behind the design of this subset is

@itemize @bullet
@item
That @code{GDB} should provide basic literals and access to operations for
arithmetic, dereferencing, field selection, indexing, and subprogram calls,
leaving more sophisticated computations to subprograms written into the
program (which therefore may be called from @code{GDB}).

@item
That type safety and strict adherence to Ada language restrictions
are not particularly important to the @code{GDB} user.

@item
That brevity is important to the @code{GDB} user.
@end itemize

@noindent
Thus, for brevity, the debugger acts as if there were
implicit @code{with} and @code{use} clauses in effect for all user-written
packages, thus making it unnecessary to fully qualify most names with
their packages, regardless of context. Where this causes ambiguity,
@code{GDB} asks the user's intent.

For details on the supported Ada syntax, see @ref{Top,, Debugging with
GDB, gdb, Debugging with GDB}.

@node Calling User-Defined Subprograms
@section Calling User-Defined Subprograms

@noindent
An important capability of @code{GDB} is the ability to call user-defined
subprograms while debugging. This is achieved simply by entering
a subprogram call statement in the form:

@smallexample
call subprogram-name (parameters)
@end smallexample

@noindent
The keyword @code{call} can be omitted in the normal case where the
@code{subprogram-name} does not coincide with any of the predefined
@code{GDB} commands.

The effect is to invoke the given subprogram, passing it the
list of parameters that is supplied. The parameters can be expressions and
can include variables from the program being debugged. The
subprogram must be defined
at the library level within your program, and @code{GDB} will call the
subprogram within the environment of your program execution (which
means that the subprogram is free to access or even modify variables
within your program).

The most important use of this facility is in allowing the inclusion of
debugging routines that are tailored to particular data structures
in your program. Such debugging routines can be written to provide a suitably
high-level description of an abstract type, rather than a low-level dump
of its physical layout. After all, the standard
@code{GDB print} command only knows the physical layout of your
types, not their abstract meaning. Debugging routines can provide information
at the desired semantic level and are thus enormously useful.

For example, when debugging GNAT itself, it is crucial to have access to
the contents of the tree nodes used to represent the program internally.
But tree nodes are represented simply by an integer value (which in turn
is an index into a table of nodes).
Using the @code{print} command on a tree node would simply print this integer
value, which is not very useful. But the PN routine (defined in file
treepr.adb in the GNAT sources) takes a tree node as input, and displays
a useful high level representation of the tree node, which includes the
syntactic category of the node, its position in the source, the integers
that denote descendant nodes and parent node, as well as varied
semantic information. To study this example in more detail, you might want to
look at the body of the PN procedure in the stated file.

Another useful application of this capability is to deal with situations of
complex data which are not handled suitably by GDB. For example, if you specify
Convention Fortran for a multi-dimensional array, GDB does not know that
the ordering of array elements has been switched and will not properly
address the array elements. In such a case, instead of trying to print the
elements directly from GDB, you can write a callable procedure that prints
the elements in the desired format.

@node Using the Next Command in a Function
@section Using the Next Command in a Function

@noindent
When you use the @code{next} command in a function, the current source
location will advance to the next statement as usual. A special case
arises in the case of a @code{return} statement.

Part of the code for a return statement is the ``epilogue'' of the function.
This is the code that returns to the caller. There is only one copy of
this epilogue code, and it is typically associated with the last return
statement in the function if there is more than one return. In some
implementations, this epilogue is associated with the first statement
of the function.

The result is that if you use the @code{next} command from a return
statement that is not the last return statement of the function you
may see a strange apparent jump to the last return statement or to
the start of the function. You should simply ignore this odd jump.
The value returned is always that from the first return statement
that was stepped through.

@node Ada Exceptions
@section Stopping when Ada Exceptions are Raised
@cindex Exceptions

@noindent
You can set catchpoints that stop the program execution when your program
raises selected exceptions.

@table @code
@item catch exception
Set a catchpoint that stops execution whenever (any task in the) program
raises any exception.

@item catch exception @var{name}
Set a catchpoint that stops execution whenever (any task in the) program
raises the exception @var{name}.

@item catch exception unhandled
Set a catchpoint that stops executing whenever (any task in the) program
raises an exception for which there is no handler.

@item info exceptions
@itemx info exceptions @var{regexp}
The @code{info exceptions} command permits the user to examine all defined
exceptions within Ada programs. With a regular expression, @var{regexp}, as
argument, prints out only those exceptions whose name matches @var{regexp}.
@end table

@node Ada Tasks
@section Ada Tasks
@cindex Tasks

@noindent
@code{GDB} allows the following task-related commands:

@table @code
@item info tasks
This command shows a list of current Ada tasks, as in the following example:

@smallexample
@iftex
@leftskip=0cm
@end iftex
(gdb) info tasks
  ID       TID P-ID   Thread Pri State                 Name
   1   8088000   0   807e000  15 Child Activation Wait main_task
   2   80a4000   1   80ae000  15 Accept/Select Wait    b
   3   809a800   1   80a4800  15 Child Activation Wait a
*  4   80ae800   3   80b8000  15 Running               c
@end smallexample

@noindent
In this listing, the asterisk before the first task indicates it to be the
currently running task. The first column lists the task ID that is used
to refer to tasks in the following commands.

@item break @var{linespec} task @var{taskid}
@itemx break @var{linespec} task @var{taskid} if @dots{}
@cindex Breakpoints and tasks
These commands are like the @code{break @dots{} thread @dots{}}.
@var{linespec} specifies source lines.

Use the qualifier @samp{task @var{taskid}} with a breakpoint command
to specify that you only want @code{GDB} to stop the program when a
particular Ada task reaches this breakpoint. @var{taskid} is one of the
numeric task identifiers assigned by @code{GDB}, shown in the first
column of the @samp{info tasks} display.

If you do not specify @samp{task @var{taskid}} when you set a
breakpoint, the breakpoint applies to @emph{all} tasks of your
program.

You can use the @code{task} qualifier on conditional breakpoints as
well; in this case, place @samp{task @var{taskid}} before the
breakpoint condition (before the @code{if}).

@item task @var{taskno}
@cindex Task switching

This command allows switching to the task referred by @var{taskno}. In
particular, this allows browsing of the backtrace of the specified
task. It is advisable to switch back to the original task before
continuing execution otherwise the scheduling of the program may be
perturbed.
@end table

@noindent
For more detailed information on the tasking support,
see @ref{Top,, Debugging with GDB, gdb, Debugging with GDB}.

@node Debugging Generic Units
@section Debugging Generic Units
@cindex Debugging Generic Units
@cindex Generics

@noindent
GNAT always uses code expansion for generic instantiation. This means that
each time an instantiation occurs, a complete copy of the original code is
made, with appropriate substitutions of formals by actuals.

It is not possible to refer to the original generic entities in
@code{GDB}, but it is always possible to debug a particular instance of
a generic, by using the appropriate expanded names. For example, if we have

@smallexample @c ada
@group
@cartouche
@b{procedure} g @b{is}

   @b{generic} @b{package} k @b{is}
      @b{procedure} kp (v1 : @b{in} @b{out} integer);
   @b{end} k;

   @b{package} @b{body} k @b{is}
      @b{procedure} kp (v1 : @b{in} @b{out} integer) @b{is}
      @b{begin}
         v1 := v1 + 1;
      @b{end} kp;
   @b{end} k;

   @b{package} k1 @b{is} @b{new} k;
   @b{package} k2 @b{is} @b{new} k;

   var : integer := 1;

@b{begin}
   k1.kp (var);
   k2.kp (var);
   k1.kp (var);
   k2.kp (var);
@b{end};
@end cartouche
@end group
@end smallexample

@noindent
Then to break on a call to procedure kp in the k2 instance, simply
use the command:

@smallexample
(gdb) break g.k2.kp
@end smallexample

@noindent
When the breakpoint occurs, you can step through the code of the
instance in the normal manner and examine the values of local variables, as for
other units.

@node Remote Debugging with gdbserver
@section Remote Debugging with gdbserver
@cindex Remote Debugging with gdbserver

@noindent
On platforms where gdbserver is supported, it is possible to use this tool
to debug your application remotely.  This can be useful in situations
where the program needs to be run on a target host that is different
from the host used for development, particularly when the target has
a limited amount of resources (either CPU and/or memory).

To do so, start your program using gdbserver on the target machine.
gdbserver then automatically suspends the execution of your program
at its entry point, waiting for a debugger to connect to it.  The
following commands starts an application and tells gdbserver to
wait for a connection with the debugger on localhost port 4444.

@smallexample
$ gdbserver localhost:4444 program
Process program created; pid = 5685
Listening on port 4444
@end smallexample

Once gdbserver has started listening, we can tell the debugger to establish
a connection with this gdbserver, and then start the same debugging session
as if the program was being debugged on the same host, directly under
the control of GDB.

@smallexample
$ gdb program
(gdb) target remote targethost:4444
Remote debugging using targethost:4444
0x00007f29936d0af0 in ?? () from /lib64/ld-linux-x86-64.so.
(gdb) b foo.adb:3
Breakpoint 1 at 0x401f0c: file foo.adb, line 3.
(gdb) continue
Continuing.

Breakpoint 1, foo () at foo.adb:4
4       end foo;
@end smallexample

It is also possible to use gdbserver to attach to an already running
program, in which case the execution of that program is simply suspended
until the connection between the debugger and gdbserver is established.

For more information on how to use gdbserver, @ref{Top, Server, Using
the gdbserver Program, gdb, Debugging with GDB}.  @value{EDITION} provides support
for gdbserver on x86-linux, x86-windows and x86_64-linux.

@node GNAT Abnormal Termination or Failure to Terminate
@section GNAT Abnormal Termination or Failure to Terminate
@cindex GNAT Abnormal Termination or Failure to Terminate

@noindent
When presented with programs that contain serious errors in syntax
or semantics,
GNAT may on rare occasions  experience problems in operation, such
as aborting with a
segmentation fault or illegal memory access, raising an internal
exception, terminating abnormally, or failing to terminate at all.
In such cases, you can activate
various features of GNAT that can help you pinpoint the construct in your
program that is the likely source of the problem.

The following strategies are presented in increasing order of
difficulty, corresponding to your experience in using GNAT and your
familiarity with compiler internals.

@enumerate
@item
Run @command{gcc} with the @option{-gnatf}. This first
switch causes all errors on a given line to be reported. In its absence,
only the first error on a line is displayed.

The @option{-gnatdO} switch causes errors to be displayed as soon as they
are encountered, rather than after compilation is terminated. If GNAT
terminates prematurely or goes into an infinite loop, the last error
message displayed may help to pinpoint the culprit.

@item
Run @command{gcc} with the @option{-v (verbose)} switch. In this
mode, @command{gcc} produces ongoing information about the progress of the
compilation and provides the name of each procedure as code is
generated. This switch allows you to find which Ada procedure was being
compiled when it encountered a code generation problem.

@item
@cindex @option{-gnatdc} switch
Run @command{gcc} with the @option{-gnatdc} switch. This is a GNAT specific
switch that does for the front-end what @option{-v} does
for the back end. The system prints the name of each unit,
either a compilation unit or nested unit, as it is being analyzed.
@item
Finally, you can start
@code{gdb} directly on the @code{gnat1} executable. @code{gnat1} is the
front-end of GNAT, and can be run independently (normally it is just
called from @command{gcc}). You can use @code{gdb} on @code{gnat1} as you
would on a C program (but @pxref{The GNAT Debugger GDB} for caveats). The
@code{where} command is the first line of attack; the variable
@code{lineno} (seen by @code{print lineno}), used by the second phase of
@code{gnat1} and by the @command{gcc} backend, indicates the source line at
which the execution stopped, and @code{input_file name} indicates the name of
the source file.
@end enumerate

@node Naming Conventions for GNAT Source Files
@section Naming Conventions for GNAT Source Files

@noindent
In order to examine the workings of the GNAT system, the following
brief description of its organization may be helpful:

@itemize @bullet
@item
Files with prefix @file{sc} contain the lexical scanner.

@item
All files prefixed with @file{par} are components of the parser. The
numbers correspond to chapters of the Ada Reference Manual. For example,
parsing of select statements can be found in @file{par-ch9.adb}.

@item
All files prefixed with @file{sem} perform semantic analysis. The
numbers correspond to chapters of the Ada standard. For example, all
issues involving context clauses can be found in @file{sem_ch10.adb}. In
addition, some features of the language require sufficient special processing
to justify their own semantic files: sem_aggr for aggregates, sem_disp for
dynamic dispatching, etc.

@item
All files prefixed with @file{exp} perform normalization and
expansion of the intermediate representation (abstract syntax tree, or AST).
these files use the same numbering scheme as the parser and semantics files.
For example, the construction of record initialization procedures is done in
@file{exp_ch3.adb}.

@item
The files prefixed with @file{bind} implement the binder, which
verifies the consistency of the compilation, determines an order of
elaboration, and generates the bind file.

@item
The files @file{atree.ads} and @file{atree.adb} detail the low-level
data structures used by the front-end.

@item
The files @file{sinfo.ads} and @file{sinfo.adb} detail the structure of
the abstract syntax tree as produced by the parser.

@item
The files @file{einfo.ads} and @file{einfo.adb} detail the attributes of
all entities, computed during semantic analysis.

@item
Library management issues are dealt with in files with prefix
@file{lib}.

@item
@findex Ada
@cindex Annex A
Ada files with the prefix @file{a-} are children of @code{Ada}, as
defined in Annex A.

@item
@findex Interfaces
@cindex Annex B
Files with prefix @file{i-} are children of @code{Interfaces}, as
defined in Annex B.

@item
@findex System
Files with prefix @file{s-} are children of @code{System}. This includes
both language-defined children and GNAT run-time routines.

@item
@findex GNAT
Files with prefix @file{g-} are children of @code{GNAT}. These are useful
general-purpose packages, fully documented in their specs. All
the other @file{.c} files are modifications of common @command{gcc} files.
@end itemize

@node Getting Internal Debugging Information
@section Getting Internal Debugging Information

@noindent
Most compilers have internal debugging switches and modes. GNAT
does also, except GNAT internal debugging switches and modes are not
secret. A summary and full description of all the compiler and binder
debug flags are in the file @file{debug.adb}. You must obtain the
sources of the compiler to see the full detailed effects of these flags.

The switches that print the source of the program (reconstructed from
the internal tree) are of general interest for user programs, as are the
options to print
the full internal tree, and the entity table (the symbol table
information). The reconstructed source provides a readable version of the
program after the front-end has completed analysis and  expansion,
and is useful when studying the performance of specific constructs.
For example, constraint checks are indicated, complex aggregates
are replaced with loops and assignments, and tasking primitives
are replaced with run-time calls.

@node Stack Traceback
@section Stack Traceback
@cindex traceback
@cindex stack traceback
@cindex stack unwinding

@noindent
Traceback is a mechanism to display the sequence of subprogram calls that
leads to a specified execution point in a program. Often (but not always)
the execution point is an instruction at which an exception has been raised.
This mechanism is also known as @i{stack unwinding} because it obtains
its information by scanning the run-time stack and recovering the activation
records of all active subprograms. Stack unwinding is one of the most
important tools for program debugging.

The first entry stored in traceback corresponds to the deepest calling level,
that is to say the subprogram currently executing the instruction
from which we want to obtain the traceback.

Note that there is no runtime performance penalty when stack traceback
is enabled, and no exception is raised during program execution.

@menu
* Non-Symbolic Traceback::
* Symbolic Traceback::
@end menu

@node Non-Symbolic Traceback
@subsection Non-Symbolic Traceback
@cindex traceback, non-symbolic

@noindent
Note: this feature is not supported on all platforms. See
@file{GNAT.Traceback spec in g-traceb.ads} for a complete list of supported
platforms.

@menu
* Tracebacks From an Unhandled Exception::
* Tracebacks From Exception Occurrences (non-symbolic)::
* Tracebacks From Anywhere in a Program (non-symbolic)::
@end menu

@node Tracebacks From an Unhandled Exception
@subsubsection Tracebacks From an Unhandled Exception

@noindent
A runtime non-symbolic traceback is a list of addresses of call instructions.
To enable this feature you must use the @option{-E}
@code{gnatbind}'s option. With this option a stack traceback is stored as part
of exception information. You can retrieve this information using the
@code{addr2line} tool.

Here is a simple example:

@smallexample @c ada
@cartouche
@b{procedure} STB @b{is}

   @b{procedure} P1 @b{is}
   @b{begin}
      @b{raise} Constraint_Error;
   @b{end} P1;

   @b{procedure} P2 @b{is}
   @b{begin}
      P1;
   @b{end} P2;

@b{begin}
   P2;
@b{end} STB;
@end cartouche
@end smallexample

@smallexample
$ gnatmake stb -bargs -E
$ stb

Execution terminated by unhandled exception
Exception name: CONSTRAINT_ERROR
Message: stb.adb:5
Call stack traceback locations:
0x401373 0x40138b 0x40139c 0x401335 0x4011c4 0x4011f1 0x77e892a4
@end smallexample

@noindent
As we see the traceback lists a sequence of addresses for the unhandled
exception @code{CONSTRAINT_ERROR} raised in procedure P1. It is easy to
guess that this exception come from procedure P1. To translate these
addresses into the source lines where the calls appear, the
@code{addr2line} tool, described below, is invaluable. The use of this tool
requires the program to be compiled with debug information.

@smallexample
$ gnatmake -g stb -bargs -E
$ stb

Execution terminated by unhandled exception
Exception name: CONSTRAINT_ERROR
Message: stb.adb:5
Call stack traceback locations:
0x401373 0x40138b 0x40139c 0x401335 0x4011c4 0x4011f1 0x77e892a4

$ addr2line --exe=stb 0x401373 0x40138b 0x40139c 0x401335 0x4011c4
   0x4011f1 0x77e892a4

00401373 at d:/stb/stb.adb:5
0040138B at d:/stb/stb.adb:10
0040139C at d:/stb/stb.adb:14
00401335 at d:/stb/b~stb.adb:104
004011C4 at /build/@dots{}/crt1.c:200
004011F1 at /build/@dots{}/crt1.c:222
77E892A4 in ?? at ??:0
@end smallexample

@noindent
The @code{addr2line} tool has several other useful options:

@table @code
@item --functions
to get the function name corresponding to any location

@item --demangle=gnat
to use the gnat decoding mode for the function names. Note that
for binutils version 2.9.x the option is simply @option{--demangle}.
@end table

@smallexample
$ addr2line --exe=stb --functions --demangle=gnat 0x401373 0x40138b
   0x40139c 0x401335 0x4011c4 0x4011f1

00401373 in stb.p1 at d:/stb/stb.adb:5
0040138B in stb.p2 at d:/stb/stb.adb:10
0040139C in stb at d:/stb/stb.adb:14
00401335 in main at d:/stb/b~stb.adb:104
004011C4 in <__mingw_CRTStartup> at /build/@dots{}/crt1.c:200
004011F1 in <mainCRTStartup> at /build/@dots{}/crt1.c:222
@end smallexample

@noindent
From this traceback we can see that the exception was raised in
@file{stb.adb} at line 5, which was reached from a procedure call in
@file{stb.adb} at line 10, and so on. The @file{b~std.adb} is the binder file,
which contains the call to the main program.
@xref{Running gnatbind}. The remaining entries are assorted runtime routines,
and the output will vary from platform to platform.

It is also possible to use @code{GDB} with these traceback addresses to debug
the program. For example, we can break at a given code location, as reported
in the stack traceback:

@smallexample
$ gdb -nw stb
@noindent
Furthermore, this feature is not implemented inside Windows DLL. Only
the non-symbolic traceback is reported in this case.

(gdb) break *0x401373
Breakpoint 1 at 0x401373: file stb.adb, line 5.
@end smallexample

@noindent
It is important to note that the stack traceback addresses
do not change when debug information is included. This is particularly useful
because it makes it possible to release software without debug information (to
minimize object size), get a field report that includes a stack traceback
whenever an internal bug occurs, and then be able to retrieve the sequence
of calls with the same program compiled with debug information.

@node Tracebacks From Exception Occurrences (non-symbolic)
@subsubsection Tracebacks From Exception Occurrences

@noindent
Non-symbolic tracebacks are obtained by using the @option{-E} binder argument.
The stack traceback is attached to the exception information string, and can
be retrieved in an exception handler within the Ada program, by means of the
Ada facilities defined in @code{Ada.Exceptions}. Here is a simple example:

@smallexample @c ada
@b{with} Ada.Text_IO;
@b{with} Ada.Exceptions;

@b{procedure} STB @b{is}

   @b{use} Ada;
   @b{use} Ada.Exceptions;

   @b{procedure} P1 @b{is}
      K : Positive := 1;
   @b{begin}
      K := K - 1;
   @b{exception}
      @b{when} E : @b{others} =>
         Text_IO.Put_Line (Exception_Information (E));
   @b{end} P1;

   @b{procedure} P2 @b{is}
   @b{begin}
      P1;
   @b{end} P2;

@b{begin}
   P2;
@b{end} STB;
@end smallexample

@noindent
This program will output:

@smallexample
$ stb

Exception name: CONSTRAINT_ERROR
Message: stb.adb:12
Call stack traceback locations:
0x4015e4 0x401633 0x401644 0x401461 0x4011c4 0x4011f1 0x77e892a4
@end smallexample

@node Tracebacks From Anywhere in a Program (non-symbolic)
@subsubsection Tracebacks From Anywhere in a Program

@noindent
It is also possible to retrieve a stack traceback from anywhere in a
program. For this you need to
use the @code{GNAT.Traceback} API. This package includes a procedure called
@code{Call_Chain} that computes a complete stack traceback, as well as useful
display procedures described below. It is not necessary to use the
@option{-E gnatbind} option in this case, because the stack traceback mechanism
is invoked explicitly.

@noindent
In the following example we compute a traceback at a specific location in
the program, and we display it using @code{GNAT.Debug_Utilities.Image} to
convert addresses to strings:

@smallexample @c ada
@b{with} Ada.Text_IO;
@b{with} GNAT.Traceback;
@b{with} GNAT.Debug_Utilities;

@b{procedure} STB @b{is}

   @b{use} Ada;
   @b{use} GNAT;
   @b{use} GNAT.Traceback;

   @b{procedure} P1 @b{is}
      TB  : Tracebacks_Array (1 .. 10);
      --@i{  We are asking for a maximum of 10 stack frames.}
      Len : Natural;
      --@i{  Len will receive the actual number of stack frames returned.}
   @b{begin}
      Call_Chain (TB, Len);

      Text_IO.Put ("In STB.P1 : ");

      @b{for} K @b{in} 1 .. Len @b{loop}
         Text_IO.Put (Debug_Utilities.Image (TB (K)));
         Text_IO.Put (' ');
      @b{end} @b{loop};

      Text_IO.New_Line;
   @b{end} P1;

   @b{procedure} P2 @b{is}
   @b{begin}
      P1;
   @b{end} P2;

@b{begin}
   P2;
@b{end} STB;
@end smallexample

@smallexample
$ gnatmake -g stb
$ stb

In STB.P1 : 16#0040_F1E4# 16#0040_14F2# 16#0040_170B# 16#0040_171C#
16#0040_1461# 16#0040_11C4# 16#0040_11F1# 16#77E8_92A4#
@end smallexample

@noindent
You can then get further information by invoking the @code{addr2line}
tool as described earlier (note that the hexadecimal addresses
need to be specified in C format, with a leading ``0x'').

@node Symbolic Traceback
@subsection Symbolic Traceback
@cindex traceback, symbolic

@noindent
A symbolic traceback is a stack traceback in which procedure names are
associated with each code location.

@noindent
Note that this feature is not supported on all platforms. See
@file{GNAT.Traceback.Symbolic spec in g-trasym.ads} for a complete
list of currently supported platforms.

@noindent
Note that the symbolic traceback requires that the program be compiled
with debug information. If it is not compiled with debug information
only the non-symbolic information will be valid.

@menu
* Tracebacks From Exception Occurrences (symbolic)::
* Tracebacks From Anywhere in a Program (symbolic)::
@end menu

@node Tracebacks From Exception Occurrences (symbolic)
@subsubsection Tracebacks From Exception Occurrences

@smallexample @c ada
@b{with} Ada.Text_IO;
@b{with} GNAT.Traceback.Symbolic;

@b{procedure} STB @b{is}

   @b{procedure} P1 @b{is}
   @b{begin}
      @b{raise} Constraint_Error;
   @b{end} P1;

   @b{procedure} P2 @b{is}
   @b{begin}
      P1;
   @b{end} P2;

   @b{procedure} P3 @b{is}
   @b{begin}
      P2;
   @b{end} P3;

@b{begin}
   P3;
@b{exception}
   @b{when} E : @b{others} =>
      Ada.Text_IO.Put_Line (GNAT.Traceback.Symbolic.Symbolic_Traceback (E));
@b{end} STB;
@end smallexample

@smallexample
$ gnatmake -g .\stb -bargs -E
$ stb

0040149F in stb.p1 at stb.adb:8
004014B7 in stb.p2 at stb.adb:13
004014CF in stb.p3 at stb.adb:18
004015DD in ada.stb at stb.adb:22
00401461 in main at b~stb.adb:168
004011C4 in __mingw_CRTStartup at crt1.c:200
004011F1 in mainCRTStartup at crt1.c:222
77E892A4 in ?? at ??:0
@end smallexample

@noindent
In the above example the ``.\'' syntax in the @command{gnatmake} command
is currently required by @command{addr2line} for files that are in
the current working directory.
Moreover, the exact sequence of linker options may vary from platform
to platform.
The above @option{-largs} section is for Windows platforms. By contrast,
under Unix there is no need for the @option{-largs} section.
Differences across platforms are due to details of linker implementation.

@node Tracebacks From Anywhere in a Program (symbolic)
@subsubsection Tracebacks From Anywhere in a Program

@noindent
It is possible to get a symbolic stack traceback
from anywhere in a program, just as for non-symbolic tracebacks.
The first step is to obtain a non-symbolic
traceback, and then call @code{Symbolic_Traceback} to compute the symbolic
information. Here is an example:

@smallexample @c ada
@b{with} Ada.Text_IO;
@b{with} GNAT.Traceback;
@b{with} GNAT.Traceback.Symbolic;

@b{procedure} STB @b{is}

   @b{use} Ada;
   @b{use} GNAT.Traceback;
   @b{use} GNAT.Traceback.Symbolic;

   @b{procedure} P1 @b{is}
      TB  : Tracebacks_Array (1 .. 10);
      --@i{  We are asking for a maximum of 10 stack frames.}
      Len : Natural;
      --@i{  Len will receive the actual number of stack frames returned.}
   @b{begin}
      Call_Chain (TB, Len);
      Text_IO.Put_Line (Symbolic_Traceback (TB (1 .. Len)));
   @b{end} P1;

   @b{procedure} P2 @b{is}
   @b{begin}
      P1;
   @b{end} P2;

@b{begin}
   P2;
@b{end} STB;
@end smallexample

@c ******************************

@c **************************************
@node Platform-Specific Information for the Run-Time Libraries
@appendix Platform-Specific Information for the Run-Time Libraries
@cindex Tasking and threads libraries
@cindex Threads libraries and tasking
@cindex Run-time libraries (platform-specific information)

@noindent
The GNAT run-time implementation may vary with respect to both the
underlying threads library and the exception handling scheme.
For threads support, one or more of the following are supplied:
@itemize @bullet
@item @b{native threads library}, a binding to the thread package from
the underlying operating system

@item @b{pthreads library} (Sparc Solaris only), a binding to the Solaris
POSIX thread package
@end itemize

@noindent
For exception handling, either or both of two models are supplied:
@itemize @bullet
@item @b{Zero-Cost Exceptions} (``ZCX''),@footnote{
Most programs should experience a substantial speed improvement by
being compiled with a ZCX run-time.
This is especially true for
tasking applications or applications with many exception handlers.}
@cindex Zero-Cost Exceptions
@cindex ZCX (Zero-Cost Exceptions)
which uses binder-generated tables that
are interrogated at run time to locate a handler

@item @b{setjmp / longjmp} (``SJLJ''),
@cindex setjmp/longjmp Exception Model
@cindex SJLJ (setjmp/longjmp Exception Model)
which uses dynamically-set data to establish
the set of handlers
@end itemize

@noindent
This appendix summarizes which combinations of threads and exception support
are supplied on various GNAT platforms.
It then shows how to select a particular library either
permanently or temporarily,
explains the properties of (and tradeoffs among) the various threads
libraries, and provides some additional
information about several specific platforms.

@menu
* Summary of Run-Time Configurations::
* Specifying a Run-Time Library::
* Choosing the Scheduling Policy::
* Solaris-Specific Considerations::
* Linux-Specific Considerations::
* AIX-Specific Considerations::
* RTX-Specific Considerations::
* HP-UX-Specific Considerations::
@end menu

@node Summary of Run-Time Configurations
@section Summary of Run-Time Configurations

@multitable @columnfractions .30 .70
@item @b{alpha-openvms}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native VMS threads
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-sjlj}
@item @code{@ @ @ @ }Tasking    @tab native TRU64 threads
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{ia64-hp_linux}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab pthread library
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @b{ia64-hpux}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native HP-UX threads
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{ia64-openvms}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native VMS threads
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @b{ia64-sgi_linux}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab pthread library
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @b{pa-hpux}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native HP-UX threads
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-sjlj}
@item @code{@ @ @ @ }Tasking    @tab native HP-UX threads
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{ppc-aix}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native AIX threads
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-sjlj}
@item @code{@ @ @ @ }Tasking    @tab native AIX threads
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{ppc-darwin}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native MacOS threads
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @b{sparc-solaris}  @tab
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native Solaris threads library
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-pthread}
@item @code{@ @ @ @ }Tasking    @tab pthread library
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-sjlj}
@item @code{@ @ @ @ }Tasking    @tab native Solaris threads library
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{sparc64-solaris}  @tab
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native Solaris threads library
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @b{x86-linux}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab pthread library
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-sjlj}
@item @code{@ @ @ @ }Tasking    @tab pthread library
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{x86-lynx}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native LynxOS threads
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{x86-solaris}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native Solaris threads
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-sjlj}
@item @code{@ @ @ @ }Tasking    @tab native Solaris threads library
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{x86-windows}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab native Win32 threads
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-sjlj}
@item @code{@ @ @ @ }Tasking    @tab native Win32 threads
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @b{x86-windows-rtx}
@item @code{@ @ }@i{rts-rtx-rtss (default)}
@item @code{@ @ @ @ }Tasking    @tab RTX real-time subsystem RTSS threads (kernel mode)
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@item @code{@ @ }@i{rts-rtx-w32}
@item @code{@ @ @ @ }Tasking    @tab RTX Win32 threads (user mode)
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @b{x86_64-linux}
@item @code{@ @ }@i{rts-native (default)}
@item @code{@ @ @ @ }Tasking    @tab pthread library
@item @code{@ @ @ @ }Exceptions @tab ZCX
@*
@item @code{@ @ }@i{rts-sjlj}
@item @code{@ @ @ @ }Tasking    @tab pthread library
@item @code{@ @ @ @ }Exceptions @tab SJLJ
@*
@end multitable

@node Specifying a Run-Time Library
@section Specifying a Run-Time Library

@noindent
The @file{adainclude} subdirectory containing the sources of the GNAT
run-time library, and the @file{adalib} subdirectory containing the
@file{ALI} files and the static and/or shared GNAT library, are located
in the gcc target-dependent area:

@smallexample
target=$prefix/lib/gcc/gcc-@i{dumpmachine}/gcc-@i{dumpversion}/
@end smallexample

@noindent
As indicated above, on some platforms several run-time libraries are supplied.
These libraries are installed in the target dependent area and
contain a complete source and binary subdirectory. The detailed description
below explains the differences between the different libraries in terms of
their thread support.

The default run-time library (when GNAT is installed) is @emph{rts-native}.
This default run time is selected by the means of soft links.
For example on x86-linux:

@smallexample
@group
 $(target-dir)
     |
     +--- adainclude----------+
     |                        |
     +--- adalib-----------+  |
     |                     |  |
     +--- rts-native       |  |
     |    |                |  |
     |    +--- adainclude <---+
     |    |                |
     |    +--- adalib <----+
     |
     +--- rts-sjlj
          |
          +--- adainclude
          |
          +--- adalib
@end group
@end smallexample

@noindent
If the @i{rts-sjlj} library is to be selected on a permanent basis,
these soft links can be modified with the following commands:

@smallexample
$ cd $target
$ rm -f adainclude adalib
$ ln -s rts-sjlj/adainclude adainclude
$ ln -s rts-sjlj/adalib adalib
@end smallexample

@noindent
Alternatively, you can specify @file{rts-sjlj/adainclude} in the file
@file{$target/ada_source_path} and @file{rts-sjlj/adalib} in
@file{$target/ada_object_path}.

Selecting another run-time library temporarily can be
achieved by using the @option{--RTS} switch, e.g., @option{--RTS=sjlj}
@cindex @option{--RTS} option

@node Choosing the Scheduling Policy
@section Choosing the Scheduling Policy

@noindent
When using a POSIX threads implementation, you have a choice of several
scheduling policies: @code{SCHED_FIFO},
@cindex @code{SCHED_FIFO} scheduling policy
@code{SCHED_RR}
@cindex @code{SCHED_RR} scheduling policy
and @code{SCHED_OTHER}.
@cindex @code{SCHED_OTHER} scheduling policy
Typically, the default is @code{SCHED_OTHER}, while using @code{SCHED_FIFO}
or @code{SCHED_RR} requires special (e.g., root) privileges.

By default, GNAT uses the @code{SCHED_OTHER} policy. To specify
@code{SCHED_FIFO},
@cindex @code{SCHED_FIFO} scheduling policy
you can use one of the following:

@itemize @bullet
@item
@code{pragma Time_Slice (0.0)}
@cindex pragma Time_Slice
@item
the corresponding binder option @option{-T0}
@cindex @option{-T0} option
@item
@code{pragma Task_Dispatching_Policy (FIFO_Within_Priorities)}
@cindex pragma Task_Dispatching_Policy
@end itemize

@noindent
To specify @code{SCHED_RR},
@cindex @code{SCHED_RR} scheduling policy
you should use @code{pragma Time_Slice} with a
value greater than @code{0.0}, or else use the corresponding @option{-T}
binder option.

@node Solaris-Specific Considerations
@section Solaris-Specific Considerations
@cindex Solaris Sparc threads libraries

@noindent
This section addresses some topics related to the various threads libraries
on Sparc Solaris.

@menu
* Solaris Threads Issues::
@end menu

@node Solaris Threads Issues
@subsection Solaris Threads Issues

@noindent
GNAT under Solaris/Sparc 32 bits comes with an alternate tasking run-time
library based on POSIX threads --- @emph{rts-pthread}.
@cindex rts-pthread threads library
This run-time library has the advantage of being mostly shared across all
POSIX-compliant thread implementations, and it also provides under
@w{Solaris 8} the @code{PTHREAD_PRIO_INHERIT}
@cindex @code{PTHREAD_PRIO_INHERIT} policy (under rts-pthread)
and @code{PTHREAD_PRIO_PROTECT}
@cindex @code{PTHREAD_PRIO_PROTECT} policy (under rts-pthread)
semantics that can be selected using the predefined pragma
@code{Locking_Policy}
@cindex pragma Locking_Policy (under rts-pthread)
with respectively
@code{Inheritance_Locking} and @code{Ceiling_Locking} as the policy.
@cindex @code{Inheritance_Locking} (under rts-pthread)
@cindex @code{Ceiling_Locking} (under rts-pthread)

As explained above, the native run-time library is based on the Solaris thread
library (@code{libthread}) and is the default library.

When the Solaris threads library is used (this is the default), programs
compiled with GNAT can automatically take advantage of
and can thus execute on multiple processors.
The user can alternatively specify a processor on which the program should run
to emulate a single-processor system. The multiprocessor / uniprocessor choice
is made by
setting the environment variable @env{GNAT_PROCESSOR}
@cindex @env{GNAT_PROCESSOR} environment variable (on Sparc Solaris)
to one of the following:

@table @code
@item -2
Use the default configuration (run the program on all
available processors) - this is the same as having @code{GNAT_PROCESSOR}
unset

@item -1
Let the run-time implementation choose one processor and run the program on
that processor

@item 0 .. Last_Proc
Run the program on the specified processor.
@code{Last_Proc} is equal to @code{_SC_NPROCESSORS_CONF - 1}
(where @code{_SC_NPROCESSORS_CONF} is a system variable).
@end table

@node Linux-Specific Considerations
@section Linux-Specific Considerations
@cindex Linux threads libraries

@noindent
On GNU/Linux without NPTL support (usually system with GNU C Library
older than 2.3), the signal model is not POSIX compliant, which means
that to send a signal to the process, you need to send the signal to all
threads, e.g.@: by using @code{killpg()}.

@node AIX-Specific Considerations
@section AIX-Specific Considerations
@cindex AIX resolver library

@noindent
On AIX, the resolver library initializes some internal structure on
the first call to @code{get*by*} functions, which are used to implement
@code{GNAT.Sockets.Get_Host_By_Name} and
@code{GNAT.Sockets.Get_Host_By_Address}.
If such initialization occurs within an Ada task, and the stack size for
the task is the default size, a stack overflow may occur.

To avoid this overflow, the user should either ensure that the first call
to @code{GNAT.Sockets.Get_Host_By_Name} or
@code{GNAT.Sockets.Get_Host_By_Addrss}
occurs in the environment task, or use @code{pragma Storage_Size} to
specify a sufficiently large size for the stack of the task that contains
this call.

@node RTX-Specific Considerations
@section RTX-Specific Considerations
@cindex RTX libraries

@noindent
The Real-time Extension (RTX) to Windows is based on the Windows Win32
API. Applications can be built to work in two different modes:

@itemize @bullet
@item
Windows executables that run in Ring 3 to utilize memory protection
(@emph{rts-rtx-w32}).

@item
Real-time subsystem (RTSS) executables that run in Ring 0, where
performance can be optimized with RTSS applications taking precedent
over all Windows applications (@emph{rts-rtx-rtss}). This mode requires
the Microsoft linker to handle RTSS libraries.

@end itemize

@node HP-UX-Specific Considerations
@section HP-UX-Specific Considerations
@cindex HP-UX Scheduling

@noindent
On HP-UX, appropriate privileges are required to change the scheduling
parameters of a task. The calling process must have appropriate
privileges or be a member of a group having @code{PRIV_RTSCHED} access to
successfully change the scheduling parameters.

By default, GNAT uses the @code{SCHED_HPUX} policy. To have access to the
priority range 0-31 either the @code{FIFO_Within_Priorities} or the
@code{Round_Robin_Within_Priorities} scheduling policies need to be set.

To specify the @code{FIFO_Within_Priorities} scheduling policy you can use
one of the following:

@itemize @bullet
@item
@code{pragma Time_Slice (0.0)}
@cindex pragma Time_Slice
@item
the corresponding binder option @option{-T0}
@cindex @option{-T0} option
@item
@code{pragma Task_Dispatching_Policy (FIFO_Within_Priorities)}
@cindex pragma Task_Dispatching_Policy
@end itemize

@noindent
To specify the @code{Round_Robin_Within_Priorities}, scheduling policy
you should use @code{pragma Time_Slice} with a
value greater than @code{0.0}, or use the corresponding @option{-T}
binder option, or set the @code{pragma Task_Dispatching_Policy
(Round_Robin_Within_Priorities)}.

@c *******************************
@node Example of Binder Output File
@appendix Example of Binder Output File

@noindent
This Appendix displays the source code for @command{gnatbind}'s output
file generated for a simple ``Hello World'' program.
Comments have been added for clarification purposes.

@smallexample @c adanocomment
@iftex
@leftskip=0cm
@end iftex
--  The package is called Ada_Main unless this name is actually used
--  as a unit name in the partition, in which case some other unique
--  name is used.

@b{with} System;
@b{package} ada_main @b{is}

   Elab_Final_Code : Integer;
   @b{pragma} Import (C, Elab_Final_Code, "__gnat_inside_elab_final_code");

   --  The main program saves the parameters (argument count,
   --  argument values, environment pointer) in global variables
   --  for later access by other units including
   --  Ada.Command_Line.

   gnat_argc : Integer;
   gnat_argv : System.Address;
   gnat_envp : System.Address;

   --  The actual variables are stored in a library routine. This
   --  is useful for some shared library situations, where there
   --  are problems if variables are not in the library.

   @b{pragma} Import (C, gnat_argc);
   @b{pragma} Import (C, gnat_argv);
   @b{pragma} Import (C, gnat_envp);

   --  The exit status is similarly an external location

   gnat_exit_status : Integer;
   @b{pragma} Import (C, gnat_exit_status);

   GNAT_Version : @b{constant} String :=
                    "GNAT Version: 6.0.0w (20061115)";
   @b{pragma} Export (C, GNAT_Version, "__gnat_version");

   --  This is the generated adafinal routine that performs
   --  finalization at the end of execution. In the case where
   --  Ada is the main program, this main program makes a call
   --  to adafinal at program termination.

   @b{procedure} adafinal;
   @b{pragma} Export (C, adafinal, "adafinal");

   --  This is the generated adainit routine that performs
   --  initialization at the start of execution. In the case
   --  where Ada is the main program, this main program makes
   --  a call to adainit at program startup.

   @b{procedure} adainit;
   @b{pragma} Export (C, adainit, "adainit");

   --  This routine is called at the start of execution. It is
   --  a dummy routine that is used by the debugger to breakpoint
   --  at the start of execution.

   @b{procedure} Break_Start;
   @b{pragma} Import (C, Break_Start, "__gnat_break_start");

   --  This is the actual generated main program (it would be
   --  suppressed if the no main program switch were used). As
   --  required by standard system conventions, this program has
   --  the external name main.

   @b{function} main
     (argc : Integer;
      argv : System.Address;
      envp : System.Address)
      @b{return} Integer;
   @b{pragma} Export (C, main, "main");

   --  The following set of constants give the version
   --  identification values for every unit in the bound
   --  partition. This identification is computed from all
   --  dependent semantic units, and corresponds to the
   --  string that would be returned by use of the
   --  Body_Version or Version attributes.

   @b{type} Version_32 @b{is} @b{mod} 2 ** 32;
   u00001 : @b{constant} Version_32 := 16#7880BEB3#;
   u00002 : @b{constant} Version_32 := 16#0D24CBD0#;
   u00003 : @b{constant} Version_32 := 16#3283DBEB#;
   u00004 : @b{constant} Version_32 := 16#2359F9ED#;
   u00005 : @b{constant} Version_32 := 16#664FB847#;
   u00006 : @b{constant} Version_32 := 16#68E803DF#;
   u00007 : @b{constant} Version_32 := 16#5572E604#;
   u00008 : @b{constant} Version_32 := 16#46B173D8#;
   u00009 : @b{constant} Version_32 := 16#156A40CF#;
   u00010 : @b{constant} Version_32 := 16#033DABE0#;
   u00011 : @b{constant} Version_32 := 16#6AB38FEA#;
   u00012 : @b{constant} Version_32 := 16#22B6217D#;
   u00013 : @b{constant} Version_32 := 16#68A22947#;
   u00014 : @b{constant} Version_32 := 16#18CC4A56#;
   u00015 : @b{constant} Version_32 := 16#08258E1B#;
   u00016 : @b{constant} Version_32 := 16#367D5222#;
   u00017 : @b{constant} Version_32 := 16#20C9ECA4#;
   u00018 : @b{constant} Version_32 := 16#50D32CB6#;
   u00019 : @b{constant} Version_32 := 16#39A8BB77#;
   u00020 : @b{constant} Version_32 := 16#5CF8FA2B#;
   u00021 : @b{constant} Version_32 := 16#2F1EB794#;
   u00022 : @b{constant} Version_32 := 16#31AB6444#;
   u00023 : @b{constant} Version_32 := 16#1574B6E9#;
   u00024 : @b{constant} Version_32 := 16#5109C189#;
   u00025 : @b{constant} Version_32 := 16#56D770CD#;
   u00026 : @b{constant} Version_32 := 16#02F9DE3D#;
   u00027 : @b{constant} Version_32 := 16#08AB6B2C#;
   u00028 : @b{constant} Version_32 := 16#3FA37670#;
   u00029 : @b{constant} Version_32 := 16#476457A0#;
   u00030 : @b{constant} Version_32 := 16#731E1B6E#;
   u00031 : @b{constant} Version_32 := 16#23C2E789#;
   u00032 : @b{constant} Version_32 := 16#0F1BD6A1#;
   u00033 : @b{constant} Version_32 := 16#7C25DE96#;
   u00034 : @b{constant} Version_32 := 16#39ADFFA2#;
   u00035 : @b{constant} Version_32 := 16#571DE3E7#;
   u00036 : @b{constant} Version_32 := 16#5EB646AB#;
   u00037 : @b{constant} Version_32 := 16#4249379B#;
   u00038 : @b{constant} Version_32 := 16#0357E00A#;
   u00039 : @b{constant} Version_32 := 16#3784FB72#;
   u00040 : @b{constant} Version_32 := 16#2E723019#;
   u00041 : @b{constant} Version_32 := 16#623358EA#;
   u00042 : @b{constant} Version_32 := 16#107F9465#;
   u00043 : @b{constant} Version_32 := 16#6843F68A#;
   u00044 : @b{constant} Version_32 := 16#63305874#;
   u00045 : @b{constant} Version_32 := 16#31E56CE1#;
   u00046 : @b{constant} Version_32 := 16#02917970#;
   u00047 : @b{constant} Version_32 := 16#6CCBA70E#;
   u00048 : @b{constant} Version_32 := 16#41CD4204#;
   u00049 : @b{constant} Version_32 := 16#572E3F58#;
   u00050 : @b{constant} Version_32 := 16#20729FF5#;
   u00051 : @b{constant} Version_32 := 16#1D4F93E8#;
   u00052 : @b{constant} Version_32 := 16#30B2EC3D#;
   u00053 : @b{constant} Version_32 := 16#34054F96#;
   u00054 : @b{constant} Version_32 := 16#5A199860#;
   u00055 : @b{constant} Version_32 := 16#0E7F912B#;
   u00056 : @b{constant} Version_32 := 16#5760634A#;
   u00057 : @b{constant} Version_32 := 16#5D851835#;

   --  The following Export pragmas export the version numbers
   --  with symbolic names ending in B (for body) or S
   --  (for spec) so that they can be located in a link. The
   --  information provided here is sufficient to track down
   --  the exact versions of units used in a given build.

   @b{pragma} Export (C, u00001, "helloB");
   @b{pragma} Export (C, u00002, "system__standard_libraryB");
   @b{pragma} Export (C, u00003, "system__standard_libraryS");
   @b{pragma} Export (C, u00004, "adaS");
   @b{pragma} Export (C, u00005, "ada__text_ioB");
   @b{pragma} Export (C, u00006, "ada__text_ioS");
   @b{pragma} Export (C, u00007, "ada__exceptionsB");
   @b{pragma} Export (C, u00008, "ada__exceptionsS");
   @b{pragma} Export (C, u00009, "gnatS");
   @b{pragma} Export (C, u00010, "gnat__heap_sort_aB");
   @b{pragma} Export (C, u00011, "gnat__heap_sort_aS");
   @b{pragma} Export (C, u00012, "systemS");
   @b{pragma} Export (C, u00013, "system__exception_tableB");
   @b{pragma} Export (C, u00014, "system__exception_tableS");
   @b{pragma} Export (C, u00015, "gnat__htableB");
   @b{pragma} Export (C, u00016, "gnat__htableS");
   @b{pragma} Export (C, u00017, "system__exceptionsS");
   @b{pragma} Export (C, u00018, "system__machine_state_operationsB");
   @b{pragma} Export (C, u00019, "system__machine_state_operationsS");
   @b{pragma} Export (C, u00020, "system__machine_codeS");
   @b{pragma} Export (C, u00021, "system__storage_elementsB");
   @b{pragma} Export (C, u00022, "system__storage_elementsS");
   @b{pragma} Export (C, u00023, "system__secondary_stackB");
   @b{pragma} Export (C, u00024, "system__secondary_stackS");
   @b{pragma} Export (C, u00025, "system__parametersB");
   @b{pragma} Export (C, u00026, "system__parametersS");
   @b{pragma} Export (C, u00027, "system__soft_linksB");
   @b{pragma} Export (C, u00028, "system__soft_linksS");
   @b{pragma} Export (C, u00029, "system__stack_checkingB");
   @b{pragma} Export (C, u00030, "system__stack_checkingS");
   @b{pragma} Export (C, u00031, "system__tracebackB");
   @b{pragma} Export (C, u00032, "system__tracebackS");
   @b{pragma} Export (C, u00033, "ada__streamsS");
   @b{pragma} Export (C, u00034, "ada__tagsB");
   @b{pragma} Export (C, u00035, "ada__tagsS");
   @b{pragma} Export (C, u00036, "system__string_opsB");
   @b{pragma} Export (C, u00037, "system__string_opsS");
   @b{pragma} Export (C, u00038, "interfacesS");
   @b{pragma} Export (C, u00039, "interfaces__c_streamsB");
   @b{pragma} Export (C, u00040, "interfaces__c_streamsS");
   @b{pragma} Export (C, u00041, "system__file_ioB");
   @b{pragma} Export (C, u00042, "system__file_ioS");
   @b{pragma} Export (C, u00043, "ada__finalizationB");
   @b{pragma} Export (C, u00044, "ada__finalizationS");
   @b{pragma} Export (C, u00045, "system__finalization_rootB");
   @b{pragma} Export (C, u00046, "system__finalization_rootS");
   @b{pragma} Export (C, u00047, "system__finalization_implementationB");
   @b{pragma} Export (C, u00048, "system__finalization_implementationS");
   @b{pragma} Export (C, u00049, "system__string_ops_concat_3B");
   @b{pragma} Export (C, u00050, "system__string_ops_concat_3S");
   @b{pragma} Export (C, u00051, "system__stream_attributesB");
   @b{pragma} Export (C, u00052, "system__stream_attributesS");
   @b{pragma} Export (C, u00053, "ada__io_exceptionsS");
   @b{pragma} Export (C, u00054, "system__unsigned_typesS");
   @b{pragma} Export (C, u00055, "system__file_control_blockS");
   @b{pragma} Export (C, u00056, "ada__finalization__list_controllerB");
   @b{pragma} Export (C, u00057, "ada__finalization__list_controllerS");

   -- BEGIN ELABORATION ORDER
   -- ada (spec)
   -- gnat (spec)
   -- gnat.heap_sort_a (spec)
   -- gnat.heap_sort_a (body)
   -- gnat.htable (spec)
   -- gnat.htable (body)
   -- interfaces (spec)
   -- system (spec)
   -- system.machine_code (spec)
   -- system.parameters (spec)
   -- system.parameters (body)
   -- interfaces.c_streams (spec)
   -- interfaces.c_streams (body)
   -- system.standard_library (spec)
   -- ada.exceptions (spec)
   -- system.exception_table (spec)
   -- system.exception_table (body)
   -- ada.io_exceptions (spec)
   -- system.exceptions (spec)
   -- system.storage_elements (spec)
   -- system.storage_elements (body)
   -- system.machine_state_operations (spec)
   -- system.machine_state_operations (body)
   -- system.secondary_stack (spec)
   -- system.stack_checking (spec)
   -- system.soft_links (spec)
   -- system.soft_links (body)
   -- system.stack_checking (body)
   -- system.secondary_stack (body)
   -- system.standard_library (body)
   -- system.string_ops (spec)
   -- system.string_ops (body)
   -- ada.tags (spec)
   -- ada.tags (body)
   -- ada.streams (spec)
   -- system.finalization_root (spec)
   -- system.finalization_root (body)
   -- system.string_ops_concat_3 (spec)
   -- system.string_ops_concat_3 (body)
   -- system.traceback (spec)
   -- system.traceback (body)
   -- ada.exceptions (body)
   -- system.unsigned_types (spec)
   -- system.stream_attributes (spec)
   -- system.stream_attributes (body)
   -- system.finalization_implementation (spec)
   -- system.finalization_implementation (body)
   -- ada.finalization (spec)
   -- ada.finalization (body)
   -- ada.finalization.list_controller (spec)
   -- ada.finalization.list_controller (body)
   -- system.file_control_block (spec)
   -- system.file_io (spec)
   -- system.file_io (body)
   -- ada.text_io (spec)
   -- ada.text_io (body)
   -- hello (body)
   -- END ELABORATION ORDER

@b{end} ada_main;

--  The following source file name pragmas allow the generated file
--  names to be unique for different main programs. They are needed
--  since the package name will always be Ada_Main.

@b{pragma} Source_File_Name (ada_main, Spec_File_Name => "b~hello.ads");
@b{pragma} Source_File_Name (ada_main, Body_File_Name => "b~hello.adb");

--  Generated package body for Ada_Main starts here

@b{package} @b{body} ada_main @b{is}

   --  The actual finalization is performed by calling the
   --  library routine in System.Standard_Library.Adafinal

   @b{procedure} Do_Finalize;
   @b{pragma} Import (C, Do_Finalize, "system__standard_library__adafinal");

   -------------
   -- adainit --
   -------------

@findex adainit
   @b{procedure} adainit @b{is}

      --  These booleans are set to True once the associated unit has
      --  been elaborated. It is also used to avoid elaborating the
      --  same unit twice.

      E040 : Boolean;
      @b{pragma} Import (Ada, E040, "interfaces__c_streams_E");

      E008 : Boolean;
      @b{pragma} Import (Ada, E008, "ada__exceptions_E");

      E014 : Boolean;
      @b{pragma} Import (Ada, E014, "system__exception_table_E");

      E053 : Boolean;
      @b{pragma} Import (Ada, E053, "ada__io_exceptions_E");

      E017 : Boolean;
      @b{pragma} Import (Ada, E017, "system__exceptions_E");

      E024 : Boolean;
      @b{pragma} Import (Ada, E024, "system__secondary_stack_E");

      E030 : Boolean;
      @b{pragma} Import (Ada, E030, "system__stack_checking_E");

      E028 : Boolean;
      @b{pragma} Import (Ada, E028, "system__soft_links_E");

      E035 : Boolean;
      @b{pragma} Import (Ada, E035, "ada__tags_E");

      E033 : Boolean;
      @b{pragma} Import (Ada, E033, "ada__streams_E");

      E046 : Boolean;
      @b{pragma} Import (Ada, E046, "system__finalization_root_E");

      E048 : Boolean;
      @b{pragma} Import (Ada, E048, "system__finalization_implementation_E");

      E044 : Boolean;
      @b{pragma} Import (Ada, E044, "ada__finalization_E");

      E057 : Boolean;
      @b{pragma} Import (Ada, E057, "ada__finalization__list_controller_E");

      E055 : Boolean;
      @b{pragma} Import (Ada, E055, "system__file_control_block_E");

      E042 : Boolean;
      @b{pragma} Import (Ada, E042, "system__file_io_E");

      E006 : Boolean;
      @b{pragma} Import (Ada, E006, "ada__text_io_E");

      --  Set_Globals is a library routine that stores away the
      --  value of the indicated set of global values in global
      --  variables within the library.

      @b{procedure} Set_Globals
        (Main_Priority            : Integer;
         Time_Slice_Value         : Integer;
         WC_Encoding              : Character;
         Locking_Policy           : Character;
         Queuing_Policy           : Character;
         Task_Dispatching_Policy  : Character;
         Adafinal                 : System.Address;
         Unreserve_All_Interrupts : Integer;
         Exception_Tracebacks     : Integer);
@findex __gnat_set_globals
      @b{pragma} Import (C, Set_Globals, "__gnat_set_globals");

      --  SDP_Table_Build is a library routine used to build the
      --  exception tables. See unit Ada.Exceptions in files
      --  a-except.ads/adb for full details of how zero cost
      --  exception handling works. This procedure, the call to
      --  it, and the two following tables are all omitted if the
      --  build is in longjmp/setjmp exception mode.

@findex SDP_Table_Build
@findex Zero Cost Exceptions
      @b{procedure} SDP_Table_Build
        (SDP_Addresses   : System.Address;
         SDP_Count       : Natural;
         Elab_Addresses  : System.Address;
         Elab_Addr_Count : Natural);
      @b{pragma} Import (C, SDP_Table_Build, "__gnat_SDP_Table_Build");

      --  Table of Unit_Exception_Table addresses. Used for zero
      --  cost exception handling to build the top level table.

      ST : @b{aliased} @b{constant} @b{array} (1 .. 23) @b{of} System.Address := (
        Hello'UET_Address,
        Ada.Text_Io'UET_Address,
        Ada.Exceptions'UET_Address,
        Gnat.Heap_Sort_A'UET_Address,
        System.Exception_Table'UET_Address,
        System.Machine_State_Operations'UET_Address,
        System.Secondary_Stack'UET_Address,
        System.Parameters'UET_Address,
        System.Soft_Links'UET_Address,
        System.Stack_Checking'UET_Address,
        System.Traceback'UET_Address,
        Ada.Streams'UET_Address,
        Ada.Tags'UET_Address,
        System.String_Ops'UET_Address,
        Interfaces.C_Streams'UET_Address,
        System.File_Io'UET_Address,
        Ada.Finalization'UET_Address,
        System.Finalization_Root'UET_Address,
        System.Finalization_Implementation'UET_Address,
        System.String_Ops_Concat_3'UET_Address,
        System.Stream_Attributes'UET_Address,
        System.File_Control_Block'UET_Address,
        Ada.Finalization.List_Controller'UET_Address);

      --  Table of addresses of elaboration routines. Used for
      --  zero cost exception handling to make sure these
      --  addresses are included in the top level procedure
      --  address table.

      EA : @b{aliased} @b{constant} @b{array} (1 .. 23) @b{of} System.Address := (
        adainit'Code_Address,
        Do_Finalize'Code_Address,
        Ada.Exceptions'Elab_Spec'Address,
        System.Exceptions'Elab_Spec'Address,
        Interfaces.C_Streams'Elab_Spec'Address,
        System.Exception_Table'Elab_Body'Address,
        Ada.Io_Exceptions'Elab_Spec'Address,
        System.Stack_Checking'Elab_Spec'Address,
        System.Soft_Links'Elab_Body'Address,
        System.Secondary_Stack'Elab_Body'Address,
        Ada.Tags'Elab_Spec'Address,
        Ada.Tags'Elab_Body'Address,
        Ada.Streams'Elab_Spec'Address,
        System.Finalization_Root'Elab_Spec'Address,
        Ada.Exceptions'Elab_Body'Address,
        System.Finalization_Implementation'Elab_Spec'Address,
        System.Finalization_Implementation'Elab_Body'Address,
        Ada.Finalization'Elab_Spec'Address,
        Ada.Finalization.List_Controller'Elab_Spec'Address,
        System.File_Control_Block'Elab_Spec'Address,
        System.File_Io'Elab_Body'Address,
        Ada.Text_Io'Elab_Spec'Address,
        Ada.Text_Io'Elab_Body'Address);

   --  Start of processing for adainit

   @b{begin}

      --  Call SDP_Table_Build to build the top level procedure
      --  table for zero cost exception handling (omitted in
      --  longjmp/setjmp mode).

      SDP_Table_Build (ST'Address, 23, EA'Address, 23);

      --  Call Set_Globals to record various information for
      --  this partition.  The values are derived by the binder
      --  from information stored in the ali files by the compiler.

@findex __gnat_set_globals
      Set_Globals
        (Main_Priority            => -1,
         --  Priority of main program, -1 if no pragma Priority used

         Time_Slice_Value         => -1,
         --  Time slice from Time_Slice pragma, -1 if none used

         WC_Encoding              => 'b',
         --  Wide_Character encoding used, default is brackets

         Locking_Policy           => ' ',
         --  Locking_Policy used, default of space means not
         --  specified, otherwise it is the first character of
         --  the policy name.

         Queuing_Policy           => ' ',
         --  Queuing_Policy used, default of space means not
         --  specified, otherwise it is the first character of
         --  the policy name.

         Task_Dispatching_Policy  => ' ',
         --  Task_Dispatching_Policy used, default of space means
         --  not specified, otherwise first character of the
         --  policy name.

         Adafinal                 => System.Null_Address,
         --  Address of Adafinal routine, not used anymore

         Unreserve_All_Interrupts => 0,
         --  Set true if pragma Unreserve_All_Interrupts was used

         Exception_Tracebacks     => 0);
         --  Indicates if exception tracebacks are enabled

      Elab_Final_Code := 1;

      --  Now we have the elaboration calls for all units in the partition.
      --  The Elab_Spec and Elab_Body attributes generate references to the
      --  implicit elaboration procedures generated by the compiler for
      --  each unit that requires elaboration.

      @b{if} @b{not} E040 @b{then}
         Interfaces.C_Streams'Elab_Spec;
      @b{end} @b{if};
      E040 := True;
      @b{if} @b{not} E008 @b{then}
         Ada.Exceptions'Elab_Spec;
      @b{end} @b{if};
      @b{if} @b{not} E014 @b{then}
         System.Exception_Table'Elab_Body;
         E014 := True;
      @b{end} @b{if};
      @b{if} @b{not} E053 @b{then}
         Ada.Io_Exceptions'Elab_Spec;
         E053 := True;
      @b{end} @b{if};
      @b{if} @b{not} E017 @b{then}
         System.Exceptions'Elab_Spec;
         E017 := True;
      @b{end} @b{if};
      @b{if} @b{not} E030 @b{then}
         System.Stack_Checking'Elab_Spec;
      @b{end} @b{if};
      @b{if} @b{not} E028 @b{then}
         System.Soft_Links'Elab_Body;
         E028 := True;
      @b{end} @b{if};
      E030 := True;
      @b{if} @b{not} E024 @b{then}
         System.Secondary_Stack'Elab_Body;
         E024 := True;
      @b{end} @b{if};
      @b{if} @b{not} E035 @b{then}
         Ada.Tags'Elab_Spec;
      @b{end} @b{if};
      @b{if} @b{not} E035 @b{then}
         Ada.Tags'Elab_Body;
         E035 := True;
      @b{end} @b{if};
      @b{if} @b{not} E033 @b{then}
         Ada.Streams'Elab_Spec;
         E033 := True;
      @b{end} @b{if};
      @b{if} @b{not} E046 @b{then}
         System.Finalization_Root'Elab_Spec;
      @b{end} @b{if};
      E046 := True;
      @b{if} @b{not} E008 @b{then}
         Ada.Exceptions'Elab_Body;
         E008 := True;
      @b{end} @b{if};
      @b{if} @b{not} E048 @b{then}
         System.Finalization_Implementation'Elab_Spec;
      @b{end} @b{if};
      @b{if} @b{not} E048 @b{then}
         System.Finalization_Implementation'Elab_Body;
         E048 := True;
      @b{end} @b{if};
      @b{if} @b{not} E044 @b{then}
         Ada.Finalization'Elab_Spec;
      @b{end} @b{if};
      E044 := True;
      @b{if} @b{not} E057 @b{then}
         Ada.Finalization.List_Controller'Elab_Spec;
      @b{end} @b{if};
      E057 := True;
      @b{if} @b{not} E055 @b{then}
         System.File_Control_Block'Elab_Spec;
         E055 := True;
      @b{end} @b{if};
      @b{if} @b{not} E042 @b{then}
         System.File_Io'Elab_Body;
         E042 := True;
      @b{end} @b{if};
      @b{if} @b{not} E006 @b{then}
         Ada.Text_Io'Elab_Spec;
      @b{end} @b{if};
      @b{if} @b{not} E006 @b{then}
         Ada.Text_Io'Elab_Body;
         E006 := True;
      @b{end} @b{if};

      Elab_Final_Code := 0;
   @b{end} adainit;

   --------------
   -- adafinal --
   --------------

@findex adafinal
   @b{procedure} adafinal @b{is}
   @b{begin}
      Do_Finalize;
   @b{end} adafinal;

   ----------
   -- main --
   ----------

   --  main is actually a function, as in the ANSI C standard,
   --  defined to return the exit status. The three parameters
   --  are the argument count, argument values and environment
   --  pointer.

@findex Main Program
   @b{function} main
     (argc : Integer;
      argv : System.Address;
      envp : System.Address)
      @b{return} Integer
   @b{is}
      --  The initialize routine performs low level system
      --  initialization using a standard library routine which
      --  sets up signal handling and performs any other
      --  required setup. The routine can be found in file
      --  a-init.c.

@findex __gnat_initialize
      @b{procedure} initialize;
      @b{pragma} Import (C, initialize, "__gnat_initialize");

      --  The finalize routine performs low level system
      --  finalization using a standard library routine. The
      --  routine is found in file a-final.c and in the standard
      --  distribution is a dummy routine that does nothing, so
      --  really this is a hook for special user finalization.

@findex __gnat_finalize
      @b{procedure} finalize;
      @b{pragma} Import (C, finalize, "__gnat_finalize");

      --  We get to the main program of the partition by using
      --  pragma Import because if we try to with the unit and
      --  call it Ada style, then not only do we waste time
      --  recompiling it, but also, we don't really know the right
      --  switches (e.g.@: identifier character set) to be used
      --  to compile it.

      @b{procedure} Ada_Main_Program;
      @b{pragma} Import (Ada, Ada_Main_Program, "_ada_hello");

   --  Start of processing for main

   @b{begin}
      --  Save global variables

      gnat_argc := argc;
      gnat_argv := argv;
      gnat_envp := envp;

      --  Call low level system initialization

      Initialize;

      --  Call our generated Ada initialization routine

      adainit;

      --  This is the point at which we want the debugger to get
      --  control

      Break_Start;

      --  Now we call the main program of the partition

      Ada_Main_Program;

      --  Perform Ada finalization

      adafinal;

      --  Perform low level system finalization

      Finalize;

      --  Return the proper exit status
      @b{return} (gnat_exit_status);
   @b{end};

--  This section is entirely comments, so it has no effect on the
--  compilation of the Ada_Main package. It provides the list of
--  object files and linker options, as well as some standard
--  libraries needed for the link. The gnatlink utility parses
--  this b~hello.adb file to read these comment lines to generate
--  the appropriate command line arguments for the call to the
--  system linker. The BEGIN/END lines are used for sentinels for
--  this parsing operation.

--  The exact file names will of course depend on the environment,
--  host/target and location of files on the host system.

@findex Object file list
-- BEGIN Object file/option list
   --   ./hello.o
   --   -L./
   --   -L/usr/local/gnat/lib/gcc-lib/i686-pc-linux-gnu/2.8.1/adalib/
   --   /usr/local/gnat/lib/gcc-lib/i686-pc-linux-gnu/2.8.1/adalib/libgnat.a
-- END Object file/option list

@b{end} ada_main;
@end smallexample

@noindent
The Ada code in the above example is exactly what is generated by the
binder. We have added comments to more clearly indicate the function
of each part of the generated @code{Ada_Main} package.

The code is standard Ada in all respects, and can be processed by any
tools that handle Ada. In particular, it is possible to use the debugger
in Ada mode to debug the generated @code{Ada_Main} package. For example,
suppose that for reasons that you do not understand, your program is crashing
during elaboration of the body of @code{Ada.Text_IO}. To locate this bug,
you can place a breakpoint on the call:

@smallexample @c ada
Ada.Text_Io'Elab_Body;
@end smallexample

@noindent
and trace the elaboration routine for this package to find out where
the problem might be (more usually of course you would be debugging
elaboration code in your own application).

@node Elaboration Order Handling in GNAT
@appendix Elaboration Order Handling in GNAT
@cindex Order of elaboration
@cindex Elaboration control

@menu
* Elaboration Code::
* Checking the Elaboration Order::
* Controlling the Elaboration Order::
* Controlling Elaboration in GNAT - Internal Calls::
* Controlling Elaboration in GNAT - External Calls::
* Default Behavior in GNAT - Ensuring Safety::
* Treatment of Pragma Elaborate::
* Elaboration Issues for Library Tasks::
* Mixing Elaboration Models::
* What to Do If the Default Elaboration Behavior Fails::
* Elaboration for Indirect Calls::
* Summary of Procedures for Elaboration Control::
* Other Elaboration Order Considerations::
* Determining the Chosen Elaboration Order::
@end menu

@noindent
This chapter describes the handling of elaboration code in Ada and
in GNAT, and discusses how the order of elaboration of program units can
be controlled in GNAT, either automatically or with explicit programming
features.

@node Elaboration Code
@section Elaboration Code

@noindent
Ada provides rather general mechanisms for executing code at elaboration
time, that is to say before the main program starts executing. Such code arises
in three contexts:

@table @asis
@item Initializers for variables.
Variables declared at the library level, in package specs or bodies, can
require initialization that is performed at elaboration time, as in:
@smallexample @c ada
@cartouche
Sqrt_Half : Float := Sqrt (0.5);
@end cartouche
@end smallexample

@item Package initialization code
Code in a @code{BEGIN-END} section at the outer level of a package body is
executed as part of the package body elaboration code.

@item Library level task allocators
Tasks that are declared using task allocators at the library level
start executing immediately and hence can execute at elaboration time.
@end table

@noindent
Subprogram calls are possible in any of these contexts, which means that
any arbitrary part of the program may be executed as part of the elaboration
code. It is even possible to write a program which does all its work at
elaboration time, with a null main program, although stylistically this
would usually be considered an inappropriate way to structure
a program.

An important concern arises in the context of elaboration code:
we have to be sure that it is executed in an appropriate order. What we
have is a series of elaboration code sections, potentially one section
for each unit in the program. It is important that these execute
in the correct order. Correctness here means that, taking the above
example of the declaration of @code{Sqrt_Half},
if some other piece of
elaboration code references @code{Sqrt_Half},
then it must run after the
section of elaboration code that contains the declaration of
@code{Sqrt_Half}.

There would never be any order of elaboration problem if we made a rule
that whenever you @code{with} a unit, you must elaborate both the spec and body
of that unit before elaborating the unit doing the @code{with}'ing:

@smallexample @c ada
@group
@cartouche
@b{with} Unit_1;
@b{package} Unit_2 @b{is} @dots{}
@end cartouche
@end group
@end smallexample

@noindent
would require that both the body and spec of @code{Unit_1} be elaborated
before the spec of @code{Unit_2}. However, a rule like that would be far too
restrictive. In particular, it would make it impossible to have routines
in separate packages that were mutually recursive.

You might think that a clever enough compiler could look at the actual
elaboration code and determine an appropriate correct order of elaboration,
but in the general case, this is not possible. Consider the following
example.

In the body of @code{Unit_1}, we have a procedure @code{Func_1}
that references
the variable @code{Sqrt_1}, which is declared in the elaboration code
of the body of @code{Unit_1}:

@smallexample @c ada
@cartouche
Sqrt_1 : Float := Sqrt (0.1);
@end cartouche
@end smallexample

@noindent
The elaboration code of the body of @code{Unit_1} also contains:

@smallexample @c ada
@group
@cartouche
@b{if} expression_1 = 1 @b{then}
   Q := Unit_2.Func_2;
@b{end} @b{if};
@end cartouche
@end group
@end smallexample

@noindent
@code{Unit_2} is exactly parallel,
it has a procedure @code{Func_2} that references
the variable @code{Sqrt_2}, which is declared in the elaboration code of
the body @code{Unit_2}:

@smallexample @c ada
@cartouche
Sqrt_2 : Float := Sqrt (0.1);
@end cartouche
@end smallexample

@noindent
The elaboration code of the body of @code{Unit_2} also contains:

@smallexample @c ada
@group
@cartouche
@b{if} expression_2 = 2 @b{then}
   Q := Unit_1.Func_1;
@b{end} @b{if};
@end cartouche
@end group
@end smallexample

@noindent
Now the question is, which of the following orders of elaboration is
acceptable:

@smallexample
@group
Spec of Unit_1
Spec of Unit_2
Body of Unit_1
Body of Unit_2
@end group
@end smallexample

@noindent
or

@smallexample
@group
Spec of Unit_2
Spec of Unit_1
Body of Unit_2
Body of Unit_1
@end group
@end smallexample

@noindent
If you carefully analyze the flow here, you will see that you cannot tell
at compile time the answer to this question.
If @code{expression_1} is not equal to 1,
and @code{expression_2} is not equal to 2,
then either order is acceptable, because neither of the function calls is
executed. If both tests evaluate to true, then neither order is acceptable
and in fact there is no correct order.

If one of the two expressions is true, and the other is false, then one
of the above orders is correct, and the other is incorrect. For example,
if @code{expression_1} /= 1 and @code{expression_2} = 2,
then the call to @code{Func_1}
will occur, but not the call to @code{Func_2.}
This means that it is essential
to elaborate the body of @code{Unit_1} before
the body of @code{Unit_2}, so the first
order of elaboration is correct and the second is wrong.

By making @code{expression_1} and @code{expression_2}
depend on input data, or perhaps
the time of day, we can make it impossible for the compiler or binder
to figure out which of these expressions will be true, and hence it
is impossible to guarantee a safe order of elaboration at run time.

@node Checking the Elaboration Order
@section Checking the Elaboration Order

@noindent
In some languages that involve the same kind of elaboration problems,
e.g.@: Java and C++, the programmer is expected to worry about these
ordering problems himself, and it is common to
write a program in which an incorrect elaboration order  gives
surprising results, because it references variables before they
are initialized.
Ada is designed to be a safe language, and a programmer-beware approach is
clearly not sufficient. Consequently, the language provides three lines
of defense:

@table @asis
@item Standard rules
Some standard rules restrict the possible choice of elaboration
order. In particular, if you @code{with} a unit, then its spec is always
elaborated before the unit doing the @code{with}. Similarly, a parent
spec is always elaborated before the child spec, and finally
a spec is always elaborated before its corresponding body.

@item Dynamic elaboration checks
@cindex Elaboration checks
@cindex Checks, elaboration
Dynamic checks are made at run time, so that if some entity is accessed
before it is elaborated (typically  by means of a subprogram call)
then the exception (@code{Program_Error}) is raised.

@item Elaboration control
Facilities are provided for the programmer to specify the desired order
of elaboration.
@end table

Let's look at these facilities in more detail. First, the rules for
dynamic checking. One possible rule would be simply to say that the
exception is raised if you access a variable which has not yet been
elaborated. The trouble with this approach is that it could require
expensive checks on every variable reference. Instead Ada has two
rules which are a little more restrictive, but easier to check, and
easier to state:

@table @asis
@item Restrictions on calls
A subprogram can only be called at elaboration time if its body
has been elaborated. The rules for elaboration given above guarantee
that the spec of the subprogram has been elaborated before the
call, but not the body. If this rule is violated, then the
exception @code{Program_Error} is raised.

@item Restrictions on instantiations
A generic unit can only be instantiated if the body of the generic
unit has been elaborated. Again, the rules for elaboration given above
guarantee that the spec of the generic unit has been elaborated
before the instantiation, but not the body. If this rule is
violated, then the exception @code{Program_Error} is raised.
@end table

@noindent
The idea is that if the body has been elaborated, then any variables
it references must have been elaborated; by checking for the body being
elaborated we guarantee that none of its references causes any
trouble. As we noted above, this is a little too restrictive, because a
subprogram that has no non-local references in its body may in fact be safe
to call. However, it really would be unsafe to rely on this, because
it would mean that the caller was aware of details of the implementation
in the body. This goes against the basic tenets of Ada.

A plausible implementation can be described as follows.
A Boolean variable is associated with each subprogram
and each generic unit. This variable is initialized to False, and is set to
True at the point body is elaborated. Every call or instantiation checks the
variable, and raises @code{Program_Error} if the variable is False.

Note that one might think that it would be good enough to have one Boolean
variable for each package, but that would not deal with cases of trying
to call a body in the same package as the call
that has not been elaborated yet.
Of course a compiler may be able to do enough analysis to optimize away
some of the Boolean variables as unnecessary, and @code{GNAT} indeed
does such optimizations, but still the easiest conceptual model is to
think of there being one variable per subprogram.

@node Controlling the Elaboration Order
@section Controlling the Elaboration Order

@noindent
In the previous section we discussed the rules in Ada which ensure
that @code{Program_Error} is raised if an incorrect elaboration order is
chosen. This prevents erroneous executions, but we need mechanisms to
specify a correct execution and avoid the exception altogether.
To achieve this, Ada provides a number of features for controlling
the order of elaboration. We discuss these features in this section.

First, there are several ways of indicating to the compiler that a given
unit has no elaboration problems:

@table @asis
@item packages that do not require a body
A library package that does not require a body does not permit
a body (this rule was introduced in Ada 95).
Thus if we have a such a package, as in:

@smallexample @c ada
@group
@cartouche
@b{package} Definitions @b{is}
   @b{generic}
      @b{type} m @b{is} @b{new} integer;
   @b{package} Subp @b{is}
      @b{type} a @b{is} @b{array} (1 .. 10) @b{of} m;
      @b{type} b @b{is} @b{array} (1 .. 20) @b{of} m;
   @b{end} Subp;
@b{end} Definitions;
@end cartouche
@end group
@end smallexample

@noindent
A package that @code{with}'s @code{Definitions} may safely instantiate
@code{Definitions.Subp} because the compiler can determine that there
definitely is no package body to worry about in this case

@item pragma Pure
@cindex pragma Pure
@findex Pure
Places sufficient restrictions on a unit to guarantee that
no call to any subprogram in the unit can result in an
elaboration problem. This means that the compiler does not need
to worry about the point of elaboration of such units, and in
particular, does not need to check any calls to any subprograms
in this unit.

@item pragma Preelaborate
@findex Preelaborate
@cindex pragma Preelaborate
This pragma places slightly less stringent restrictions on a unit than
does pragma Pure,
but these restrictions are still sufficient to ensure that there
are no elaboration problems with any calls to the unit.

@item pragma Elaborate_Body
@findex Elaborate_Body
@cindex pragma Elaborate_Body
This pragma requires that the body of a unit be elaborated immediately
after its spec. Suppose a unit @code{A} has such a pragma,
and unit @code{B} does
a @code{with} of unit @code{A}. Recall that the standard rules require
the spec of unit @code{A}
to be elaborated before the @code{with}'ing unit; given the pragma in
@code{A}, we also know that the body of @code{A}
will be elaborated before @code{B}, so
that calls to @code{A} are safe and do not need a check.
@end table

@noindent
Note that,
unlike pragma @code{Pure} and pragma @code{Preelaborate},
the use of
@code{Elaborate_Body} does not guarantee that the program is
free of elaboration problems, because it may not be possible
to satisfy the requested elaboration order.
Let's go back to the example with @code{Unit_1} and @code{Unit_2}.
If a programmer
marks @code{Unit_1} as @code{Elaborate_Body},
and not @code{Unit_2,} then the order of
elaboration will be:

@smallexample
@group
Spec of Unit_2
Spec of Unit_1
Body of Unit_1
Body of Unit_2
@end group
@end smallexample

@noindent
Now that means that the call to @code{Func_1} in @code{Unit_2}
need not be checked,
it must be safe. But the call to @code{Func_2} in
@code{Unit_1} may still fail if
@code{Expression_1} is equal to 1,
and the programmer must still take
responsibility for this not being the case.

If all units carry a pragma @code{Elaborate_Body}, then all problems are
eliminated, except for calls entirely within a body, which are
in any case fully under programmer control. However, using the pragma
everywhere is not always possible.
In particular, for our @code{Unit_1}/@code{Unit_2} example, if
we marked both of them as having pragma @code{Elaborate_Body}, then
clearly there would be no possible elaboration order.

The above pragmas allow a server to guarantee safe use by clients, and
clearly this is the preferable approach. Consequently a good rule
is to mark units as @code{Pure} or @code{Preelaborate} if possible,
and if this is not possible,
mark them as @code{Elaborate_Body} if possible.
As we have seen, there are situations where neither of these
three pragmas can be used.
So we also provide methods for clients to control the
order of elaboration of the servers on which they depend:

@table @asis
@item pragma Elaborate (unit)
@findex Elaborate
@cindex pragma Elaborate
This pragma is placed in the context clause, after a @code{with} clause,
and it requires that the body of the named unit be elaborated before
the unit in which the pragma occurs. The idea is to use this pragma
if the current unit calls at elaboration time, directly or indirectly,
some subprogram in the named unit.

@item pragma Elaborate_All (unit)
@findex Elaborate_All
@cindex pragma Elaborate_All
This is a stronger version of the Elaborate pragma. Consider the
following example:

@smallexample
Unit A @code{with}'s unit B and calls B.Func in elab code
Unit B @code{with}'s unit C, and B.Func calls C.Func
@end smallexample

@noindent
Now if we put a pragma @code{Elaborate (B)}
in unit @code{A}, this ensures that the
body of @code{B} is elaborated before the call, but not the
body of @code{C}, so
the call to @code{C.Func} could still cause @code{Program_Error} to
be raised.

The effect of a pragma @code{Elaborate_All} is stronger, it requires
not only that the body of the named unit be elaborated before the
unit doing the @code{with}, but also the bodies of all units that the
named unit uses, following @code{with} links transitively. For example,
if we put a pragma @code{Elaborate_All (B)} in unit @code{A},
then it requires
not only that the body of @code{B} be elaborated before @code{A},
but also the
body of @code{C}, because @code{B} @code{with}'s @code{C}.
@end table

@noindent
We are now in a position to give a usage rule in Ada for avoiding
elaboration problems, at least if dynamic dispatching and access to
subprogram values are not used. We will handle these cases separately
later.

The rule is simple. If a unit has elaboration code that can directly or
indirectly make a call to a subprogram in a @code{with}'ed unit, or instantiate
a generic package in a @code{with}'ed unit,
then if the @code{with}'ed unit does not have
pragma @code{Pure} or @code{Preelaborate}, then the client should have
a pragma @code{Elaborate_All}
for the @code{with}'ed unit. By following this rule a client is
assured that calls can be made without risk of an exception.

For generic subprogram instantiations, the rule can be relaxed to
require only a pragma @code{Elaborate} since elaborating the body
of a subprogram cannot cause any transitive elaboration (we are
not calling the subprogram in this case, just elaborating its
declaration).

If this rule is not followed, then a program may be in one of four
states:

@table @asis
@item No order exists
No order of elaboration exists which follows the rules, taking into
account any @code{Elaborate}, @code{Elaborate_All},
or @code{Elaborate_Body} pragmas. In
this case, an Ada compiler must diagnose the situation at bind
time, and refuse to build an executable program.

@item One or more orders exist, all incorrect
One or more acceptable elaboration orders exist, and all of them
generate an elaboration order problem. In this case, the binder
can build an executable program, but @code{Program_Error} will be raised
when the program is run.

@item Several orders exist, some right, some incorrect
One or more acceptable elaboration orders exists, and some of them
work, and some do not. The programmer has not controlled
the order of elaboration, so the binder may or may not pick one of
the correct orders, and the program may or may not raise an
exception when it is run. This is the worst case, because it means
that the program may fail when moved to another compiler, or even
another version of the same compiler.

@item One or more orders exists, all correct
One ore more acceptable elaboration orders exist, and all of them
work. In this case the program runs successfully. This state of
affairs can be guaranteed by following the rule we gave above, but
may be true even if the rule is not followed.
@end table

@noindent
Note that one additional advantage of following our rules on the use
of @code{Elaborate} and @code{Elaborate_All}
is that the program continues to stay in the ideal (all orders OK) state
even if maintenance
changes some bodies of some units. Conversely, if a program that does
not follow this rule happens to be safe at some point, this state of affairs
may deteriorate silently as a result of maintenance changes.

You may have noticed that the above discussion did not mention
the use of @code{Elaborate_Body}. This was a deliberate omission. If you
@code{with} an @code{Elaborate_Body} unit, it still may be the case that
code in the body makes calls to some other unit, so it is still necessary
to use @code{Elaborate_All} on such units.

@node Controlling Elaboration in GNAT - Internal Calls
@section Controlling Elaboration in GNAT - Internal Calls

@noindent
In the case of internal calls, i.e., calls within a single package, the
programmer has full control over the order of elaboration, and it is up
to the programmer to elaborate declarations in an appropriate order. For
example writing:

@smallexample @c ada
@group
@cartouche
@b{function} One @b{return} Float;

Q : Float := One;

@b{function} One @b{return} Float @b{is}
@b{begin}
     @b{return} 1.0;
@b{end} One;
@end cartouche
@end group
@end smallexample

@noindent
will obviously raise @code{Program_Error} at run time, because function
One will be called before its body is elaborated. In this case GNAT will
generate a warning that the call will raise @code{Program_Error}:

@smallexample
@group
@cartouche
 1. procedure y is
 2.    function One return Float;
 3.
 4.    Q : Float := One;
                    |
    >>> warning: cannot call "One" before body is elaborated
    >>> warning: Program_Error will be raised at run time

 5.
 6.    function One return Float is
 7.    begin
 8.         return 1.0;
 9.    end One;
10.
11. begin
12.    null;
13. end;
@end cartouche
@end group
@end smallexample

@noindent
Note that in this particular case, it is likely that the call is safe, because
the function @code{One} does not access any global variables.
Nevertheless in Ada, we do not want the validity of the check to depend on
the contents of the body (think about the separate compilation case), so this
is still wrong, as we discussed in the previous sections.

The error is easily corrected by rearranging the declarations so that the
body of @code{One} appears before the declaration containing the call
(note that in Ada 95 and Ada 2005,
declarations can appear in any order, so there is no restriction that
would prevent this reordering, and if we write:

@smallexample @c ada
@group
@cartouche
@b{function} One @b{return} Float;

@b{function} One @b{return} Float @b{is}
@b{begin}
     @b{return} 1.0;
@b{end} One;

Q : Float := One;
@end cartouche
@end group
@end smallexample

@noindent
then all is well, no warning is generated, and no
@code{Program_Error} exception
will be raised.
Things are more complicated when a chain of subprograms is executed:

@smallexample @c ada
@group
@cartouche
@b{function} A @b{return} Integer;
@b{function} B @b{return} Integer;
@b{function} C @b{return} Integer;

@b{function} B @b{return} Integer @b{is} @b{begin} @b{return} A; @b{end};
@b{function} C @b{return} Integer @b{is} @b{begin} @b{return} B; @b{end};

X : Integer := C;

@b{function} A @b{return} Integer @b{is} @b{begin} @b{return} 1; @b{end};
@end cartouche
@end group
@end smallexample

@noindent
Now the call to @code{C}
at elaboration time in the declaration of @code{X} is correct, because
the body of @code{C} is already elaborated,
and the call to @code{B} within the body of
@code{C} is correct, but the call
to @code{A} within the body of @code{B} is incorrect, because the body
of @code{A} has not been elaborated, so @code{Program_Error}
will be raised on the call to @code{A}.
In this case GNAT will generate a
warning that @code{Program_Error} may be
raised at the point of the call. Let's look at the warning:

@smallexample
@group
@cartouche
 1. procedure x is
 2.    function A return Integer;
 3.    function B return Integer;
 4.    function C return Integer;
 5.
 6.    function B return Integer is begin return A; end;
                                                    |
    >>> warning: call to "A" before body is elaborated may
                 raise Program_Error
    >>> warning: "B" called at line 7
    >>> warning: "C" called at line 9

 7.    function C return Integer is begin return B; end;
 8.
 9.    X : Integer := C;
10.
11.    function A return Integer is begin return 1; end;
12.
13. begin
14.    null;
15. end;
@end cartouche
@end group
@end smallexample

@noindent
Note that the message here says ``may raise'', instead of the direct case,
where the message says ``will be raised''. That's because whether
@code{A} is
actually called depends in general on run-time flow of control.
For example, if the body of @code{B} said

@smallexample @c ada
@group
@cartouche
@b{function} B @b{return} Integer @b{is}
@b{begin}
   @b{if} some-condition-depending-on-input-data @b{then}
      @b{return} A;
   @b{else}
      @b{return} 1;
   @b{end} @b{if};
@b{end} B;
@end cartouche
@end group
@end smallexample

@noindent
then we could not know until run time whether the incorrect call to A would
actually occur, so @code{Program_Error} might
or might not be raised. It is possible for a compiler to
do a better job of analyzing bodies, to
determine whether or not @code{Program_Error}
might be raised, but it certainly
couldn't do a perfect job (that would require solving the halting problem
and is provably impossible), and because this is a warning anyway, it does
not seem worth the effort to do the analysis. Cases in which it
would be relevant are rare.

In practice, warnings of either of the forms given
above will usually correspond to
real errors, and should be examined carefully and eliminated.
In the rare case where a warning is bogus, it can be suppressed by any of
the following methods:

@itemize @bullet
@item
Compile with the @option{-gnatws} switch set

@item
Suppress @code{Elaboration_Check} for the called subprogram

@item
Use pragma @code{Warnings_Off} to turn warnings off for the call
@end itemize

@noindent
For the internal elaboration check case,
GNAT by default generates the
necessary run-time checks to ensure
that @code{Program_Error} is raised if any
call fails an elaboration check. Of course this can only happen if a
warning has been issued as described above. The use of pragma
@code{Suppress (Elaboration_Check)} may (but is not guaranteed to) suppress
some of these checks, meaning that it may be possible (but is not
guaranteed) for a program to be able to call a subprogram whose body
is not yet elaborated, without raising a @code{Program_Error} exception.

@node Controlling Elaboration in GNAT - External Calls
@section Controlling Elaboration in GNAT - External Calls

@noindent
The previous section discussed the case in which the execution of a
particular thread of elaboration code occurred entirely within a
single unit. This is the easy case to handle, because a programmer
has direct and total control over the order of elaboration, and
furthermore, checks need only be generated in cases which are rare
and which the compiler can easily detect.
The situation is more complex when separate compilation is taken into account.
Consider the following:

@smallexample @c ada
@cartouche
@group
@b{package} Math @b{is}
   @b{function} Sqrt (Arg : Float) @b{return} Float;
@b{end} Math;

@b{package} @b{body} Math @b{is}
   @b{function} Sqrt (Arg : Float) @b{return} Float @b{is}
   @b{begin}
         @dots{}
   @b{end} Sqrt;
@b{end} Math;
@end group
@group
@b{with} Math;
@b{package} Stuff @b{is}
   X : Float := Math.Sqrt (0.5);
@b{end} Stuff;

@b{with} Stuff;
@b{procedure} Main @b{is}
@b{begin}
   @dots{}
@b{end} Main;
@end group
@end cartouche
@end smallexample

@noindent
where @code{Main} is the main program. When this program is executed, the
elaboration code must first be executed, and one of the jobs of the
binder is to determine the order in which the units of a program are
to be elaborated. In this case we have four units: the spec and body
of @code{Math},
the spec of @code{Stuff} and the body of @code{Main}).
In what order should the four separate sections of elaboration code
be executed?

There are some restrictions in the order of elaboration that the binder
can choose. In particular, if unit U has a @code{with}
for a package @code{X}, then you
are assured that the spec of @code{X}
is elaborated before U , but you are
not assured that the body of @code{X}
is elaborated before U.
This means that in the above case, the binder is allowed to choose the
order:

@smallexample
spec of Math
spec of Stuff
body of Math
body of Main
@end smallexample

@noindent
but that's not good, because now the call to @code{Math.Sqrt}
that happens during
the elaboration of the @code{Stuff}
spec happens before the body of @code{Math.Sqrt} is
elaborated, and hence causes @code{Program_Error} exception to be raised.
At first glance, one might say that the binder is misbehaving, because
obviously you want to elaborate the body of something you @code{with}
first, but
that is not a general rule that can be followed in all cases. Consider

@smallexample @c ada
@group
@cartouche
@b{package} X @b{is} @dots{}

@b{package} Y @b{is} @dots{}

@b{with} X;
@b{package} @b{body} Y @b{is} @dots{}

@b{with} Y;
@b{package} @b{body} X @b{is} @dots{}
@end cartouche
@end group
@end smallexample

@noindent
This is a common arrangement, and, apart from the order of elaboration
problems that might arise in connection with elaboration code, this works fine.
A rule that says that you must first elaborate the body of anything you
@code{with} cannot work in this case:
the body of @code{X} @code{with}'s @code{Y},
which means you would have to
elaborate the body of @code{Y} first, but that @code{with}'s @code{X},
which means
you have to elaborate the body of @code{X} first, but @dots{} and we have a
loop that cannot be broken.

It is true that the binder can in many cases guess an order of elaboration
that is unlikely to cause a @code{Program_Error}
exception to be raised, and it tries to do so (in the
above example of @code{Math/Stuff/Spec}, the GNAT binder will
by default
elaborate the body of @code{Math} right after its spec, so all will be well).

However, a program that blindly relies on the binder to be helpful can
get into trouble, as we discussed in the previous sections, so
GNAT
provides a number of facilities for assisting the programmer in
developing programs that are robust with respect to elaboration order.

@node Default Behavior in GNAT - Ensuring Safety
@section Default Behavior in GNAT - Ensuring Safety

@noindent
The default behavior in GNAT ensures elaboration safety. In its
default mode GNAT implements the
rule we previously described as the right approach. Let's restate it:

@itemize
@item
@emph{If a unit has elaboration code that can directly or indirectly make a
call to a subprogram in a @code{with}'ed unit, or instantiate a generic
package in a @code{with}'ed unit, then if the @code{with}'ed unit
does not have pragma @code{Pure} or
@code{Preelaborate}, then the client should have an
@code{Elaborate_All} pragma for the @code{with}'ed unit.}

@emph{In the case of instantiating a generic subprogram, it is always
sufficient to have only an @code{Elaborate} pragma for the
@code{with}'ed unit.}
@end itemize

@noindent
By following this rule a client is assured that calls and instantiations
can be made without risk of an exception.

In this mode GNAT traces all calls that are potentially made from
elaboration code, and puts in any missing implicit @code{Elaborate}
and @code{Elaborate_All} pragmas.
The advantage of this approach is that no elaboration problems
are possible if the binder can find an elaboration order that is
consistent with these implicit @code{Elaborate} and
@code{Elaborate_All} pragmas. The
disadvantage of this approach is that no such order may exist.

If the binder does not generate any diagnostics, then it means that it has
found an elaboration order that is guaranteed to be safe. However, the binder
may still be relying on implicitly generated @code{Elaborate} and
@code{Elaborate_All} pragmas so portability to other compilers than GNAT is not
guaranteed.

If it is important to guarantee portability, then the compilations should
use the
@option{-gnatel}
(info messages for elaboration prag mas) switch. This will cause info messages
to be generated indicating the missing @code{Elaborate} and
@code{Elaborate_All} pragmas.
Consider the following source program:

@smallexample @c ada
@group
@cartouche
@b{with} k;
@b{package} j @b{is}
  m : integer := k.r;
@b{end};
@end cartouche
@end group
@end smallexample

@noindent
where it is clear that there
should be a pragma @code{Elaborate_All}
for unit @code{k}. An implicit pragma will be generated, and it is
likely that the binder will be able to honor it. However, if you want
to port this program to some other Ada compiler than GNAT.
it is safer to include the pragma explicitly in the source. If this
unit is compiled with the
@option{-gnatel}
switch, then the compiler outputs an information message:

@smallexample
@group
@cartouche
1. with k;
2. package j is
3.   m : integer := k.r;
                     |
   >>> info: call to "r" may raise Program_Error
   >>> info: missing pragma Elaborate_All for "k"

4. end;
@end cartouche
@end group
@end smallexample

@noindent
and these messages can be used as a guide for supplying manually
the missing pragmas. It is usually a bad idea to use this
option during development. That's because it will tell you when
you need to put in a pragma, but cannot tell you when it is time
to take it out. So the use of pragma @code{Elaborate_All} may lead to
unnecessary dependencies and even false circularities.

This default mode is more restrictive than the Ada Reference
Manual, and it is possible to construct programs which will compile
using the dynamic model described there, but will run into a
circularity using the safer static model we have described.

Of course any Ada compiler must be able to operate in a mode
consistent with the requirements of the Ada Reference Manual,
and in particular must have the capability of implementing the
standard dynamic model of elaboration with run-time checks.

In GNAT, this standard mode can be achieved either by the use of
the @option{-gnatE} switch on the compiler (@command{gcc} or
@command{gnatmake}) command, or by the use of the configuration pragma:

@smallexample @c ada
@b{pragma} Elaboration_Checks (DYNAMIC);
@end smallexample

@noindent
Either approach will cause the unit affected to be compiled using the
standard dynamic run-time elaboration checks described in the Ada
Reference Manual. The static model is generally preferable, since it
is clearly safer to rely on compile and link time checks rather than
run-time checks. However, in the case of legacy code, it may be
difficult to meet the requirements of the static model. This
issue is further discussed in
@ref{What to Do If the Default Elaboration Behavior Fails}.

Note that the static model provides a strict subset of the allowed
behavior and programs of the Ada Reference Manual, so if you do
adhere to the static model and no circularities exist,
then you are assured that your program will
work using the dynamic model, providing that you remove any
pragma Elaborate statements from the source.

@node Treatment of Pragma Elaborate
@section Treatment of Pragma Elaborate
@cindex Pragma Elaborate

@noindent
The use of @code{pragma Elaborate}
should generally be avoided in Ada 95 and Ada 2005 programs,
since there is no guarantee that transitive calls
will be properly handled. Indeed at one point, this pragma was placed
in Annex J (Obsolescent Features), on the grounds that it is never useful.

Now that's a bit restrictive. In practice, the case in which
@code{pragma Elaborate} is useful is when the caller knows that there
are no transitive calls, or that the called unit contains all necessary
transitive @code{pragma Elaborate} statements, and legacy code often
contains such uses.

Strictly speaking the static mode in GNAT should ignore such pragmas,
since there is no assurance at compile time that the necessary safety
conditions are met. In practice, this would cause GNAT to be incompatible
with correctly written Ada 83 code that had all necessary
@code{pragma Elaborate} statements in place. Consequently, we made the
decision that GNAT in its default mode will believe that if it encounters
a @code{pragma Elaborate} then the programmer knows what they are doing,
and it will trust that no elaboration errors can occur.

The result of this decision is two-fold. First to be safe using the
static mode, you should remove all @code{pragma Elaborate} statements.
Second, when fixing circularities in existing code, you can selectively
use @code{pragma Elaborate} statements to convince the static mode of
GNAT that it need not generate an implicit @code{pragma Elaborate_All}
statement.

When using the static mode with @option{-gnatwl}, any use of
@code{pragma Elaborate} will generate a warning about possible
problems.

@node Elaboration Issues for Library Tasks
@section Elaboration Issues for Library Tasks
@cindex Library tasks, elaboration issues
@cindex Elaboration of library tasks

@noindent
In this section we examine special elaboration issues that arise for
programs that declare library level tasks.

Generally the model of execution of an Ada program is that all units are
elaborated, and then execution of the program starts. However, the
declaration of library tasks definitely does not fit this model. The
reason for this is that library tasks start as soon as they are declared
(more precisely, as soon as the statement part of the enclosing package
body is reached), that is to say before elaboration
of the program is complete. This means that if such a task calls a
subprogram, or an entry in another task, the callee may or may not be
elaborated yet, and in the standard
Reference Manual model of dynamic elaboration checks, you can even
get timing dependent Program_Error exceptions, since there can be
a race between the elaboration code and the task code.

The static model of elaboration in GNAT seeks to avoid all such
dynamic behavior, by being conservative, and the conservative
approach in this particular case is to assume that all the code
in a task body is potentially executed at elaboration time if
a task is declared at the library level.

This can definitely result in unexpected circularities. Consider
the following example

@smallexample @c ada
@b{package} Decls @b{is}
  @b{task} Lib_Task @b{is}
     @b{entry} Start;
  @b{end} Lib_Task;

  @b{type} My_Int @b{is} @b{new} Integer;

  @b{function} Ident (M : My_Int) @b{return} My_Int;
@b{end} Decls;

@b{with} Utils;
@b{package} @b{body} Decls @b{is}
  @b{task} @b{body} Lib_Task @b{is}
  @b{begin}
     @b{accept} Start;
     Utils.Put_Val (2);
  @b{end} Lib_Task;

  @b{function} Ident (M : My_Int) @b{return} My_Int @b{is}
  @b{begin}
     @b{return} M;
  @b{end} Ident;
@b{end} Decls;

@b{with} Decls;
@b{package} Utils @b{is}
  @b{procedure} Put_Val (Arg : Decls.My_Int);
@b{end} Utils;

@b{with} Text_IO;
@b{package} @b{body} Utils @b{is}
  @b{procedure} Put_Val (Arg : Decls.My_Int) @b{is}
  @b{begin}
     Text_IO.Put_Line (Decls.My_Int'Image (Decls.Ident (Arg)));
  @b{end} Put_Val;
@b{end} Utils;

@b{with} Decls;
@b{procedure} Main @b{is}
@b{begin}
   Decls.Lib_Task.Start;
@b{end};
@end smallexample

@noindent
If the above example is compiled in the default static elaboration
mode, then a circularity occurs. The circularity comes from the call
@code{Utils.Put_Val} in the task body of @code{Decls.Lib_Task}. Since
this call occurs in elaboration code, we need an implicit pragma
@code{Elaborate_All} for @code{Utils}. This means that not only must
the spec and body of @code{Utils} be elaborated before the body
of @code{Decls}, but also the spec and body of any unit that is
@code{with'ed} by the body of @code{Utils} must also be elaborated before
the body of @code{Decls}. This is the transitive implication of
pragma @code{Elaborate_All} and it makes sense, because in general
the body of @code{Put_Val} might have a call to something in a
@code{with'ed} unit.

In this case, the body of Utils (actually its spec) @code{with's}
@code{Decls}. Unfortunately this means that the body of @code{Decls}
must be elaborated before itself, in case there is a call from the
body of @code{Utils}.

Here is the exact chain of events we are worrying about:

@enumerate
@item
In the body of @code{Decls} a call is made from within the body of a library
task to a subprogram in the package @code{Utils}. Since this call may
occur at elaboration time (given that the task is activated at elaboration
time), we have to assume the worst, i.e., that the
call does happen at elaboration time.

@item
This means that the body and spec of @code{Util} must be elaborated before
the body of @code{Decls} so that this call does not cause an access before
elaboration.

@item
Within the body of @code{Util}, specifically within the body of
@code{Util.Put_Val} there may be calls to any unit @code{with}'ed
by this package.

@item
One such @code{with}'ed package is package @code{Decls}, so there
might be a call to a subprogram in @code{Decls} in @code{Put_Val}.
In fact there is such a call in this example, but we would have to
assume that there was such a call even if it were not there, since
we are not supposed to write the body of @code{Decls} knowing what
is in the body of @code{Utils}; certainly in the case of the
static elaboration model, the compiler does not know what is in
other bodies and must assume the worst.

@item
This means that the spec and body of @code{Decls} must also be
elaborated before we elaborate the unit containing the call, but
that unit is @code{Decls}! This means that the body of @code{Decls}
must be elaborated before itself, and that's a circularity.
@end enumerate

@noindent
Indeed, if you add an explicit pragma @code{Elaborate_All} for @code{Utils} in
the body of @code{Decls} you will get a true Ada Reference Manual
circularity that makes the program illegal.

In practice, we have found that problems with the static model of
elaboration in existing code often arise from library tasks, so
we must address this particular situation.

Note that if we compile and run the program above, using the dynamic model of
elaboration (that is to say use the @option{-gnatE} switch),
then it compiles, binds,
links, and runs, printing the expected result of 2. Therefore in some sense
the circularity here is only apparent, and we need to capture
the properties of this program that  distinguish it from other library-level
tasks that have real elaboration problems.

We have four possible answers to this question:

@itemize @bullet

@item
Use the dynamic model of elaboration.

If we use the @option{-gnatE} switch, then as noted above, the program works.
Why is this? If we examine the task body, it is apparent that the task cannot
proceed past the
@code{accept} statement until after elaboration has been completed, because
the corresponding entry call comes from the main program, not earlier.
This is why the dynamic model works here. But that's really giving
up on a precise analysis, and we prefer to take this approach only if we cannot
solve the
problem in any other manner. So let us examine two ways to reorganize
the program to avoid the potential elaboration problem.

@item
Split library tasks into separate packages.

Write separate packages, so that library tasks are isolated from
other declarations as much as possible. Let us look at a variation on
the above program.

@smallexample @c ada
@b{package} Decls1 @b{is}
  @b{task} Lib_Task @b{is}
     @b{entry} Start;
  @b{end} Lib_Task;
@b{end} Decls1;

@b{with} Utils;
@b{package} @b{body} Decls1 @b{is}
  @b{task} @b{body} Lib_Task @b{is}
  @b{begin}
     @b{accept} Start;
     Utils.Put_Val (2);
  @b{end} Lib_Task;
@b{end} Decls1;

@b{package} Decls2 @b{is}
  @b{type} My_Int @b{is} @b{new} Integer;
  @b{function} Ident (M : My_Int) @b{return} My_Int;
@b{end} Decls2;

@b{with} Utils;
@b{package} @b{body} Decls2 @b{is}
  @b{function} Ident (M : My_Int) @b{return} My_Int @b{is}
  @b{begin}
     @b{return} M;
  @b{end} Ident;
@b{end} Decls2;

@b{with} Decls2;
@b{package} Utils @b{is}
  @b{procedure} Put_Val (Arg : Decls2.My_Int);
@b{end} Utils;

@b{with} Text_IO;
@b{package} @b{body} Utils @b{is}
  @b{procedure} Put_Val (Arg : Decls2.My_Int) @b{is}
  @b{begin}
     Text_IO.Put_Line (Decls2.My_Int'Image (Decls2.Ident (Arg)));
  @b{end} Put_Val;
@b{end} Utils;

@b{with} Decls1;
@b{procedure} Main @b{is}
@b{begin}
   Decls1.Lib_Task.Start;
@b{end};
@end smallexample

@noindent
All we have done is to split @code{Decls} into two packages, one
containing the library task, and one containing everything else. Now
there is no cycle, and the program compiles, binds, links and executes
using the default static model of elaboration.

@item
Declare separate task types.

A significant part of the problem arises because of the use of the
single task declaration form. This means that the elaboration of
the task type, and the elaboration of the task itself (i.e.@: the
creation of the task) happen at the same time. A good rule
of style in Ada is to always create explicit task types. By
following the additional step of placing task objects in separate
packages from the task type declaration, many elaboration problems
are avoided. Here is another modified example of the example program:

@smallexample @c ada
@b{package} Decls @b{is}
  @b{task} @b{type} Lib_Task_Type @b{is}
     @b{entry} Start;
  @b{end} Lib_Task_Type;

  @b{type} My_Int @b{is} @b{new} Integer;

  @b{function} Ident (M : My_Int) @b{return} My_Int;
@b{end} Decls;

@b{with} Utils;
@b{package} @b{body} Decls @b{is}
  @b{task} @b{body} Lib_Task_Type @b{is}
  @b{begin}
     @b{accept} Start;
     Utils.Put_Val (2);
  @b{end} Lib_Task_Type;

  @b{function} Ident (M : My_Int) @b{return} My_Int @b{is}
  @b{begin}
     @b{return} M;
  @b{end} Ident;
@b{end} Decls;

@b{with} Decls;
@b{package} Utils @b{is}
  @b{procedure} Put_Val (Arg : Decls.My_Int);
@b{end} Utils;

@b{with} Text_IO;
@b{package} @b{body} Utils @b{is}
  @b{procedure} Put_Val (Arg : Decls.My_Int) @b{is}
  @b{begin}
     Text_IO.Put_Line (Decls.My_Int'Image (Decls.Ident (Arg)));
  @b{end} Put_Val;
@b{end} Utils;

@b{with} Decls;
@b{package} Declst @b{is}
   Lib_Task : Decls.Lib_Task_Type;
@b{end} Declst;

@b{with} Declst;
@b{procedure} Main @b{is}
@b{begin}
   Declst.Lib_Task.Start;
@b{end};
@end smallexample

@noindent
What we have done here is to replace the @code{task} declaration in
package @code{Decls} with a @code{task type} declaration. Then we
introduce a separate package @code{Declst} to contain the actual
task object. This separates the elaboration issues for
the @code{task type}
declaration, which causes no trouble, from the elaboration issues
of the task object, which is also unproblematic, since it is now independent
of the elaboration of  @code{Utils}.
This separation of concerns also corresponds to
a generally sound engineering principle of separating declarations
from instances. This version of the program also compiles, binds, links,
and executes, generating the expected output.

@item
Use No_Entry_Calls_In_Elaboration_Code restriction.
@cindex No_Entry_Calls_In_Elaboration_Code

The previous two approaches described how a program can be restructured
to avoid the special problems caused by library task bodies. in practice,
however, such restructuring may be difficult to apply to existing legacy code,
so we must consider solutions that do not require massive rewriting.

Let us consider more carefully why our original sample program works
under the dynamic model of elaboration. The reason is that the code
in the task body blocks immediately on the @code{accept}
statement. Now of course there is nothing to prohibit elaboration
code from making entry calls (for example from another library level task),
so we cannot tell in isolation that
the task will not execute the accept statement  during elaboration.

However, in practice it is very unusual to see elaboration code
make any entry calls, and the pattern of tasks starting
at elaboration time and then immediately blocking on @code{accept} or
@code{select} statements is very common. What this means is that
the compiler is being too pessimistic when it analyzes the
whole package body as though it might be executed at elaboration
time.

If we know that the elaboration code contains no entry calls, (a very safe
assumption most of the time, that could almost be made the default
behavior), then we can compile all units of the program under control
of the following configuration pragma:

@smallexample
pragma Restrictions (No_Entry_Calls_In_Elaboration_Code);
@end smallexample

@noindent
This pragma can be placed in the @file{gnat.adc} file in the usual
manner. If we take our original unmodified program and compile it
in the presence of a @file{gnat.adc} containing the above pragma,
then once again, we can compile, bind, link, and execute, obtaining
the expected result. In the presence of this pragma, the compiler does
not trace calls in a task body, that appear after the first @code{accept}
or @code{select} statement, and therefore does not report a potential
circularity in the original program.

The compiler will check to the extent it can that the above
restriction is not violated, but it is not always possible to do a
complete check at compile time, so it is important to use this
pragma only if the stated restriction is in fact met, that is to say
no task receives an entry call before elaboration of all units is completed.

@end itemize

@node Mixing Elaboration Models
@section Mixing Elaboration Models
@noindent
So far, we have assumed that the entire program is either compiled
using the dynamic model or static model, ensuring consistency. It
is possible to mix the two models, but rules have to be followed
if this mixing is done to ensure that elaboration checks are not
omitted.

The basic rule is that @emph{a unit compiled with the static model cannot
be @code{with'ed} by a unit compiled with the dynamic model}. The
reason for this is that in the static model, a unit assumes that
its clients guarantee to use (the equivalent of) pragma
@code{Elaborate_All} so that no elaboration checks are required
in inner subprograms, and this assumption is violated if the
client is compiled with dynamic checks.

The precise rule is as follows. A unit that is compiled with dynamic
checks can only @code{with} a unit that meets at least one of the
following criteria:

@itemize @bullet

@item
The @code{with'ed} unit is itself compiled with dynamic elaboration
checks (that is with the @option{-gnatE} switch.

@item
The @code{with'ed} unit is an internal GNAT implementation unit from
the System, Interfaces, Ada, or GNAT hierarchies.

@item
The @code{with'ed} unit has pragma Preelaborate or pragma Pure.

@item
The @code{with'ing} unit (that is the client) has an explicit pragma
@code{Elaborate_All} for the @code{with'ed} unit.

@end itemize

@noindent
If this rule is violated, that is if a unit with dynamic elaboration
checks @code{with's} a unit that does not meet one of the above four
criteria, then the binder (@code{gnatbind}) will issue a warning
similar to that in the following example:

@smallexample
warning: "x.ads" has dynamic elaboration checks and with's
warning:   "y.ads" which has static elaboration checks
@end smallexample

@noindent
These warnings indicate that the rule has been violated, and that as a result
elaboration checks may be missed in the resulting executable file.
This warning may be suppressed using the @option{-ws} binder switch
in the usual manner.

One useful application of this mixing rule is in the case of a subsystem
which does not itself @code{with} units from the remainder of the
application. In this case, the entire subsystem can be compiled with
dynamic checks to resolve a circularity in the subsystem, while
allowing the main application that uses this subsystem to be compiled
using the more reliable default static model.

@node What to Do If the Default Elaboration Behavior Fails
@section What to Do If the Default Elaboration Behavior Fails

@noindent
If the binder cannot find an acceptable order, it outputs detailed
diagnostics. For example:
@smallexample
@group
@iftex
@leftskip=0cm
@end iftex
error: elaboration circularity detected
info:   "proc (body)" must be elaborated before "pack (body)"
info:     reason: Elaborate_All probably needed in unit "pack (body)"
info:     recompile "pack (body)" with -gnatel
info:                             for full details
info:       "proc (body)"
info:         is needed by its spec:
info:       "proc (spec)"
info:         which is withed by:
info:       "pack (body)"
info:  "pack (body)" must be elaborated before "proc (body)"
info:     reason: pragma Elaborate in unit "proc (body)"
@end group

@end smallexample

@noindent
In this case we have a cycle that the binder cannot break. On the one
hand, there is an explicit pragma Elaborate in @code{proc} for
@code{pack}. This means that the body of @code{pack} must be elaborated
before the body of @code{proc}. On the other hand, there is elaboration
code in @code{pack} that calls a subprogram in @code{proc}. This means
that for maximum safety, there should really be a pragma
Elaborate_All in @code{pack} for @code{proc} which would require that
the body of @code{proc} be elaborated before the body of
@code{pack}. Clearly both requirements cannot be satisfied.
Faced with a circularity of this kind, you have three different options.

@table @asis
@item Fix the program
The most desirable option from the point of view of long-term maintenance
is to rearrange the program so that the elaboration problems are avoided.
One useful technique is to place the elaboration code into separate
child packages. Another is to move some of the initialization code to
explicitly called subprograms, where the program controls the order
of initialization explicitly. Although this is the most desirable option,
it may be impractical and involve too much modification, especially in
the case of complex legacy code.

@item Perform dynamic checks
If the compilations are done using the
@option{-gnatE}
(dynamic elaboration check) switch, then GNAT behaves in a quite different
manner. Dynamic checks are generated for all calls that could possibly result
in raising an exception. With this switch, the compiler does not generate
implicit @code{Elaborate} or @code{Elaborate_All} pragmas. The behavior then is
exactly as specified in the @cite{Ada Reference Manual}.
The binder will generate
an executable program that may or may not raise @code{Program_Error}, and then
it is the programmer's job to ensure that it does not raise an exception. Note
that it is important to compile all units with the switch, it cannot be used
selectively.

@item Suppress checks
The drawback of dynamic checks is that they generate a
significant overhead at run time, both in space and time. If you
are absolutely sure that your program cannot raise any elaboration
exceptions, and you still want to use the dynamic elaboration model,
then you can use the configuration pragma
@code{Suppress (Elaboration_Check)} to suppress all such checks. For
example this pragma could be placed in the @file{gnat.adc} file.

@item Suppress checks selectively
When you know that certain calls or instantiations in elaboration code cannot
possibly lead to an elaboration error, and the binder nevertheless complains
about implicit @code{Elaborate} and @code{Elaborate_All} pragmas that lead to
elaboration circularities, it is possible to remove those warnings locally and
obtain a program that will bind. Clearly this can be unsafe, and it is the
responsibility of the programmer to make sure that the resulting program has no
elaboration anomalies. The pragma @code{Suppress (Elaboration_Check)} can be
used with different granularity to suppress warnings and break elaboration
circularities:

@itemize @bullet
@item
Place the pragma that names the called subprogram in the declarative part
that contains the call.

@item
Place the pragma in the declarative part, without naming an entity. This
disables warnings on all calls in the corresponding  declarative region.

@item
Place the pragma in the package spec that declares the called subprogram,
and name the subprogram. This disables warnings on all elaboration calls to
that subprogram.

@item
Place the pragma in the package spec that declares the called subprogram,
without naming any entity. This disables warnings on all elaboration calls to
all subprograms declared in this spec.

@item Use Pragma Elaborate
As previously described in section @xref{Treatment of Pragma Elaborate},
GNAT in static mode assumes that a @code{pragma} Elaborate indicates correctly
that no elaboration checks are required on calls to the designated unit.
There may be cases in which the caller knows that no transitive calls
can occur, so that a @code{pragma Elaborate} will be sufficient in a
case where @code{pragma Elaborate_All} would cause a circularity.
@end itemize

@noindent
These five cases are listed in order of decreasing safety, and therefore
require increasing programmer care in their application. Consider the
following program:

@smallexample @c adanocomment
@b{package} Pack1 @b{is}
  @b{function} F1 @b{return} Integer;
  X1 : Integer;
@b{end} Pack1;

@b{package} Pack2 @b{is}
  @b{function} F2 @b{return} Integer;
  @b{function} Pure (x : integer) @b{return} integer;
  --  pragma Suppress (Elaboration_Check, On => Pure);  -- (3)
  --  pragma Suppress (Elaboration_Check);              -- (4)
@b{end} Pack2;

@b{with} Pack2;
@b{package} @b{body} Pack1 @b{is}
  @b{function} F1 @b{return} Integer @b{is}
  @b{begin}
    @b{return} 100;
  @b{end} F1;
  Val : integer := Pack2.Pure (11);    --  Elab. call (1)
@b{begin}
  @b{declare}
    --  pragma Suppress(Elaboration_Check, Pack2.F2);   -- (1)
    --  pragma Suppress(Elaboration_Check);             -- (2)
  @b{begin}
    X1 := Pack2.F2 + 1;                --  Elab. call (2)
  @b{end};
@b{end} Pack1;

@b{with} Pack1;
@b{package} @b{body} Pack2 @b{is}
  @b{function} F2 @b{return} Integer @b{is}
  @b{begin}
     @b{return} Pack1.F1;
  @b{end} F2;
  @b{function} Pure (x : integer) @b{return} integer @b{is}
  @b{begin}
     @b{return} x ** 3 - 3 * x;
  @b{end};
@b{end} Pack2;

@b{with} Pack1, Ada.Text_IO;
@b{procedure} Proc3 @b{is}
@b{begin}
  Ada.Text_IO.Put_Line(Pack1.X1'Img); -- 101
@b{end} Proc3;
@end smallexample
In the absence of any pragmas, an attempt to bind this program produces
the following diagnostics:
@smallexample
@group
@iftex
@leftskip=.5cm
@end iftex
error: elaboration circularity detected
info:    "pack1 (body)" must be elaborated before "pack1 (body)"
info:       reason: Elaborate_All probably needed in unit "pack1 (body)"
info:       recompile "pack1 (body)" with -gnatel for full details
info:          "pack1 (body)"
info:             must be elaborated along with its spec:
info:          "pack1 (spec)"
info:             which is withed by:
info:          "pack2 (body)"
info:             which must be elaborated along with its spec:
info:          "pack2 (spec)"
info:             which is withed by:
info:          "pack1 (body)"
@end group
@end smallexample
The sources of the circularity are the two calls to @code{Pack2.Pure} and
@code{Pack2.F2} in the body of @code{Pack1}. We can see that the call to
F2 is safe, even though F2 calls F1, because the call appears after the
elaboration of the body of F1. Therefore the pragma (1) is safe, and will
remove the warning on the call. It is also possible to use pragma (2)
because there are no other potentially unsafe calls in the block.

@noindent
The call to @code{Pure} is safe because this function does not depend on the
state of @code{Pack2}. Therefore any call to this function is safe, and it
is correct to place pragma (3) in the corresponding package spec.

@noindent
Finally, we could place pragma (4) in the spec of @code{Pack2} to disable
warnings on all calls to functions declared therein. Note that this is not
necessarily safe, and requires more detailed examination of the subprogram
bodies involved. In particular, a call to @code{F2} requires that @code{F1}
be already elaborated.
@end table

@noindent
It is hard to generalize on which of these four approaches should be
taken. Obviously if it is possible to fix the program so that the default
treatment works, this is preferable, but this may not always be practical.
It is certainly simple enough to use
@option{-gnatE}
but the danger in this case is that, even if the GNAT binder
finds a correct elaboration order, it may not always do so,
and certainly a binder from another Ada compiler might not. A
combination of testing and analysis (for which the
information messages generated with the
@option{-gnatel}
switch can be useful) must be used to ensure that the program is free
of errors. One switch that is useful in this testing is the
@option{-p (pessimistic elaboration order)}
switch for
@code{gnatbind}.
Normally the binder tries to find an order that has the best chance
of avoiding elaboration problems. However, if this switch is used, the binder
plays a devil's advocate role, and tries to choose the order that
has the best chance of failing. If your program works even with this
switch, then it has a better chance of being error free, but this is still
not a guarantee.

For an example of this approach in action, consider the C-tests (executable
tests) from the ACVC suite. If these are compiled and run with the default
treatment, then all but one of them succeed without generating any error
diagnostics from the binder. However, there is one test that fails, and
this is not surprising, because the whole point of this test is to ensure
that the compiler can handle cases where it is impossible to determine
a correct order statically, and it checks that an exception is indeed
raised at run time.

This one test must be compiled and run using the
@option{-gnatE}
switch, and then it passes. Alternatively, the entire suite can
be run using this switch. It is never wrong to run with the dynamic
elaboration switch if your code is correct, and we assume that the
C-tests are indeed correct (it is less efficient, but efficiency is
not a factor in running the ACVC tests.)

@node Elaboration for Indirect Calls
@section Elaboration for Indirect Calls
@cindex Dispatching calls
@cindex Indirect calls

@noindent
In rare cases, the static elaboration model fails to prevent
dispatching calls to not-yet-elaborated subprograms. In such cases, we
fall back to run-time checks; premature calls to any primitive
operation of a tagged type before the body of the operation has been
elaborated will raise @code{Program_Error}.

Access-to-subprogram types, however, are handled conservatively, and
do not require run-time checks. This was not true in earlier versions
of the compiler; you can use the @option{-gnatd.U} debug switch to
revert to the old behavior if the new conservative behavior causes
elaboration cycles. Here, ``conservative'' means that if you do
@code{P'Access} during elaboration, the compiler will assume that you
might call @code{P} indirectly during elaboration, so it adds an
implicit @code{pragma Elaborate_All} on the library unit containing
@code{P}. The @option{-gnatd.U} switch is safe if you know there are
no such calls. If the program worked before, it will continue to work
with @option{-gnatd.U}. But beware that code modifications such as
adding an indirect call can cause erroneous behavior in the presence
of @option{-gnatd.U}.

@node Summary of Procedures for Elaboration Control
@section Summary of Procedures for Elaboration Control
@cindex Elaboration control

@noindent
First, compile your program with the default options, using none of
the special elaboration control switches. If the binder successfully
binds your program, then you can be confident that, apart from issues
raised by the use of access-to-subprogram types and dynamic dispatching,
the program is free of elaboration errors. If it is important that the
program be portable to other compilers than GNAT, then use the
@option{-gnatel}
switch to generate messages about missing @code{Elaborate} or
@code{Elaborate_All} pragmas, and supply the missing pragmas.

If the program fails to bind using the default static elaboration
handling, then you can fix the program to eliminate the binder
message, or recompile the entire program with the
@option{-gnatE} switch to generate dynamic elaboration checks,
and, if you are sure there really are no elaboration problems,
use a global pragma @code{Suppress (Elaboration_Check)}.

@node Other Elaboration Order Considerations
@section Other Elaboration Order Considerations
@noindent
This section has been entirely concerned with the issue of finding a valid
elaboration order, as defined by the Ada Reference Manual. In a case
where several elaboration orders are valid, the task is to find one
of the possible valid elaboration orders (and the static model in GNAT
will ensure that this is achieved).

The purpose of the elaboration rules in the Ada Reference Manual is to
make sure that no entity is accessed before it has been elaborated. For
a subprogram, this means that the spec and body must have been elaborated
before the subprogram is called. For an object, this means that the object
must have been elaborated before its value is read or written. A violation
of either of these two requirements is an access before elaboration order,
and this section has been all about avoiding such errors.

In the case where more than one order of elaboration is possible, in the
sense that access before elaboration errors are avoided, then any one of
the orders is ``correct'' in the sense that it meets the requirements of
the Ada Reference Manual, and no such error occurs.

However, it may be the case for a given program, that there are
constraints on the order of elaboration that come not from consideration
of avoiding elaboration errors, but rather from extra-lingual logic
requirements. Consider this example:

@smallexample @c ada
@b{with} Init_Constants;
@b{package} Constants @b{is}
   X : Integer := 0;
   Y : Integer := 0;
@b{end} Constants;

@b{package} Init_Constants @b{is}
   @b{procedure} P; --@i{ require a body}
@b{end} Init_Constants;

@b{with} Constants;
@b{package} @b{body} Init_Constants @b{is}
   @b{procedure} P @b{is} @b{begin} @b{null}; @b{end};
@b{begin}
   Constants.X := 3;
   Constants.Y := 4;
@b{end} Init_Constants;

@b{with} Constants;
@b{package} Calc @b{is}
   Z : Integer := Constants.X + Constants.Y;
@b{end} Calc;

@b{with} Calc;
@b{with} Text_IO; @b{use} Text_IO;
@b{procedure} Main @b{is}
@b{begin}
   Put_Line (Calc.Z'Img);
@b{end} Main;
@end smallexample

@noindent
In this example, there is more than one valid order of elaboration. For
example both the following are correct orders:

@smallexample
Init_Constants spec
Constants spec
Calc spec
Init_Constants body
Main body

  and

Init_Constants spec
Init_Constants body
Constants spec
Calc spec
Main body
@end smallexample

@noindent
There is no language rule to prefer one or the other, both are correct
from an order of elaboration point of view. But the programmatic effects
of the two orders are very different. In the first, the elaboration routine
of @code{Calc} initializes @code{Z} to zero, and then the main program
runs with this value of zero. But in the second order, the elaboration
routine of @code{Calc} runs after the body of Init_Constants has set
@code{X} and @code{Y} and thus @code{Z} is set to 7 before @code{Main}
runs.

One could perhaps by applying pretty clever non-artificial intelligence
to the situation guess that it is more likely that the second order of
elaboration is the one desired, but there is no formal linguistic reason
to prefer one over the other. In fact in this particular case, GNAT will
prefer the second order, because of the rule that bodies are elaborated
as soon as possible, but it's just luck that this is what was wanted
(if indeed the second order was preferred).

If the program cares about the order of elaboration routines in a case like
this, it is important to specify the order required. In this particular
case, that could have been achieved by adding to the spec of Calc:

@smallexample @c ada
@b{pragma} Elaborate_All (Constants);
@end smallexample

@noindent
which requires that the body (if any) and spec of @code{Constants},
as well as the body and spec of any unit @code{with}'ed by
@code{Constants} be elaborated before @code{Calc} is elaborated.

Clearly no automatic method can always guess which alternative you require,
and if you are working with legacy code that had constraints of this kind
which were not properly specified by adding @code{Elaborate} or
@code{Elaborate_All} pragmas, then indeed it is possible that two different
compilers can choose different orders.

However, GNAT does attempt to diagnose the common situation where there
are uninitialized variables in the visible part of a package spec, and the
corresponding package body has an elaboration block that directly or
indirectly initialized one or more of these variables. This is the situation
in which a pragma Elaborate_Body is usually desirable, and GNAT will generate
a warning that suggests this addition if it detects this situation.

The @code{gnatbind}
@option{-p} switch may be useful in smoking
out problems. This switch causes bodies to be elaborated as late as possible
instead of as early as possible. In the example above, it would have forced
the choice of the first elaboration order. If you get different results
when using this switch, and particularly if one set of results is right,
and one is wrong as far as you are concerned, it shows that you have some
missing @code{Elaborate} pragmas. For the example above, we have the
following output:

@smallexample
gnatmake -f -q main
main
 7
gnatmake -f -q main -bargs -p
main
 0
@end smallexample

@noindent
It is of course quite unlikely that both these results are correct, so
it is up to you in a case like this to investigate the source of the
difference, by looking at the two elaboration orders that are chosen,
and figuring out which is correct, and then adding the necessary
@code{Elaborate} or @code{Elaborate_All} pragmas to ensure the desired order.

@node Determining the Chosen Elaboration Order
@section Determining the Chosen Elaboration Order
@noindent

To see the elaboration order that the binder chooses, you can look at
the last part of the b~xxx.adb binder output file. Here is an example:

@smallexample @c ada
System.Soft_Links'Elab_Body;
E14 := True;
System.Secondary_Stack'Elab_Body;
E18 := True;
System.Exception_Table'Elab_Body;
E24 := True;
Ada.Io_Exceptions'Elab_Spec;
E67 := True;
Ada.Tags'Elab_Spec;
Ada.Streams'Elab_Spec;
E43 := True;
Interfaces.C'Elab_Spec;
E69 := True;
System.Finalization_Root'Elab_Spec;
E60 := True;
System.Os_Lib'Elab_Body;
E71 := True;
System.Finalization_Implementation'Elab_Spec;
System.Finalization_Implementation'Elab_Body;
E62 := True;
Ada.Finalization'Elab_Spec;
E58 := True;
Ada.Finalization.List_Controller'Elab_Spec;
E76 := True;
System.File_Control_Block'Elab_Spec;
E74 := True;
System.File_Io'Elab_Body;
E56 := True;
Ada.Tags'Elab_Body;
E45 := True;
Ada.Text_Io'Elab_Spec;
Ada.Text_Io'Elab_Body;
E07 := True;
@end smallexample

@noindent
Here Elab_Spec elaborates the spec
and Elab_Body elaborates the body. The assignments to the Exx flags
flag that the corresponding body is now elaborated.

You can also ask the binder to generate a more
readable list of the elaboration order using the
@code{-l} switch when invoking the binder. Here is
an example of the output generated by this switch:

@smallexample
ada (spec)
interfaces (spec)
system (spec)
system.case_util (spec)
system.case_util (body)
system.concat_2 (spec)
system.concat_2 (body)
system.concat_3 (spec)
system.concat_3 (body)
system.htable (spec)
system.parameters (spec)
system.parameters (body)
system.crtl (spec)
interfaces.c_streams (spec)
interfaces.c_streams (body)
system.restrictions (spec)
system.restrictions (body)
system.standard_library (spec)
system.exceptions (spec)
system.exceptions (body)
system.storage_elements (spec)
system.storage_elements (body)
system.secondary_stack (spec)
system.stack_checking (spec)
system.stack_checking (body)
system.string_hash (spec)
system.string_hash (body)
system.htable (body)
system.strings (spec)
system.strings (body)
system.traceback (spec)
system.traceback (body)
system.traceback_entries (spec)
system.traceback_entries (body)
ada.exceptions (spec)
ada.exceptions.last_chance_handler (spec)
system.soft_links (spec)
system.soft_links (body)
ada.exceptions.last_chance_handler (body)
system.secondary_stack (body)
system.exception_table (spec)
system.exception_table (body)
ada.io_exceptions (spec)
ada.tags (spec)
ada.streams (spec)
interfaces.c (spec)
interfaces.c (body)
system.finalization_root (spec)
system.finalization_root (body)
system.memory (spec)
system.memory (body)
system.standard_library (body)
system.os_lib (spec)
system.os_lib (body)
system.unsigned_types (spec)
system.stream_attributes (spec)
system.stream_attributes (body)
system.finalization_implementation (spec)
system.finalization_implementation (body)
ada.finalization (spec)
ada.finalization (body)
ada.finalization.list_controller (spec)
ada.finalization.list_controller (body)
system.file_control_block (spec)
system.file_io (spec)
system.file_io (body)
system.val_uns (spec)
system.val_util (spec)
system.val_util (body)
system.val_uns (body)
system.wch_con (spec)
system.wch_con (body)
system.wch_cnv (spec)
system.wch_jis (spec)
system.wch_jis (body)
system.wch_cnv (body)
system.wch_stw (spec)
system.wch_stw (body)
ada.tags (body)
ada.exceptions (body)
ada.text_io (spec)
ada.text_io (body)
text_io (spec)
gdbstr (body)
@end smallexample

@c **********************************
@node Overflow Check Handling in GNAT
@appendix Overflow Check Handling in GNAT
@cindex Overflow checks
@cindex Checks (overflow)
@c **********************************

@menu
* Background::
* Overflow Checking Modes in GNAT::
* Specifying the Desired Mode::
* Default Settings::
* Implementation Notes::
@end menu


@node Background
@section Background

@noindent
Overflow checks are checks that the compiler may make to ensure
that intermediate results are not out of range. For example:

@smallexample @c ada
   A : Integer;
   ...
   A := A + 1;
@end smallexample

@noindent
if @code{A} has the value @code{Integer'Last}, then the addition may cause
overflow since the result is out of range of the type @code{Integer}.
In this case @code{Constraint_Error} will be raised if checks are
enabled.

A trickier situation arises in examples like the following:

@smallexample @c ada
  A, C : Integer;
  ...
  A := (A + 1) + C;
@end smallexample

@noindent
where @code{A} is @code{Integer'Last} and @code{C} is @code{-1}.
Now the final result of the expression on the right hand side is
@code{Integer'Last} which is in range, but the question arises whether the
intermediate addition of @code{(A + 1)} raises an overflow error.

The (perhaps surprising) answer is that the Ada language
definition does not answer this question. Instead it leaves
it up to the implementation to do one of two things if overflow
checks are enabled.

@itemize @bullet
@item
raise an exception (@code{Constraint_Error}), or

@item
yield the correct mathematical result which is then used in
subsequent operations.
@end itemize

@noindent
If the compiler chooses the first approach, then the assignment of this
example will indeed raise @code{Constraint_Error} if overflow checking is
enabled, or result in erroneous execution if overflow checks are suppressed.

But if the compiler
chooses the second approach, then it can perform both additions yielding
the correct mathematical result, which is in range, so no exception
will be raised, and the right result is obtained, regardless of whether
overflow checks are suppressed.

Note that in the first example an
exception will be raised in either case, since if the compiler
gives the correct mathematical result for the addition, it will
be out of range of the target type of the assignment, and thus
fails the range check.

This lack of specified behavior in the handling of overflow for
intermediate results is a source of non-portability, and can thus
be problematic when programs are ported. Most typically this arises
in a situation where the original compiler did not raise an exception,
and then the application is moved to a compiler where the check is
performed on the intermediate result and an unexpected exception is
raised.

Furthermore, when using Ada 2012's preconditions and other
assertion forms, another issue arises. Consider:

@smallexample @c ada
     @b{procedure} P (A, B : Integer) @b{with}
       Pre => A + B <= Integer'Last;
@end smallexample

@noindent
One often wants to regard arithmetic in a context like this from
a mathematical point of view. So for example, if the two actual parameters
for a call to @code{P} are both @code{Integer'Last}, then
the precondition should be regarded as False. If we are executing
in a mode with run-time checks enabled for preconditions, then we would
like this precondition to fail, rather than raising an exception
because of the intermediate overflow.

However, the language definition leaves the specification of
whether the above condition fails (raising @code{Assert_Error}) or
causes an intermediate overflow (raising @code{Constraint_Error})
up to the implementation.

The situation is worse in a case such as the following:

@smallexample @c ada
     @b{procedure} Q (A, B, C : Integer) @b{with}
       Pre => A + B + C <= Integer'Last;
@end smallexample

@noindent
Consider the call

@smallexample @c ada
     Q (A => Integer'Last, B => 1, C => -1);
@end smallexample

@noindent
From a mathematical point of view the precondition
is True, but at run time we may (but are not guaranteed to) get an
exception raised because of the intermediate overflow (and we really
would prefer this precondition to be considered True at run time).

@node Overflow Checking Modes in GNAT
@section Overflow Checking Modes in GNAT

@noindent
To deal with the portability issue, and with the problem of
mathematical versus run-time interpretation of the expressions in
assertions, GNAT provides comprehensive control over the handling
of intermediate overflow. GNAT can operate in three modes, and
furthemore, permits separate selection of operating modes for
the expressions within assertions (here the term ``assertions''
is used in the technical sense, which includes preconditions and so forth)
and for expressions appearing outside assertions.

The three modes are:

@itemize @bullet
@item   @i{Use base type for intermediate operations} (@code{STRICT})

     In this mode, all intermediate results for predefined arithmetic
     operators are computed using the base type, and the result must
     be in range of the base type. If this is not the
     case then either an exception is raised (if overflow checks are
     enabled) or the execution is erroneous (if overflow checks are suppressed).
     This is the normal default mode.

@item   @i{Most intermediate overflows avoided} (@code{MINIMIZED})

     In this mode, the compiler attempts to avoid intermediate overflows by
     using a larger integer type, typically @code{Long_Long_Integer},
     as the type in which arithmetic is
     performed for predefined arithmetic operators. This may be slightly more
     expensive at
     run time (compared to suppressing intermediate overflow checks), though
     the cost is negligible on modern 64-bit machines. For the examples given
     earlier, no intermediate overflows would have resulted in exceptions,
     since the intermediate results are all in the range of
     @code{Long_Long_Integer} (typically 64-bits on nearly all implementations
     of GNAT). In addition, if checks are enabled, this reduces the number of
     checks that must be made, so this choice may actually result in an
     improvement in space and time behavior.

     However, there are cases where @code{Long_Long_Integer} is not large
     enough, consider the following example:

@smallexample @c ada
       @b{procedure} R (A, B, C, D : Integer) @b{with}
         Pre => (A**2 * B**2) / (C**2 * D**2) <= 10;
@end smallexample

     where @code{A} = @code{B} = @code{C} = @code{D} = @code{Integer'Last}.
     Now the intermediate results are
     out of the range of @code{Long_Long_Integer} even though the final result
     is in range and the precondition is True (from a mathematical point
     of view). In such a case, operating in this mode, an overflow occurs
     for the intermediate computation (which is why this mode
     says @i{most} intermediate overflows are avoided). In this case,
     an exception is raised if overflow checks are enabled, and the
     execution is erroneous if overflow checks are suppressed.

@item   @i{All intermediate overflows avoided} (@code{ELIMINATED})

     In this mode, the compiler  avoids all intermediate overflows
     by using arbitrary precision arithmetic as required. In this
     mode, the above example with @code{A**2 * B**2} would
     not cause intermediate overflow, because the intermediate result
     would be evaluated using sufficient precision, and the result
     of evaluating the precondition would be True.

     This mode has the advantage of avoiding any intermediate
     overflows, but at the expense of significant run-time overhead,
     including the use of a library (included automatically in this
     mode) for multiple-precision arithmetic.

     This mode provides cleaner semantics for assertions, since now
     the run-time behavior emulates true arithmetic behavior for the
     predefined arithmetic operators, meaning that there is never a
     conflict between the mathematical view of the assertion, and its
     run-time behavior.

     Note that in this mode, the behavior is unaffected by whether or
     not overflow checks are suppressed, since overflow does not occur.
     It is possible for gigantic intermediate expressions to raise
     @code{Storage_Error} as a result of attempting to compute the
     results of such expressions (e.g. @code{Integer'Last ** Integer'Last})
     but overflow is impossible.


@end itemize

@noindent
  Note that these modes apply only to the evaluation of predefined
  arithmetic, membership, and comparison operators for signed integer
  aritmetic.

  For fixed-point arithmetic, checks can be suppressed. But if checks
  are enabled
  then fixed-point values are always checked for overflow against the
  base type for intermediate expressions (that is such checks always
  operate in the equivalent of @code{STRICT} mode).

  For floating-point, on nearly all architectures, @code{Machine_Overflows}
  is False, and IEEE infinities are generated, so overflow exceptions
  are never raised. If you want to avoid infinities, and check that
  final results of expressions are in range, then you can declare a
  constrained floating-point type, and range checks will be carried
  out in the normal manner (with infinite values always failing all
  range checks).


@c -------------------------
@node Specifying the Desired Mode
@section Specifying the Desired Mode

@noindent
The desired mode of for handling intermediate overflow can be specified using
either the @code{Overflow_Mode} pragma or an equivalent compiler switch.
The pragma has the form
@cindex pragma @code{Overflow_Mode}

@smallexample @c ada
    @b{pragma} Overflow_Mode ([General =>] MODE [, [Assertions =>] MODE]);
@end smallexample

@noindent
where @code{MODE} is one of

@itemize @bullet
@item   @code{STRICT}:  intermediate overflows checked (using base type)
@item   @code{MINIMIZED}: minimize intermediate overflows
@item   @code{ELIMINATED}: eliminate intermediate overflows
@end itemize

@noindent
The case is ignored, so @code{MINIMIZED}, @code{Minimized} and
@code{minimized} all have the same effect.

If only the @code{General} parameter is present, then the given @code{MODE}
applies
to expressions both within and outside assertions. If both arguments
are present, then @code{General} applies to expressions outside assertions,
and @code{Assertions} applies to expressions within assertions. For example:

@smallexample @c ada
   @b{pragma} Overflow_Mode
     (General => Minimized, Assertions => Eliminated);
@end smallexample

@noindent
specifies that general expressions outside assertions be evaluated
in ``minimize intermediate overflows'' mode, and expressions within
assertions be evaluated in ``eliminate intermediate overflows'' mode.
This is often a reasonable choice, avoiding excessive overhead
outside assertions, but assuring a high degree of portability
when importing code from another compiler, while incurring
the extra overhead for assertion expressions to ensure that
the behavior at run time matches the expected mathematical
behavior.

The @code{Overflow_Mode} pragma has the same scoping and placement
rules as pragma @code{Suppress}, so it can occur either as a
configuration pragma, specifying a default for the whole
program, or in a declarative scope, where it applies to the
remaining declarations and statements in that scope.

Note that pragma @code{Overflow_Mode} does not affect whether
overflow checks are enabled or suppressed. It only controls the
method used to compute intermediate values. To control whether
overflow checking is enabled or suppressed, use pragma @code{Suppress}
or @code{Unsuppress} in the usual manner

Additionally, a compiler switch @option{-gnato?} or @option{-gnato??}
can be used to control the checking mode default (which can be subsequently
overridden using pragmas).
@cindex @option{-gnato?} (gcc)
@cindex @option{-gnato??} (gcc)

Here `@code{?}' is one of the digits `@code{1}' through `@code{3}':

@itemize @bullet
@item   @code{1}:
use base type for intermediate operations (@code{STRICT})
@item   @code{2}:
minimize intermediate overflows (@code{MINIMIZED})
@item   @code{3}:
eliminate intermediate overflows (@code{ELIMINATED})
@end itemize

@noindent
As with the pragma, if only one digit appears then it applies to all
cases; if two digits are given, then the first applies outside
assertions, and the second within assertions. Thus the equivalent
of the example pragma above would be
@option{-gnato23}.

If no digits follow the @option{-gnato}, then it is equivalent to
@option{-gnato11},
causing all intermediate operations to be computed using the base
type (@code{STRICT} mode).

In addition to setting the mode used for computation of intermediate
results, the @code{-gnato} switch also enables overflow checking (which
is suppressed by default). It thus combines the effect of using
a pragma @code{Overflow_Mode} and pragma @code{Unsuppress}.


@c -------------------------
@node Default Settings
@section Default Settings

The default mode for overflow checks is

@smallexample
   General => Strict
@end smallexample

@noindent
which causes all computations both inside and outside assertions to use
the base type. In addition overflow checks are suppressed.

This retains compatibility with previous versions of
GNAT which suppressed overflow checks by default and always
used the base type for computation of intermediate results.

The switch @option{-gnato} (with no digits following) is equivalent to
@cindex @option{-gnato} (gcc)

@smallexample
   General => Strict
@end smallexample

@noindent
which causes overflow checking of all intermediate overflows
both inside and outside assertions against the base type.
This provides compatibility
with this switch as implemented in previous versions of GNAT.

The pragma @code{Suppress (Overflow_Check)} disables overflow
checking, but it has no effect on the method used for computing
intermediate results.

The pragma @code{Unsuppress (Overflow_Check)} enables overflow
checking, but it has no effect on the method used for computing
intermediate results.

@c -------------------------
@node Implementation Notes
@section Implementation Notes

In practice on typical 64-bit machines, the @code{MINIMIZED} mode is
reasonably efficient, and can be generally used. It also helps
to ensure compatibility with code imported from some other
compiler to GNAT.

Setting all intermediate overflows checking (@code{CHECKED} mode)
makes sense if you want to
make sure that your code is compatible with any other possible
Ada implementation. This may be useful in ensuring portability
for code that is to be exported to some other compiler than GNAT.


The Ada standard allows the reassociation of expressions at
the same precedence level if no parentheses are present. For
example, @w{@code{A+B+C}} parses as though it were @w{@code{(A+B)+C}}, but
the compiler can reintepret this as @w{@code{A+(B+C)}}, possibly
introducing or eliminating an overflow exception. The GNAT
compiler never takes advantage of this freedom, and the
expression @w{@code{A+B+C}} will be evaluated as @w{@code{(A+B)+C}}.
If you need the other order, you can write the parentheses
explicitly @w{@code{A+(B+C)}} and GNAT will respect this order.

The use of @code{ELIMINATED} mode will cause the compiler to
automatically include an appropriate arbitrary precision
integer arithmetic package. The compiler will make calls
to this package, though only in cases where it cannot be
sure that @code{Long_Long_Integer} is sufficient to guard against
intermediate overflows. This package does not use dynamic
alllocation, but it does use the secondary stack, so an
appropriate secondary stack package must be present (this
is always true for standard full Ada, but may require
specific steps for restricted run times such as ZFP).

Although @code{ELIMINATED} mode causes expressions to use arbitrary
precision arithmetic, avoiding overflow, the final result
must be in an appropriate range. This is true even if the
final result is of type @code{[Long_[Long_]]Integer'Base}, which
still has the same bounds as its associated constrained
type at run-time.

Currently, the @code{ELIMINATED} mode is only available on target
platforms for which @code{Long_Long_Integer} is 64-bits (nearly all GNAT
platforms).

@c *******************************
@node Conditional Compilation
@appendix Conditional Compilation
@c *******************************
@cindex Conditional compilation

@noindent
It is often necessary to arrange for a single source program
to serve multiple purposes, where it is compiled in different
ways to achieve these different goals. Some examples of the
need for this feature are

@itemize @bullet
@item  Adapting a program to a different hardware environment
@item  Adapting a program to a different target architecture
@item  Turning debugging features on and off
@item  Arranging for a program to compile with different compilers
@end itemize

@noindent
In C, or C++, the typical approach would be to use the preprocessor
that is defined as part of the language. The Ada language does not
contain such a feature. This is not an oversight, but rather a very
deliberate design decision, based on the experience that overuse of
the preprocessing features in C and C++ can result in programs that
are extremely difficult to maintain. For example, if we have ten
switches that can be on or off, this means that there are a thousand
separate programs, any one of which might not even be syntactically
correct, and even if syntactically correct, the resulting program
might not work correctly. Testing all combinations can quickly become
impossible.

Nevertheless, the need to tailor programs certainly exists, and in
this Appendix we will discuss how this can
be achieved using Ada in general, and GNAT in particular.

@menu
* Use of Boolean Constants::
* Debugging - A Special Case::
* Conditionalizing Declarations::
* Use of Alternative Implementations::
* Preprocessing::
@end menu

@node Use of Boolean Constants
@section Use of Boolean Constants

@noindent
In the case where the difference is simply which code
sequence is executed, the cleanest solution is to use Boolean
constants to control which code is executed.

@smallexample @c ada
@group
FP_Initialize_Required : @b{constant} Boolean := True;
@dots{}
@b{if} FP_Initialize_Required @b{then}
@dots{}
@b{end} @b{if};
@end group
@end smallexample

@noindent
Not only will the code inside the @code{if} statement not be executed if
the constant Boolean is @code{False}, but it will also be completely
deleted from the program.
However, the code is only deleted after the @code{if} statement
has been checked for syntactic and semantic correctness.
(In contrast, with preprocessors the code is deleted before the
compiler ever gets to see it, so it is not checked until the switch
is turned on.)
@cindex Preprocessors (contrasted with conditional compilation)

Typically the Boolean constants will be in a separate package,
something like:

@smallexample @c ada
@group
@b{package} Config @b{is}
   FP_Initialize_Required : @b{constant} Boolean := True;
   Reset_Available        : @b{constant} Boolean := False;
   @dots{}
@b{end} Config;
@end group
@end smallexample

@noindent
The @code{Config} package exists in multiple forms for the various targets,
with an appropriate script selecting the version of @code{Config} needed.
Then any other unit requiring conditional compilation can do a @code{with}
of @code{Config} to make the constants visible.


@node Debugging - A Special Case
@section Debugging - A Special Case

@noindent
A common use of conditional code is to execute statements (for example
dynamic checks, or output of intermediate results) under control of a
debug switch, so that the debugging behavior can be turned on and off.
This can be done using a Boolean constant to control whether the code
is active:

@smallexample @c ada
@group
@b{if} Debugging @b{then}
   Put_Line ("got to the first stage!");
@b{end} @b{if};
@end group
@end smallexample

@noindent
or

@smallexample @c ada
@group
@b{if} Debugging @b{and} @b{then} Temperature > 999.0 @b{then}
   @b{raise} Temperature_Crazy;
@b{end} @b{if};
@end group
@end smallexample

@noindent
Since this is a common case, there are special features to deal with
this in a convenient manner. For the case of tests, Ada 2005 has added
a pragma @code{Assert} that can be used for such tests. This pragma is modeled
@cindex pragma @code{Assert}
on the @code{Assert} pragma that has always been available in GNAT, so this
feature may be used with GNAT even if you are not using Ada 2005 features.
The use of pragma @code{Assert} is described in
@ref{Pragma Assert,,, gnat_rm, GNAT Reference Manual}, but as an
example, the last test could be written:

@smallexample @c ada
@b{pragma} Assert (Temperature <= 999.0, "Temperature Crazy");
@end smallexample

@noindent
or simply

@smallexample @c ada
@b{pragma} Assert (Temperature <= 999.0);
@end smallexample

@noindent
In both cases, if assertions are active and the temperature is excessive,
the exception @code{Assert_Failure} will be raised, with the given string in
the first case or a string indicating the location of the pragma in the second
case used as the exception message.

You can turn assertions on and off by using the @code{Assertion_Policy}
pragma.
@cindex pragma @code{Assertion_Policy}
This is an Ada 2005 pragma which is implemented in all modes by
GNAT, but only in the latest versions of GNAT which include Ada 2005
capability. Alternatively, you can use the @option{-gnata} switch
@cindex @option{-gnata} switch
to enable assertions from the command line (this is recognized by all versions
of GNAT).

For the example above with the @code{Put_Line}, the GNAT-specific pragma
@code{Debug} can be used:
@cindex pragma @code{Debug}

@smallexample @c ada
@b{pragma} Debug (Put_Line ("got to the first stage!"));
@end smallexample

@noindent
If debug pragmas are enabled, the argument, which must be of the form of
a procedure call, is executed (in this case, @code{Put_Line} will be called).
Only one call can be present, but of course a special debugging procedure
containing any code you like can be included in the program and then
called in a pragma @code{Debug} argument as needed.

One advantage of pragma @code{Debug} over the @code{if Debugging then}
construct is that pragma @code{Debug} can appear in declarative contexts,
such as at the very beginning of a procedure, before local declarations have
been elaborated.

Debug pragmas are enabled using either the @option{-gnata} switch that also
controls assertions, or with a separate Debug_Policy pragma.
@cindex pragma @code{Debug_Policy}
The latter pragma is new in the Ada 2005 versions of GNAT (but it can be used
in Ada 95 and Ada 83 programs as well), and is analogous to
pragma @code{Assertion_Policy} to control assertions.

@code{Assertion_Policy} and @code{Debug_Policy} are configuration pragmas,
and thus they can appear in @file{gnat.adc} if you are not using a
project file, or in the file designated to contain configuration pragmas
in a project file.
They then apply to all subsequent compilations. In practice the use of
the @option{-gnata} switch is often the most convenient method of controlling
the status of these pragmas.

Note that a pragma is not a statement, so in contexts where a statement
sequence is required, you can't just write a pragma on its own. You have
to add a @code{null} statement.

@smallexample @c ada
@group
@b{if} @dots{} @b{then}
   @dots{} -- some statements
@b{else}
   @b{pragma} Assert (Num_Cases < 10);
   @b{null};
@b{end} @b{if};
@end group
@end smallexample


@node Conditionalizing Declarations
@section Conditionalizing Declarations

@noindent
In some cases, it may be necessary to conditionalize declarations to meet
different requirements. For example we might want a bit string whose length
is set to meet some hardware message requirement.

In some cases, it may be possible to do this using declare blocks controlled
by conditional constants:

@smallexample @c ada
@group
@b{if} Small_Machine @b{then}
   @b{declare}
      X : Bit_String (1 .. 10);
   @b{begin}
      @dots{}
   @b{end};
@b{else}
   @b{declare}
      X : Large_Bit_String (1 .. 1000);
   @b{begin}
      @dots{}
   @b{end};
@b{end} @b{if};
@end group
@end smallexample

@noindent
Note that in this approach, both declarations are analyzed by the
compiler so this can only be used where both declarations are legal,
even though one of them will not be used.

Another approach is to define integer constants, e.g.@: @code{Bits_Per_Word},
or Boolean constants, e.g.@: @code{Little_Endian}, and then write declarations
that are parameterized by these constants. For example

@smallexample @c ada
@group
@b{for} Rec @b{use}
  Field1 @b{at} 0 @b{range} Boolean'Pos (Little_Endian) * 10 .. Bits_Per_Word;
@b{end} @b{record};
@end group
@end smallexample

@noindent
If @code{Bits_Per_Word} is set to 32, this generates either

@smallexample @c ada
@group
@b{for} Rec @b{use}
  Field1 @b{at} 0 @b{range} 0 .. 32;
@b{end} @b{record};
@end group
@end smallexample

@noindent
for the big endian case, or

@smallexample @c ada
@group
@b{for} Rec @b{use} @b{record}
  Field1 @b{at} 0 @b{range} 10 .. 32;
@b{end} @b{record};
@end group
@end smallexample

@noindent
for the little endian case. Since a powerful subset of Ada expression
notation is usable for creating static constants, clever use of this
feature can often solve quite difficult problems in conditionalizing
compilation (note incidentally that in Ada 95, the little endian
constant was introduced as @code{System.Default_Bit_Order}, so you do not
need to define this one yourself).


@node Use of Alternative Implementations
@section Use of Alternative Implementations

@noindent
In some cases, none of the approaches described above are adequate. This
can occur for example if the set of declarations required is radically
different for two different configurations.

In this situation, the official Ada way of dealing with conditionalizing
such code is to write separate units for the different cases. As long as
this does not result in excessive duplication of code, this can be done
without creating maintenance problems. The approach is to share common
code as far as possible, and then isolate the code and declarations
that are different. Subunits are often a convenient method for breaking
out a piece of a unit that is to be conditionalized, with separate files
for different versions of the subunit for different targets, where the
build script selects the right one to give to the compiler.
@cindex Subunits (and conditional compilation)

As an example, consider a situation where a new feature in Ada 2005
allows something to be done in a really nice way. But your code must be able
to compile with an Ada 95 compiler. Conceptually you want to say:

@smallexample @c ada
@group
@b{if} Ada_2005 @b{then}
   @dots{} neat Ada 2005 code
@b{else}
   @dots{} not quite as neat Ada 95 code
@b{end} @b{if};
@end group
@end smallexample

@noindent
where @code{Ada_2005} is a Boolean constant.

But this won't work when @code{Ada_2005} is set to @code{False},
since the @code{then} clause will be illegal for an Ada 95 compiler.
(Recall that although such unreachable code would eventually be deleted
by the compiler, it still needs to be legal.  If it uses features
introduced in Ada 2005, it will be illegal in Ada 95.)

So instead we write

@smallexample @c ada
@b{procedure} Insert @b{is} @b{separate};
@end smallexample

@noindent
Then we have two files for the subunit @code{Insert}, with the two sets of
code.
If the package containing this is called @code{File_Queries}, then we might
have two files

@itemize @bullet
@item    @file{file_queries-insert-2005.adb}
@item    @file{file_queries-insert-95.adb}
@end itemize

@noindent
and the build script renames the appropriate file to

@smallexample
file_queries-insert.adb
@end smallexample

@noindent
and then carries out the compilation.

This can also be done with project files' naming schemes. For example:

@smallexample @c project
For Body ("File_Queries.Insert") use "file_queries-insert-2005.ada";
@end smallexample

@noindent
Note also that with project files it is desirable to use a different extension
than @file{ads} / @file{adb} for alternative versions. Otherwise a naming
conflict may arise through another commonly used feature: to declare as part
of the project a set of directories containing all the sources obeying the
default naming scheme.

The use of alternative units is certainly feasible in all situations,
and for example the Ada part of the GNAT run-time is conditionalized
based on the target architecture using this approach. As a specific example,
consider the implementation of the AST feature in VMS. There is one
spec:

@smallexample
s-asthan.ads
@end smallexample

@noindent
which is the same for all architectures, and three bodies:

@table @file
@item    s-asthan.adb
used for all non-VMS operating systems
@item    s-asthan-vms-alpha.adb
used for VMS on the Alpha
@item    s-asthan-vms-ia64.adb
used for VMS on the ia64
@end table

@noindent
The dummy version @file{s-asthan.adb} simply raises exceptions noting that
this operating system feature is not available, and the two remaining
versions interface with the corresponding versions of VMS to provide
VMS-compatible AST handling. The GNAT build script knows the architecture
and operating system, and automatically selects the right version,
renaming it if necessary to @file{s-asthan.adb} before the run-time build.

Another style for arranging alternative implementations is through Ada's
access-to-subprogram facility.
In case some functionality is to be conditionally included,
you can declare an access-to-procedure variable @code{Ref} that is initialized
to designate a ``do nothing'' procedure, and then invoke @code{Ref.all}
when appropriate.
In some library package, set @code{Ref} to @code{Proc'Access} for some
procedure @code{Proc} that performs the relevant processing.
The initialization only occurs if the library package is included in the
program.
The same idea can also be implemented using tagged types and dispatching
calls.


@node Preprocessing
@section Preprocessing
@cindex Preprocessing

@noindent
Although it is quite possible to conditionalize code without the use of
C-style preprocessing, as described earlier in this section, it is
nevertheless convenient in some cases to use the C approach. Moreover,
older Ada compilers have often provided some preprocessing capability,
so legacy code may depend on this approach, even though it is not
standard.

To accommodate such use, GNAT provides a preprocessor (modeled to a large
extent on the various preprocessors that have been used
with legacy code on other compilers, to enable easier transition).

The preprocessor may be used in two separate modes. It can be used quite
separately from the compiler, to generate a separate output source file
that is then fed to the compiler as a separate step. This is the
@code{gnatprep} utility, whose use is fully described in
@ref{Preprocessing with gnatprep}.
@cindex @code{gnatprep}

The preprocessing language allows such constructs as

@smallexample
@group
#if DEBUG or else (PRIORITY > 4) then
   bunch of declarations
#else
   completely different bunch of declarations
#end if;
@end group
@end smallexample

@noindent
The values of the symbols @code{DEBUG} and @code{PRIORITY} can be
defined either on the command line or in a separate file.

The other way of running the preprocessor is even closer to the C style and
often more convenient. In this approach the preprocessing is integrated into
the compilation process. The compiler is fed the preprocessor input which
includes @code{#if} lines etc, and then the compiler carries out the
preprocessing internally and processes the resulting output.
For more details on this approach, see @ref{Integrated Preprocessing}.


@c *******************************
@node Inline Assembler
@appendix Inline Assembler
@c *******************************

@noindent
If you need to write low-level software that interacts directly
with the hardware, Ada provides two ways to incorporate assembly
language code into your program.  First, you can import and invoke
external routines written in assembly language, an Ada feature fully
supported by GNAT@.  However, for small sections of code it may be simpler
or more efficient to include assembly language statements directly
in your Ada source program, using the facilities of the implementation-defined
package @code{System.Machine_Code}, which incorporates the gcc
Inline Assembler.  The Inline Assembler approach offers a number of advantages,
including the following:

@itemize @bullet
@item No need to use non-Ada tools
@item Consistent interface over different targets
@item Automatic usage of the proper calling conventions
@item Access to Ada constants and variables
@item Definition of intrinsic routines
@item Possibility of inlining a subprogram comprising assembler code
@item Code optimizer can take Inline Assembler code into account
@end itemize

This chapter presents a series of examples to show you how to use
the Inline Assembler.  Although it focuses on the Intel x86,
the general approach applies also to other processors.
It is assumed that you are familiar with Ada
and with assembly language programming.

@menu
* Basic Assembler Syntax::
* A Simple Example of Inline Assembler::
* Output Variables in Inline Assembler::
* Input Variables in Inline Assembler::
* Inlining Inline Assembler Code::
* Other Asm Functionality::
@end menu

@c ---------------------------------------------------------------------------
@node Basic Assembler Syntax
@section Basic Assembler Syntax

@noindent
The assembler used by GNAT and gcc is based not on the Intel assembly
language, but rather on a language that descends from the AT&T Unix
assembler @emph{as} (and which is often referred to as ``AT&T syntax'').
The following table summarizes the main features of @emph{as} syntax
and points out the differences from the Intel conventions.
See the gcc @emph{as} and @emph{gas} (an @emph{as} macro
pre-processor) documentation for further information.

@table @asis
@item Register names
gcc / @emph{as}: Prefix with ``%''; for example @code{%eax}
@*
Intel: No extra punctuation; for example @code{eax}

@item Immediate operand
gcc / @emph{as}: Prefix with ``$''; for example @code{$4}
@*
Intel: No extra punctuation; for example @code{4}

@item Address
gcc / @emph{as}: Prefix with ``$''; for example @code{$loc}
@*
Intel: No extra punctuation; for example @code{loc}

@item Memory contents
gcc / @emph{as}: No extra punctuation; for example @code{loc}
@*
Intel: Square brackets; for example @code{[loc]}

@item Register contents
gcc / @emph{as}: Parentheses; for example @code{(%eax)}
@*
Intel: Square brackets; for example @code{[eax]}

@item Hexadecimal numbers
gcc / @emph{as}: Leading ``0x'' (C language syntax); for example @code{0xA0}
@*
Intel: Trailing ``h''; for example @code{A0h}

@item Operand size
gcc / @emph{as}: Explicit in op code; for example @code{movw} to move
a 16-bit word
@*
Intel: Implicit, deduced by assembler; for example @code{mov}

@item Instruction repetition
gcc / @emph{as}: Split into two lines; for example
@*
@code{rep}
@*
@code{stosl}
@*
Intel: Keep on one line; for example @code{rep stosl}

@item Order of operands
gcc / @emph{as}: Source first; for example @code{movw $4, %eax}
@*
Intel: Destination first; for example @code{mov eax, 4}
@end table

@c ---------------------------------------------------------------------------
@node A Simple Example of Inline Assembler
@section A Simple Example of Inline Assembler

@noindent
The following example will generate a single assembly language statement,
@code{nop}, which does nothing.  Despite its lack of run-time effect,
the example will be useful in illustrating the basics of
the Inline Assembler facility.

@smallexample @c ada
@group
@b{with} System.Machine_Code; @b{use} System.Machine_Code;
@b{procedure} Nothing @b{is}
@b{begin}
   Asm ("nop");
@b{end} Nothing;
@end group
@end smallexample

@code{Asm} is a procedure declared in package @code{System.Machine_Code};
here it takes one parameter, a @emph{template string} that must be a static
expression and that will form the generated instruction.
@code{Asm} may be regarded as a compile-time procedure that parses
the template string and additional parameters (none here),
from which it generates a sequence of assembly language instructions.

The examples in this chapter will illustrate several of the forms
for invoking @code{Asm}; a complete specification of the syntax
is found in @ref{Machine Code Insertions,,, gnat_rm, GNAT Reference
Manual}.

Under the standard GNAT conventions, the @code{Nothing} procedure
should be in a file named @file{nothing.adb}.
You can build the executable in the usual way:
@smallexample
gnatmake nothing
@end smallexample
However, the interesting aspect of this example is not its run-time behavior
but rather the generated assembly code.
To see this output, invoke the compiler as follows:
@smallexample
   gcc -c -S -fomit-frame-pointer -gnatp @file{nothing.adb}
@end smallexample
where the options are:

@table @code
@item -c
compile only (no bind or link)
@item -S
generate assembler listing
@item -fomit-frame-pointer
do not set up separate stack frames
@item -gnatp
do not add runtime checks
@end table

This gives a human-readable assembler version of the code. The resulting
file will have the same name as the Ada source file, but with a @code{.s}
extension. In our example, the file @file{nothing.s} has the following
contents:

@smallexample
@group
.file "nothing.adb"
gcc2_compiled.:
___gnu_compiled_ada:
.text
   .align 4
.globl __ada_nothing
__ada_nothing:
#APP
   nop
#NO_APP
   jmp L1
   .align 2,0x90
L1:
   ret
@end group
@end smallexample

The assembly code you included is clearly indicated by
the compiler, between the @code{#APP} and @code{#NO_APP}
delimiters. The character before the 'APP' and 'NOAPP'
can differ on different targets. For example, GNU/Linux uses '#APP' while
on NT you will see '/APP'.

If you make a mistake in your assembler code (such as using the
wrong size modifier, or using a wrong operand for the instruction) GNAT
will report this error in a temporary file, which will be deleted when
the compilation is finished.  Generating an assembler file will help
in such cases, since you can assemble this file separately using the
@emph{as} assembler that comes with gcc.

Assembling the file using the command

@smallexample
as @file{nothing.s}
@end smallexample
@noindent
will give you error messages whose lines correspond to the assembler
input file, so you can easily find and correct any mistakes you made.
If there are no errors, @emph{as} will generate an object file
@file{nothing.out}.

@c ---------------------------------------------------------------------------
@node Output Variables in Inline Assembler
@section Output Variables in Inline Assembler

@noindent
The examples in this section, showing how to access the processor flags,
illustrate how to specify the destination operands for assembly language
statements.

@smallexample @c ada
@group
@b{with} Interfaces; @b{use} Interfaces;
@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{with} System.Machine_Code; @b{use} System.Machine_Code;
@b{procedure} Get_Flags @b{is}
   Flags : Unsigned_32;
   @b{use} ASCII;
@b{begin}
   Asm ("pushfl"          & LF & HT & --@i{ push flags on stack}
        "popl %%eax"      & LF & HT & --@i{ load eax with flags}
        "movl %%eax, %0",             --@i{ store flags in variable}
        Outputs => Unsigned_32'Asm_Output ("=g", Flags));
   Put_Line ("Flags register:" & Flags'Img);
@b{end} Get_Flags;
@end group
@end smallexample

In order to have a nicely aligned assembly listing, we have separated
multiple assembler statements in the Asm template string with linefeed
(ASCII.LF) and horizontal tab (ASCII.HT) characters.
The resulting section of the assembly output file is:

@smallexample
@group
#APP
   pushfl
   popl %eax
   movl %eax, -40(%ebp)
#NO_APP
@end group
@end smallexample

It would have been legal to write the Asm invocation as:

@smallexample
Asm ("pushfl popl %%eax movl %%eax, %0")
@end smallexample

but in the generated assembler file, this would come out as:

@smallexample
#APP
   pushfl popl %eax movl %eax, -40(%ebp)
#NO_APP
@end smallexample

which is not so convenient for the human reader.

We use Ada comments
at the end of each line to explain what the assembler instructions
actually do.  This is a useful convention.

When writing Inline Assembler instructions, you need to precede each register
and variable name with a percent sign.  Since the assembler already requires
a percent sign at the beginning of a register name, you need two consecutive
percent signs for such names in the Asm template string, thus @code{%%eax}.
In the generated assembly code, one of the percent signs will be stripped off.

Names such as @code{%0}, @code{%1}, @code{%2}, etc., denote input or output
variables: operands you later define using @code{Input} or @code{Output}
parameters to @code{Asm}.
An output variable is illustrated in
the third statement in the Asm template string:
@smallexample
movl %%eax, %0
@end smallexample
The intent is to store the contents of the eax register in a variable that can
be accessed in Ada.  Simply writing @code{movl %%eax, Flags} would not
necessarily work, since the compiler might optimize by using a register
to hold Flags, and the expansion of the @code{movl} instruction would not be
aware of this optimization.  The solution is not to store the result directly
but rather to advise the compiler to choose the correct operand form;
that is the purpose of the @code{%0} output variable.

Information about the output variable is supplied in the @code{Outputs}
parameter to @code{Asm}:
@smallexample
Outputs => Unsigned_32'Asm_Output ("=g", Flags));
@end smallexample

The output is defined by the @code{Asm_Output} attribute of the target type;
the general format is
@smallexample
Type'Asm_Output (constraint_string, variable_name)
@end smallexample

The constraint string directs the compiler how
to store/access the associated variable.  In the example
@smallexample
Unsigned_32'Asm_Output ("=m", Flags);
@end smallexample
the @code{"m"} (memory) constraint tells the compiler that the variable
@code{Flags} should be stored in a memory variable, thus preventing
the optimizer from keeping it in a register.  In contrast,
@smallexample
Unsigned_32'Asm_Output ("=r", Flags);
@end smallexample
uses the @code{"r"} (register) constraint, telling the compiler to
store the variable in a register.

If the constraint is preceded by the equal character (@strong{=}), it tells
the compiler that the variable will be used to store data into it.

In the @code{Get_Flags} example, we used the @code{"g"} (global) constraint,
allowing the optimizer to choose whatever it deems best.

There are a fairly large number of constraints, but the ones that are
most useful (for the Intel x86 processor) are the following:

@table @code
@item =
output constraint
@item g
global (i.e.@: can be stored anywhere)
@item m
in memory
@item I
a constant
@item a
use eax
@item b
use ebx
@item c
use ecx
@item d
use edx
@item S
use esi
@item D
use edi
@item r
use one of eax, ebx, ecx or edx
@item q
use one of eax, ebx, ecx, edx, esi or edi
@end table

The full set of constraints is described in the gcc and @emph{as}
documentation; note that it is possible to combine certain constraints
in one constraint string.

You specify the association of an output variable with an assembler operand
through the @code{%}@emph{n} notation, where @emph{n} is a non-negative
integer.  Thus in
@smallexample @c ada
@group
Asm ("pushfl"          & LF & HT & --@i{ push flags on stack}
     "popl %%eax"      & LF & HT & --@i{ load eax with flags}
     "movl %%eax, %0",             --@i{ store flags in variable}
     Outputs => Unsigned_32'Asm_Output ("=g", Flags));
@end group
@end smallexample
@noindent
@code{%0} will be replaced in the expanded code by the appropriate operand,
whatever
the compiler decided for the @code{Flags} variable.

In general, you may have any number of output variables:
@itemize @bullet
@item
Count the operands starting at 0; thus @code{%0}, @code{%1}, etc.
@item
Specify the @code{Outputs} parameter as a parenthesized comma-separated list
of @code{Asm_Output} attributes
@end itemize

For example:
@smallexample @c ada
@group
Asm ("movl %%eax, %0" & LF & HT &
     "movl %%ebx, %1" & LF & HT &
     "movl %%ecx, %2",
     Outputs => (Unsigned_32'Asm_Output ("=g", Var_A),   --@i{  %0 = Var_A}
                 Unsigned_32'Asm_Output ("=g", Var_B),   --@i{  %1 = Var_B}
                 Unsigned_32'Asm_Output ("=g", Var_C))); --@i{  %2 = Var_C}
@end group
@end smallexample
@noindent
where @code{Var_A}, @code{Var_B}, and @code{Var_C} are variables
in the Ada program.

As a variation on the @code{Get_Flags} example, we can use the constraints
string to direct the compiler to store the eax register into the @code{Flags}
variable, instead of including the store instruction explicitly in the
@code{Asm} template string:

@smallexample @c ada
@group
@b{with} Interfaces; @b{use} Interfaces;
@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{with} System.Machine_Code; @b{use} System.Machine_Code;
@b{procedure} Get_Flags_2 @b{is}
   Flags : Unsigned_32;
   @b{use} ASCII;
@b{begin}
   Asm ("pushfl"      & LF & HT & --@i{ push flags on stack}
        "popl %%eax",             --@i{ save flags in eax}
        Outputs => Unsigned_32'Asm_Output ("=a", Flags));
   Put_Line ("Flags register:" & Flags'Img);
@b{end} Get_Flags_2;
@end group
@end smallexample

@noindent
The @code{"a"} constraint tells the compiler that the @code{Flags}
variable will come from the eax register. Here is the resulting code:

@smallexample
@group
#APP
   pushfl
   popl %eax
#NO_APP
   movl %eax,-40(%ebp)
@end group
@end smallexample

@noindent
The compiler generated the store of eax into Flags after
expanding the assembler code.

Actually, there was no need to pop the flags into the eax register;
more simply, we could just pop the flags directly into the program variable:

@smallexample @c ada
@group
@b{with} Interfaces; @b{use} Interfaces;
@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{with} System.Machine_Code; @b{use} System.Machine_Code;
@b{procedure} Get_Flags_3 @b{is}
   Flags : Unsigned_32;
   @b{use} ASCII;
@b{begin}
   Asm ("pushfl"  & LF & HT & --@i{ push flags on stack}
        "pop %0",             --@i{ save flags in Flags}
        Outputs => Unsigned_32'Asm_Output ("=g", Flags));
   Put_Line ("Flags register:" & Flags'Img);
@b{end} Get_Flags_3;
@end group
@end smallexample

@c ---------------------------------------------------------------------------
@node Input Variables in Inline Assembler
@section Input Variables in Inline Assembler

@noindent
The example in this section illustrates how to specify the source operands
for assembly language statements.
The program simply increments its input value by 1:

@smallexample @c ada
@group
@b{with} Interfaces; @b{use} Interfaces;
@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{with} System.Machine_Code; @b{use} System.Machine_Code;
@b{procedure} Increment @b{is}

   @b{function} Incr (Value : Unsigned_32) @b{return} Unsigned_32 @b{is}
      Result : Unsigned_32;
   @b{begin}
      Asm ("incl %0",
           Outputs => Unsigned_32'Asm_Output ("=a", Result),
           Inputs  => Unsigned_32'Asm_Input ("a", Value));
      @b{return} Result;
   @b{end} Incr;

   Value : Unsigned_32;

@b{begin}
   Value := 5;
   Put_Line ("Value before is" & Value'Img);
   Value := Incr (Value);
   Put_Line ("Value after is" & Value'Img);
@b{end} Increment;
@end group
@end smallexample

The @code{Outputs} parameter to @code{Asm} specifies
that the result will be in the eax register and that it is to be stored
in the @code{Result} variable.

The @code{Inputs} parameter looks much like the @code{Outputs} parameter,
but with an @code{Asm_Input} attribute.
The @code{"="} constraint, indicating an output value, is not present.

You can have multiple input variables, in the same way that you can have more
than one output variable.

The parameter count (%0, %1) etc, still starts at the first output statement,
and continues with the input statements.

Just as the @code{Outputs} parameter causes the register to be stored into the
target variable after execution of the assembler statements, so does the
@code{Inputs} parameter cause its variable to be loaded into the register
before execution of the assembler statements.

Thus the effect of the @code{Asm} invocation is:
@enumerate
@item load the 32-bit value of @code{Value} into eax
@item execute the @code{incl %eax} instruction
@item store the contents of eax into the @code{Result} variable
@end enumerate

The resulting assembler file (with @option{-O2} optimization) contains:
@smallexample
@group
_increment__incr.1:
   subl $4,%esp
   movl 8(%esp),%eax
#APP
   incl %eax
#NO_APP
   movl %eax,%edx
   movl %ecx,(%esp)
   addl $4,%esp
   ret
@end group
@end smallexample

@c ---------------------------------------------------------------------------
@node Inlining Inline Assembler Code
@section Inlining Inline Assembler Code

@noindent
For a short subprogram such as the @code{Incr} function in the previous
section, the overhead of the call and return (creating / deleting the stack
frame) can be significant, compared to the amount of code in the subprogram
body.  A solution is to apply Ada's @code{Inline} pragma to the subprogram,
which directs the compiler to expand invocations of the subprogram at the
point(s) of call, instead of setting up a stack frame for out-of-line calls.
Here is the resulting program:

@smallexample @c ada
@group
@b{with} Interfaces; @b{use} Interfaces;
@b{with} Ada.Text_IO; @b{use} Ada.Text_IO;
@b{with} System.Machine_Code; @b{use} System.Machine_Code;
@b{procedure} Increment_2 @b{is}

   @b{function} Incr (Value : Unsigned_32) @b{return} Unsigned_32 @b{is}
      Result : Unsigned_32;
   @b{begin}
      Asm ("incl %0",
           Outputs => Unsigned_32'Asm_Output ("=a", Result),
           Inputs  => Unsigned_32'Asm_Input ("a", Value));
      @b{return} Result;
   @b{end} Incr;
   @b{pragma} Inline (Increment);

   Value : Unsigned_32;

@b{begin}
   Value := 5;
   Put_Line ("Value before is" & Value'Img);
   Value := Increment (Value);
   Put_Line ("Value after is" & Value'Img);
@b{end} Increment_2;
@end group
@end smallexample

Compile the program with both optimization (@option{-O2}) and inlining
(@option{-gnatn}) enabled.

The @code{Incr} function is still compiled as usual, but at the
point in @code{Increment} where our function used to be called:

@smallexample
@group
pushl %edi
call _increment__incr.1
@end group
@end smallexample

@noindent
the code for the function body directly appears:

@smallexample
@group
movl %esi,%eax
#APP
   incl %eax
#NO_APP
   movl %eax,%edx
@end group
@end smallexample

@noindent
thus saving the overhead of stack frame setup and an out-of-line call.

@c ---------------------------------------------------------------------------
@node Other Asm Functionality
@section Other @code{Asm} Functionality

@noindent
This section describes two important parameters to the @code{Asm}
procedure: @code{Clobber}, which identifies register usage;
and @code{Volatile}, which inhibits unwanted optimizations.

@menu
* The Clobber Parameter::
* The Volatile Parameter::
@end menu

@c ---------------------------------------------------------------------------
@node The Clobber Parameter
@subsection The @code{Clobber} Parameter

@noindent
One of the dangers of intermixing assembly language and a compiled language
such as Ada is that the compiler needs to be aware of which registers are
being used by the assembly code.  In some cases, such as the earlier examples,
the constraint string is sufficient to indicate register usage (e.g.,
@code{"a"} for
the eax register).  But more generally, the compiler needs an explicit
identification of the registers that are used by the Inline Assembly
statements.

Using a register that the compiler doesn't know about
could be a side effect of an instruction (like @code{mull}
storing its result in both eax and edx).
It can also arise from explicit register usage in your
assembly code; for example:
@smallexample
@group
Asm ("movl %0, %%ebx" & LF & HT &
     "movl %%ebx, %1",
     Outputs => Unsigned_32'Asm_Output ("=g", Var_Out),
     Inputs  => Unsigned_32'Asm_Input  ("g", Var_In));
@end group
@end smallexample
@noindent
where the compiler (since it does not analyze the @code{Asm} template string)
does not know you are using the ebx register.

In such cases you need to supply the @code{Clobber} parameter to @code{Asm},
to identify the registers that will be used by your assembly code:

@smallexample
@group
Asm ("movl %0, %%ebx" & LF & HT &
     "movl %%ebx, %1",
     Outputs => Unsigned_32'Asm_Output ("=g", Var_Out),
     Inputs  => Unsigned_32'Asm_Input  ("g", Var_In),
     Clobber => "ebx");
@end group
@end smallexample

The Clobber parameter is a static string expression specifying the
register(s) you are using.  Note that register names are @emph{not} prefixed
by a percent sign. Also, if more than one register is used then their names
are separated by commas; e.g., @code{"eax, ebx"}

The @code{Clobber} parameter has several additional uses:
@enumerate
@item Use ``register'' name @code{cc} to indicate that flags might have changed
@item Use ``register'' name @code{memory} if you changed a memory location
@end enumerate

@c ---------------------------------------------------------------------------
@node The Volatile Parameter
@subsection The @code{Volatile} Parameter
@cindex Volatile parameter

@noindent
Compiler optimizations in the presence of Inline Assembler may sometimes have
unwanted effects.  For example, when an @code{Asm} invocation with an input
variable is inside a loop, the compiler might move the loading of the input
variable outside the loop, regarding it as a one-time initialization.

If this effect is not desired, you can disable such optimizations by setting
the @code{Volatile} parameter to @code{True}; for example:

@smallexample @c ada
@group
Asm ("movl %0, %%ebx" & LF & HT &
     "movl %%ebx, %1",
     Outputs  => Unsigned_32'Asm_Output ("=g", Var_Out),
     Inputs   => Unsigned_32'Asm_Input  ("g", Var_In),
     Clobber  => "ebx",
     Volatile => True);
@end group
@end smallexample

By default, @code{Volatile} is set to @code{False} unless there is no
@code{Outputs} parameter.

Although setting @code{Volatile} to @code{True} prevents unwanted
optimizations, it will also disable other optimizations that might be
important for efficiency. In general, you should set @code{Volatile}
to @code{True} only if the compiler's optimizations have created
problems.
@c END OF INLINE ASSEMBLER CHAPTER
@c ===============================


@c *****************************************
@c Writing Portable Fixed-Point Declarations
@c *****************************************
@node Writing Portable Fixed-Point Declarations
@appendix Writing Portable Fixed-Point Declarations
@cindex Fixed-point types (writing portable declarations)

@noindent
The Ada Reference Manual gives an implementation freedom to choose bounds
that are narrower by @code{Small} from the given bounds.
For example, if we write

@smallexample @c ada
   type F1 is delta 1.0 range -128.0 .. +128.0;
@end smallexample

@noindent
then the implementation is allowed to choose -128.0 .. +127.0 if it
likes, but is not required to do so.

This leads to possible portability problems, so let's have a closer
look at this, and figure out how to avoid these problems.

First, why does this freedom exist, and why would an implementation
take advantage of it? To answer this, take a closer look at the type
declaration for @code{F1} above. If the compiler uses the given bounds,
it would need 9 bits to hold the largest positive value (and typically
that means 16 bits on all machines). But if the implementation chooses
the +127.0 bound then it can fit values of the type in 8 bits.

Why not make the user write +127.0 if that's what is wanted?
The rationale is that if you are thinking of fixed point
as a kind of ``poor man's floating-point'', then you don't want
to be thinking about the scaled integers that are used in its
representation. Let's take another example:

@smallexample @c ada
   type F2 is delta 2.0**(-15) range -1.0 .. +1.0;
@end smallexample

@noindent
Looking at this declaration, it seems casually as though
it should fit in 16 bits, but again that extra positive value
+1.0 has the scaled integer equivalent of 2**15 which is one too
big for signed 16 bits. The implementation can treat this as:

@smallexample @c ada
   type F2 is delta 2.0**(-15) range -1.0 .. +1.0-(2.0**(-15));
@end smallexample

@noindent
and the Ada language design team felt that this was too annoying
to require. We don't need to debate this decision at this point,
since it is well established (the rule about narrowing the ranges
dates to Ada 83).

But the important point is that an implementation is not required
to do this narrowing, so we have a potential portability problem.
We could imagine three types of implementation:

@enumerate a
@item
those that narrow the range automatically if they can figure
out that the narrower range will allow storage in a smaller machine unit,

@item
those that will narrow only if forced to by a @code{'Size} clause, and

@item
those that will never narrow.
@end enumerate

@noindent
Now if we are language theoreticians, we can imagine a fourth
approach: is to narrow all the time, e.g. to treat

@smallexample @c ada
   type F3 is delta 1.0 range -10.0 .. +23.0;
@end smallexample

@noindent
as though it had been written:

@smallexample @c ada
   type F3 is delta 1.0 range -9.0 .. +22.0;
@end smallexample

@noindent
But although technically allowed, such a behavior would be hostile and silly,
and no real compiler would do this. All real compilers will fall into one of
the categories (a), (b) or (c) above.

So, how do you get the compiler to do what you want? The answer is give the
actual bounds you want, and then use a @code{'Small} clause and a
@code{'Size} clause to absolutely pin down what the compiler does.
E.g., for @code{F2} above, we will write:

@smallexample @c ada
@group
   My_Small : constant := 2.0**(-15);
   My_First : constant := -1.0;
   My_Last  : constant := +1.0 - My_Small;

   type F2 is delta My_Small range My_First .. My_Last;
@end group
@end smallexample

@noindent
and then add

@smallexample @c ada
@group
   for F2'Small use my_Small;
   for F2'Size  use 16;
@end group
@end smallexample

@noindent
In practice all compilers will do the same thing here and will give you
what you want, so the above declarations are fully portable. If you really
want to play language lawyer and guard against ludicrous behavior by the
compiler you could add

@smallexample @c ada
@group
   Test1 : constant := 1 / Boolean'Pos (F2'First = My_First);
   Test2 : constant := 1 / Boolean'Pos (F2'Last  = My_Last);
@end group
@end smallexample

@noindent
One or other or both are allowed to be illegal if the compiler is
behaving in a silly manner, but at least the silly compiler will not
get away with silently messing with your (very clear) intentions.

If you follow this scheme you will be guaranteed that your fixed-point
types will be portable.


@c ***********************************
@c * Compatibility and Porting Guide *
@c ***********************************
@node Compatibility and Porting Guide
@appendix Compatibility and Porting Guide

@noindent
This chapter describes the compatibility issues that may arise between
GNAT and other Ada compilation systems (including those for Ada 83),
and shows how GNAT can expedite porting
applications developed in other Ada environments.

@menu
* Compatibility with Ada 83::
* Compatibility between Ada 95 and Ada 2005::
* Implementation-dependent characteristics::
* Compatibility with Other Ada Systems::
* Representation Clauses::
@c Brief section is only in non-VMS version
@c Full chapter is in VMS version
* Compatibility with HP Ada 83::
@end menu

@node Compatibility with Ada 83
@section Compatibility with Ada 83
@cindex Compatibility (between Ada 83 and Ada 95 / Ada 2005)

@noindent
Ada 95 and Ada 2005 are highly upwards compatible with Ada 83.  In
particular, the design intention was that the difficulties associated
with moving from Ada 83 to Ada 95 or Ada 2005 should be no greater than those
that occur when moving from one Ada 83 system to another.

However, there are a number of points at which there are minor
incompatibilities.  The @cite{Ada 95 Annotated Reference Manual} contains
full details of these issues,
and should be consulted for a complete treatment.
In practice the
following subsections treat the most likely issues to be encountered.

@menu
* Legal Ada 83 programs that are illegal in Ada 95::
* More deterministic semantics::
* Changed semantics::
* Other language compatibility issues::
@end menu

@node Legal Ada 83 programs that are illegal in Ada 95
@subsection Legal Ada 83 programs that are illegal in Ada 95

Some legal Ada 83 programs are illegal (i.e., they will fail to compile) in
Ada 95 and thus also in Ada 2005:

@table @emph
@item Character literals
Some uses of character literals are ambiguous.  Since Ada 95 has introduced
@code{Wide_Character} as a new predefined character type, some uses of
character literals that were legal in Ada 83 are illegal in Ada 95.
For example:
@smallexample @c ada
   @b{for} Char @b{in} 'A' .. 'Z' @b{loop} @dots{} @b{end} @b{loop};
@end smallexample

@noindent
The problem is that @code{'A'} and @code{'Z'} could be from either
@code{Character} or @code{Wide_Character}.  The simplest correction
is to make the type explicit; e.g.:
@smallexample @c ada
   @b{for} Char @b{in} Character @b{range} 'A' .. 'Z' @b{loop} @dots{} @b{end} @b{loop};
@end smallexample

@item New reserved words
The identifiers @code{abstract}, @code{aliased}, @code{protected},
@code{requeue}, @code{tagged}, and @code{until} are reserved in Ada 95.
Existing Ada 83 code using any of these identifiers must be edited to
use some alternative name.

@item Freezing rules
The rules in Ada 95 are slightly different with regard to the point at
which entities are frozen, and representation pragmas and clauses are
not permitted past the freeze point.  This shows up most typically in
the form of an error message complaining that a representation item
appears too late, and the appropriate corrective action is to move
the item nearer to the declaration of the entity to which it refers.

A particular case is that representation pragmas
cannot be applied to a subprogram body.  If necessary, a separate subprogram
declaration must be introduced to which the pragma can be applied.

@item Optional bodies for library packages
In Ada 83, a package that did not require a package body was nevertheless
allowed to have one.  This lead to certain surprises in compiling large
systems (situations in which the body could be unexpectedly ignored by the
binder).  In Ada 95, if a package does not require a body then it is not
permitted to have a body.  To fix this problem, simply remove a redundant
body if it is empty, or, if it is non-empty, introduce a dummy declaration
into the spec that makes the body required.  One approach is to add a private
part to the package declaration (if necessary), and define a parameterless
procedure called @code{Requires_Body}, which must then be given a dummy
procedure body in the package body, which then becomes required.
Another approach (assuming that this does not introduce elaboration
circularities) is to add an @code{Elaborate_Body} pragma to the package spec,
since one effect of this pragma is to require the presence of a package body.

@item @code{Numeric_Error} is now the same as @code{Constraint_Error}
In Ada 95, the exception @code{Numeric_Error} is a renaming of
@code{Constraint_Error}.
This means that it is illegal to have separate exception handlers for
the two exceptions.  The fix is simply to remove the handler for the
@code{Numeric_Error} case (since even in Ada 83, a compiler was free to raise
@code{Constraint_Error} in place of @code{Numeric_Error} in all cases).

@item Indefinite subtypes in generics
In Ada 83, it was permissible to pass an indefinite type (e.g.@: @code{String})
as the actual for a generic formal private type, but then the instantiation
would be illegal if there were any instances of declarations of variables
of this type in the generic body.  In Ada 95, to avoid this clear violation
of the methodological principle known as the ``contract model'',
the generic declaration explicitly indicates whether
or not such instantiations are permitted.  If a generic formal parameter
has explicit unknown discriminants, indicated by using @code{(<>)} after the
subtype name, then it can be instantiated with indefinite types, but no
stand-alone variables can be declared of this type.  Any attempt to declare
such a variable will result in an illegality at the time the generic is
declared.  If the @code{(<>)} notation is not used, then it is illegal
to instantiate the generic with an indefinite type.
This is the potential incompatibility issue when porting Ada 83 code to Ada 95.
It will show up as a compile time error, and
the fix is usually simply to add the @code{(<>)} to the generic declaration.
@end table

@node More deterministic semantics
@subsection More deterministic semantics

@table @emph
@item Conversions
Conversions from real types to integer types round away from 0.  In Ada 83
the conversion Integer(2.5) could deliver either 2 or 3 as its value.  This
implementation freedom was intended to support unbiased rounding in
statistical applications, but in practice it interfered with portability.
In Ada 95 the conversion semantics are unambiguous, and rounding away from 0
is required.  Numeric code may be affected by this change in semantics.
Note, though, that this issue is no worse than already existed in Ada 83
when porting code from one vendor to another.

@item Tasking
The Real-Time Annex introduces a set of policies that define the behavior of
features that were implementation dependent in Ada 83, such as the order in
which open select branches are executed.
@end table

@node Changed semantics
@subsection Changed semantics

@noindent
The worst kind of incompatibility is one where a program that is legal in
Ada 83 is also legal in Ada 95 but can have an effect in Ada 95 that was not
possible in Ada 83.  Fortunately this is extremely rare, but the one
situation that you should be alert to is the change in the predefined type
@code{Character} from 7-bit ASCII to 8-bit Latin-1.

@table @emph
@item Range of type @code{Character}
The range of @code{Standard.Character} is now the full 256 characters
of Latin-1, whereas in most Ada 83 implementations it was restricted
to 128 characters. Although some of the effects of
this change will be manifest in compile-time rejection of legal
Ada 83 programs it is possible for a working Ada 83 program to have
a different effect in Ada 95, one that was not permitted in Ada 83.
As an example, the expression
@code{Character'Pos(Character'Last)} returned @code{127} in Ada 83 and now
delivers @code{255} as its value.
In general, you should look at the logic of any
character-processing Ada 83 program and see whether it needs to be adapted
to work correctly with Latin-1.  Note that the predefined Ada 95 API has a
character handling package that may be relevant if code needs to be adapted
to account for the additional Latin-1 elements.
The desirable fix is to
modify the program to accommodate the full character set, but in some cases
it may be convenient to define a subtype or derived type of Character that
covers only the restricted range.
@cindex Latin-1
@end table

@node Other language compatibility issues
@subsection Other language compatibility issues

@table @emph
@item @option{-gnat83} switch
All implementations of GNAT provide a switch that causes GNAT to operate
in Ada 83 mode.  In this mode, some but not all compatibility problems
of the type described above are handled automatically.  For example, the
new reserved words introduced in Ada 95 and Ada 2005 are treated simply
as identifiers as in Ada 83.
However,
in practice, it is usually advisable to make the necessary modifications
to the program to remove the need for using this switch.
See @ref{Compiling Different Versions of Ada}.

@item Support for removed Ada 83 pragmas and attributes
A number of pragmas and attributes from Ada 83 were removed from Ada 95,
generally because they were replaced by other mechanisms.  Ada 95 and Ada 2005
compilers are allowed, but not required, to implement these missing
elements.  In contrast with some other compilers, GNAT implements all
such pragmas and attributes, eliminating this compatibility concern.  These
include @code{pragma Interface} and the floating point type attributes
(@code{Emax}, @code{Mantissa}, etc.), among other items.
@end table


@node Compatibility between Ada 95 and Ada 2005
@section Compatibility between Ada 95 and Ada 2005
@cindex Compatibility between Ada 95 and Ada 2005

@noindent
Although Ada 2005 was designed to be upwards compatible with Ada 95, there are
a number of incompatibilities. Several are enumerated below;
for a complete description please see the
Annotated Ada 2005 Reference Manual, or section 9.1.1 in
@cite{Rationale for Ada 2005}.

@table @emph
@item New reserved words.
The words @code{interface}, @code{overriding} and @code{synchronized} are
reserved in Ada 2005.
A pre-Ada 2005 program that uses any of these as an identifier will be
illegal.

@item New declarations in predefined packages.
A number of packages in the predefined environment contain new declarations:
@code{Ada.Exceptions}, @code{Ada.Real_Time}, @code{Ada.Strings},
@code{Ada.Strings.Fixed}, @code{Ada.Strings.Bounded},
@code{Ada.Strings.Unbounded}, @code{Ada.Strings.Wide_Fixed},
@code{Ada.Strings.Wide_Bounded}, @code{Ada.Strings.Wide_Unbounded},
@code{Ada.Tags}, @code{Ada.Text_IO}, and @code{Interfaces.C}.
If an Ada 95 program does a @code{with} and @code{use} of any of these
packages, the new declarations may cause name clashes.

@item Access parameters.
A nondispatching subprogram with an access parameter cannot be renamed
as a dispatching operation.  This was permitted in Ada 95.

@item Access types, discriminants, and constraints.
Rule changes in this area have led to some incompatibilities; for example,
constrained subtypes of some access types are not permitted in Ada 2005.

@item Aggregates for limited types.
The allowance of aggregates for limited types in Ada 2005 raises the
possibility of ambiguities in legal Ada 95 programs, since additional types
now need to be considered in expression resolution.

@item Fixed-point multiplication and division.
Certain expressions involving ``*'' or ``/'' for a fixed-point type, which
were legal in Ada 95 and invoked the predefined versions of these operations,
are now ambiguous.
The ambiguity may be resolved either by applying a type conversion to the
expression, or by explicitly invoking the operation from package
@code{Standard}.

@item Return-by-reference types.
The Ada 95 return-by-reference mechanism has been removed.  Instead, the user
can declare a function returning a value from an anonymous access type.
@end table


@node Implementation-dependent characteristics
@section Implementation-dependent characteristics
@noindent
Although the Ada language defines the semantics of each construct as
precisely as practical, in some situations (for example for reasons of
efficiency, or where the effect is heavily dependent on the host or target
platform) the implementation is allowed some freedom.  In porting Ada 83
code to GNAT, you need to be aware of whether / how the existing code
exercised such implementation dependencies.  Such characteristics fall into
several categories, and GNAT offers specific support in assisting the
transition from certain Ada 83 compilers.

@menu
* Implementation-defined pragmas::
* Implementation-defined attributes::
* Libraries::
* Elaboration order::
* Target-specific aspects::
@end menu

@node Implementation-defined pragmas
@subsection Implementation-defined pragmas

@noindent
Ada compilers are allowed to supplement the language-defined pragmas, and
these are a potential source of non-portability.  All GNAT-defined pragmas
are described in @ref{Implementation Defined Pragmas,,, gnat_rm, GNAT
Reference Manual}, and these include several that are specifically
intended to correspond to other vendors' Ada 83 pragmas.
For migrating from VADS, the pragma @code{Use_VADS_Size} may be useful.
For compatibility with HP Ada 83, GNAT supplies the pragmas
@code{Extend_System}, @code{Ident}, @code{Inline_Generic},
@code{Interface_Name}, @code{Passive}, @code{Suppress_All},
and @code{Volatile}.
Other relevant pragmas include @code{External} and @code{Link_With}.
Some vendor-specific
Ada 83 pragmas (@code{Share_Generic}, @code{Subtitle}, and @code{Title}) are
recognized, thus
avoiding compiler rejection of units that contain such pragmas; they are not
relevant in a GNAT context and hence are not otherwise implemented.

@node Implementation-defined attributes
@subsection Implementation-defined attributes

Analogous to pragmas, the set of attributes may be extended by an
implementation.  All GNAT-defined attributes are described in
@ref{Implementation Defined Attributes,,, gnat_rm, GNAT Reference
Manual}, and these include several that are specifically intended
to correspond to other vendors' Ada 83 attributes.  For migrating from VADS,
the attribute @code{VADS_Size} may be useful.  For compatibility with HP
Ada 83, GNAT supplies the attributes @code{Bit}, @code{Machine_Size} and
@code{Type_Class}.

@node Libraries
@subsection Libraries
@noindent
Vendors may supply libraries to supplement the standard Ada API.  If Ada 83
code uses vendor-specific libraries then there are several ways to manage
this in Ada 95 or Ada 2005:
@enumerate
@item
If the source code for the libraries (specs and bodies) are
available, then the libraries can be migrated in the same way as the
application.
@item
If the source code for the specs but not the bodies are
available, then you can reimplement the bodies.
@item
Some features introduced by Ada 95 obviate the need for library support.  For
example most Ada 83 vendors supplied a package for unsigned integers.  The
Ada 95 modular type feature is the preferred way to handle this need, so
instead of migrating or reimplementing the unsigned integer package it may
be preferable to retrofit the application using modular types.
@end enumerate

@node Elaboration order
@subsection Elaboration order
@noindent
The implementation can choose any elaboration order consistent with the unit
dependency relationship.  This freedom means that some orders can result in
Program_Error being raised due to an ``Access Before Elaboration'': an attempt
to invoke a subprogram its body has been elaborated, or to instantiate a
generic before the generic body has been elaborated.  By default GNAT
attempts to choose a safe order (one that will not encounter access before
elaboration problems) by implicitly inserting @code{Elaborate} or
@code{Elaborate_All} pragmas where
needed.  However, this can lead to the creation of elaboration circularities
and a resulting rejection of the program by gnatbind.  This issue is
thoroughly described in @ref{Elaboration Order Handling in GNAT}.
In brief, there are several
ways to deal with this situation:

@itemize @bullet
@item
Modify the program to eliminate the circularities, e.g.@: by moving
elaboration-time code into explicitly-invoked procedures
@item
Constrain the elaboration order by including explicit @code{Elaborate_Body} or
@code{Elaborate} pragmas, and then inhibit the generation of implicit
@code{Elaborate_All}
pragmas either globally (as an effect of the @option{-gnatE} switch) or locally
(by selectively suppressing elaboration checks via pragma
@code{Suppress(Elaboration_Check)} when it is safe to do so).
@end itemize

@node Target-specific aspects
@subsection Target-specific aspects
@noindent
Low-level applications need to deal with machine addresses, data
representations, interfacing with assembler code, and similar issues.  If
such an Ada 83 application is being ported to different target hardware (for
example where the byte endianness has changed) then you will need to
carefully examine the program logic; the porting effort will heavily depend
on the robustness of the original design.  Moreover, Ada 95 (and thus
Ada 2005) are sometimes
incompatible with typical Ada 83 compiler practices regarding implicit
packing, the meaning of the Size attribute, and the size of access values.
GNAT's approach to these issues is described in @ref{Representation Clauses}.

@node Compatibility with Other Ada Systems
@section Compatibility with Other Ada Systems

@noindent
If programs avoid the use of implementation dependent and
implementation defined features, as documented in the @cite{Ada
Reference Manual}, there should be a high degree of portability between
GNAT and other Ada systems.  The following are specific items which
have proved troublesome in moving Ada 95 programs from GNAT to other Ada 95
compilers, but do not affect porting code to GNAT@.
(As of @value{NOW}, GNAT is the only compiler available for Ada 2005;
the following issues may or may not arise for Ada 2005 programs
when other compilers appear.)

@table @emph
@item Ada 83 Pragmas and Attributes
Ada 95 compilers are allowed, but not required, to implement the missing
Ada 83 pragmas and attributes that are no longer defined in Ada 95.
GNAT implements all such pragmas and attributes, eliminating this as
a compatibility concern, but some other Ada 95 compilers reject these
pragmas and attributes.

@item Specialized Needs Annexes
GNAT implements the full set of special needs annexes.  At the
current time, it is the only Ada 95 compiler to do so.  This means that
programs making use of these features may not be portable to other Ada
95 compilation systems.

@item Representation Clauses
Some other Ada 95 compilers implement only the minimal set of
representation clauses required by the Ada 95 reference manual.  GNAT goes
far beyond this minimal set, as described in the next section.
@end table

@node Representation Clauses
@section Representation Clauses

@noindent
The Ada 83 reference manual was quite vague in describing both the minimal
required implementation of representation clauses, and also their precise
effects.  Ada 95 (and thus also Ada 2005) are much more explicit, but the
minimal set of capabilities required is still quite limited.

GNAT implements the full required set of capabilities in
Ada 95 and Ada 2005, but also goes much further, and in particular
an effort has been made to be compatible with existing Ada 83 usage to the
greatest extent possible.

A few cases exist in which Ada 83 compiler behavior is incompatible with
the requirements in Ada 95 (and thus also Ada 2005).  These are instances of
intentional or accidental dependence on specific implementation dependent
characteristics of these Ada 83 compilers.  The following is a list of
the cases most likely to arise in existing Ada 83 code.

@table @emph
@item Implicit Packing
Some Ada 83 compilers allowed a Size specification to cause implicit
packing of an array or record.  This could cause expensive implicit
conversions for change of representation in the presence of derived
types, and the Ada design intends to avoid this possibility.
Subsequent AI's were issued to make it clear that such implicit
change of representation in response to a Size clause is inadvisable,
and this recommendation is represented explicitly in the Ada 95 (and Ada 2005)
Reference Manuals as implementation advice that is followed by GNAT@.
The problem will show up as an error
message rejecting the size clause.  The fix is simply to provide
the explicit pragma @code{Pack}, or for more fine tuned control, provide
a Component_Size clause.

@item Meaning of Size Attribute
The Size attribute in Ada 95 (and Ada 2005) for discrete types is defined as
the minimal number of bits required to hold values of the type.  For example,
on a 32-bit machine, the size of @code{Natural} will typically be 31 and not
32 (since no sign bit is required).  Some Ada 83 compilers gave 31, and
some 32 in this situation.  This problem will usually show up as a compile
time error, but not always.  It is a good idea to check all uses of the
'Size attribute when porting Ada 83 code.  The GNAT specific attribute
Object_Size can provide a useful way of duplicating the behavior of
some Ada 83 compiler systems.

@item Size of Access Types
A common assumption in Ada 83 code is that an access type is in fact a pointer,
and that therefore it will be the same size as a System.Address value.  This
assumption is true for GNAT in most cases with one exception.  For the case of
a pointer to an unconstrained array type (where the bounds may vary from one
value of the access type to another), the default is to use a ``fat pointer'',
which is represented as two separate pointers, one to the bounds, and one to
the array.  This representation has a number of advantages, including improved
efficiency.  However, it may cause some difficulties in porting existing Ada 83
code which makes the assumption that, for example, pointers fit in 32 bits on
a machine with 32-bit addressing.

To get around this problem, GNAT also permits the use of ``thin pointers'' for
access types in this case (where the designated type is an unconstrained array
type).  These thin pointers are indeed the same size as a System.Address value.
To specify a thin pointer, use a size clause for the type, for example:

@smallexample @c ada
@b{type} X @b{is} @b{access} @b{all} String;
@b{for} X'Size @b{use} Standard'Address_Size;
@end smallexample

@noindent
which will cause the type X to be represented using a single pointer.
When using this representation, the bounds are right behind the array.
This representation is slightly less efficient, and does not allow quite
such flexibility in the use of foreign pointers or in using the
Unrestricted_Access attribute to create pointers to non-aliased objects.
But for any standard portable use of the access type it will work in
a functionally correct manner and allow porting of existing code.
Note that another way of forcing a thin pointer representation
is to use a component size clause for the element size in an array,
or a record representation clause for an access field in a record.

See the documentation of Unrestricted_Access in the GNAT RM for a
full discussion of possible problems using this attribute in conjunction
with thin pointers.
@end table

@c This brief section is only in the non-VMS version
@c The complete chapter on HP Ada is in the VMS version
@node Compatibility with HP Ada 83
@section Compatibility with HP Ada 83

@noindent
The VMS version of GNAT fully implements all the pragmas and attributes
provided by HP Ada 83, as well as providing the standard HP Ada 83
libraries, including Starlet.  In addition, data layouts and parameter
passing conventions are highly compatible.  This means that porting
existing HP Ada 83 code to GNAT in VMS systems should be easier than
most other porting efforts.  The following are some of the most
significant differences between GNAT and HP Ada 83.

@table @emph
@item Default floating-point representation
In GNAT, the default floating-point format is IEEE, whereas in HP Ada 83,
it is VMS format.  GNAT does implement the necessary pragmas
(Long_Float, Float_Representation) for changing this default.

@item System
The package System in GNAT exactly corresponds to the definition in the
Ada 95 reference manual, which means that it excludes many of the
HP Ada 83 extensions.  However, a separate package Aux_DEC is provided
that contains the additional definitions, and a special pragma,
Extend_System allows this package to be treated transparently as an
extension of package System.

@item To_Address
The definitions provided by Aux_DEC are exactly compatible with those
in the HP Ada 83 version of System, with one exception.
HP Ada provides the following declarations:

@smallexample @c ada
TO_ADDRESS (INTEGER)
TO_ADDRESS (UNSIGNED_LONGWORD)
TO_ADDRESS (@i{universal_integer})
@end smallexample

@noindent
The version of TO_ADDRESS taking a @i{universal integer} argument is in fact
an extension to Ada 83 not strictly compatible with the reference manual.
In GNAT, we are constrained to be exactly compatible with the standard,
and this means we cannot provide this capability.  In HP Ada 83, the
point of this definition is to deal with a call like:

@smallexample @c ada
TO_ADDRESS (16#12777#);
@end smallexample

@noindent
Normally, according to the Ada 83 standard, one would expect this to be
ambiguous, since it matches both the INTEGER and UNSIGNED_LONGWORD forms
of TO_ADDRESS@.  However, in HP Ada 83, there is no ambiguity, since the
definition using @i{universal_integer} takes precedence.

In GNAT, since the version with @i{universal_integer} cannot be supplied, it
is not possible to be 100% compatible.  Since there are many programs using
numeric constants for the argument to TO_ADDRESS, the decision in GNAT was
to change the name of the function in the UNSIGNED_LONGWORD case, so the
declarations provided in the GNAT version of AUX_Dec are:

@smallexample @c ada
@b{function} To_Address (X : Integer) @b{return} Address;
@b{pragma} Pure_Function (To_Address);

@b{function} To_Address_Long (X : Unsigned_Longword)
 @b{return} Address;
@b{pragma} Pure_Function (To_Address_Long);
@end smallexample

@noindent
This means that programs using TO_ADDRESS for UNSIGNED_LONGWORD must
change the name to TO_ADDRESS_LONG@.

@item Task_Id values
The Task_Id values assigned will be different in the two systems, and GNAT
does not provide a specified value for the Task_Id of the environment task,
which in GNAT is treated like any other declared task.
@end table

@noindent
For full details on these and other less significant compatibility issues,
see appendix E of the HP publication entitled @cite{HP Ada, Technical
Overview and Comparison on HP Platforms}.

For GNAT running on other than VMS systems, all the HP Ada 83 pragmas and
attributes are recognized, although only a subset of them can sensibly
be implemented.  The description of pragmas in @ref{Implementation
Defined Pragmas,,, gnat_rm, GNAT Reference Manual}
indicates whether or not they are applicable to non-VMS systems.


@c ************************************************
@node Microsoft Windows Topics
@appendix Microsoft Windows Topics
@cindex Windows NT
@cindex Windows 95
@cindex Windows 98

@noindent
This chapter describes topics that are specific to the Microsoft Windows
platforms (NT, 2000, and XP Professional).

@menu
@ifclear FSFEDITION
* Installing from the Command Line::
@end ifclear
* Using GNAT on Windows::
* Using a network installation of GNAT::
* CONSOLE and WINDOWS subsystems::
* Temporary Files::
* Mixed-Language Programming on Windows::
* Windows Calling Conventions::
* Introduction to Dynamic Link Libraries (DLLs)::
* Using DLLs with GNAT::
* Building DLLs with GNAT Project files::
* Building DLLs with GNAT::
* Building DLLs with gnatdll::
* GNAT and Windows Resources::
* Debugging a DLL::
* Setting Stack Size from gnatlink::
* Setting Heap Size from gnatlink::
@end menu

@ifclear FSFEDITION
@node Installing from the Command Line
@section Installing from the Command Line
@cindex Batch installation
@cindex Silent installation
@cindex Unassisted installation

@noindent
By default the @value{EDITION} installers display a GUI that prompts the user
to enter installation path and similar information, and guide him through the
installation process. It is also possible to perform silent installations
using the command-line interface.

In order to install one of the @value{EDITION} installers from the command
line you should pass parameter @code{/S} (and, optionally,
@code{/D=<directory>}) as command-line arguments.

@ifset PROEDITION
For example, for an unattended installation of
@value{EDITION} 7.0.2 into the default directory
@code{C:\GNATPRO\7.0.2} you would run:

@smallexample
gnatpro-7.0.2-i686-pc-mingw32-bin.exe /S
@end smallexample

To install into a custom directory, say, @code{C:\TOOLS\GNATPRO\7.0.2}:

@smallexample
gnatpro-7.0.2-i686-pc-mingw32-bin /S /D=C:\TOOLS\GNATPRO\7.0.2
@end smallexample
@end ifset

@ifset GPLEDITION
For example, for an unattended installation of
@value{EDITION} 2012 into @code{C:\GNAT\2012}:

@smallexample
gnat-gpl-2012-i686-pc-mingw32-bin /S /D=C:\GNAT\2012
@end smallexample
@end ifset

You can use the same syntax for all installers.

Note that unattended installations don't modify system path, nor create file
associations, so such activities need to be done by hand.
@end ifclear

@node Using GNAT on Windows
@section Using GNAT on Windows

@noindent
One of the strengths of the GNAT technology is that its tool set
(@command{gcc}, @command{gnatbind}, @command{gnatlink}, @command{gnatmake}, the
@code{gdb} debugger, etc.) is used in the same way regardless of the
platform.

On Windows this tool set is complemented by a number of Microsoft-specific
tools that have been provided to facilitate interoperability with Windows
when this is required. With these tools:

@itemize @bullet

@item
You can build applications using the @code{CONSOLE} or @code{WINDOWS}
subsystems.

@item
You can use any Dynamically Linked Library (DLL) in your Ada code (both
relocatable and non-relocatable DLLs are supported).

@item
You can build Ada DLLs for use in other applications. These applications
can be written in a language other than Ada (e.g., C, C++, etc). Again both
relocatable and non-relocatable Ada DLLs are supported.

@item
You can include Windows resources in your Ada application.

@item
You can use or create COM/DCOM objects.
@end itemize

@noindent
Immediately below are listed all known general GNAT-for-Windows restrictions.
Other restrictions about specific features like Windows Resources and DLLs
are listed in separate sections below.

@itemize @bullet

@item
It is not possible to use @code{GetLastError} and @code{SetLastError}
when tasking, protected records, or exceptions are used. In these
cases, in order to implement Ada semantics, the GNAT run-time system
calls certain Win32 routines that set the last error variable to 0 upon
success. It should be possible to use @code{GetLastError} and
@code{SetLastError} when tasking, protected record, and exception
features are not used, but it is not guaranteed to work.

@item
It is not possible to link against Microsoft C++ libraries except for
import libraries. Interfacing must be done by the mean of DLLs.

@item
It is possible to link against Microsoft C libraries. Yet the preferred
solution is to use C/C++ compiler that comes with @value{EDITION}, since it
doesn't require having two different development environments and makes the
inter-language debugging experience smoother.

@item
When the compilation environment is located on FAT32 drives, users may
experience recompilations of the source files that have not changed if
Daylight Saving Time (DST) state has changed since the last time files
were compiled. NTFS drives do not have this problem.

@item
No components of the GNAT toolset use any entries in the Windows
registry. The only entries that can be created are file associations and
PATH settings, provided the user has chosen to create them at installation
time, as well as some minimal book-keeping information needed to correctly
uninstall or integrate different GNAT products.
@end itemize

@node Using a network installation of GNAT
@section Using a network installation of GNAT

@noindent
Make sure the system on which GNAT is installed is accessible from the
current machine, i.e., the install location is shared over the network.
Shared resources are accessed on Windows by means of UNC paths, which
have the format @code{\\server\sharename\path}

In order to use such a network installation, simply add the UNC path of the
@file{bin} directory of your GNAT installation in front of your PATH. For
example, if GNAT is installed in @file{\GNAT} directory of a share location
called @file{c-drive} on a machine @file{LOKI}, the following command will
make it available:

@code{@ @ @ path \\loki\c-drive\gnat\bin;%path%}

Be aware that every compilation using the network installation results in the
transfer of large amounts of data across the network and will likely cause
serious performance penalty.

@node CONSOLE and WINDOWS subsystems
@section CONSOLE and WINDOWS subsystems
@cindex CONSOLE Subsystem
@cindex WINDOWS Subsystem
@cindex -mwindows

@noindent
There are two main subsystems under Windows. The @code{CONSOLE} subsystem
(which is the default subsystem) will always create a console when
launching the application. This is not something desirable when the
application has a Windows GUI. To get rid of this console the
application must be using the @code{WINDOWS} subsystem. To do so
the @option{-mwindows} linker option must be specified.

@smallexample
$ gnatmake winprog -largs -mwindows
@end smallexample

@node Temporary Files
@section Temporary Files
@cindex Temporary files

@noindent
It is possible to control where temporary files gets created by setting
the @env{TMP} environment variable. The file will be created:

@itemize
@item Under the directory pointed to by the @env{TMP} environment variable if
this directory exists.

@item Under @file{c:\temp}, if the @env{TMP} environment variable is not
set (or not pointing to a directory) and if this directory exists.

@item Under the current working directory otherwise.
@end itemize

@noindent
This allows you to determine exactly where the temporary
file will be created. This is particularly useful in networked
environments where you may not have write access to some
directories.

@node Mixed-Language Programming on Windows
@section Mixed-Language Programming on Windows

@noindent
Developing pure Ada applications on Windows is no different than on
other GNAT-supported platforms. However, when developing or porting an
application that contains a mix of Ada and C/C++, the choice of your
Windows C/C++ development environment conditions your overall
interoperability strategy.

If you use @command{gcc} or Microsoft C to compile the non-Ada part of
your application, there are no Windows-specific restrictions that
affect the overall interoperability with your Ada code. If you do want
to use the Microsoft tools for your C++ code, you have two choices:

@enumerate
@item
Encapsulate your C++ code in a DLL to be linked with your Ada
application. In this case, use the Microsoft or whatever environment to
build the DLL and use GNAT to build your executable
(@pxref{Using DLLs with GNAT}).

@item
Or you can encapsulate your Ada code in a DLL to be linked with the
other part of your application. In this case, use GNAT to build the DLL
(@pxref{Building DLLs with GNAT Project files}) and use the Microsoft
or whatever environment to build your executable.
@end enumerate

In addition to the description about C main in
@pxref{Mixed Language Programming} section, if the C main uses a
stand-alone library it is required on x86-windows to
setup the SEH context. For this the C main must looks like this:

@smallexample
/* main.c */
extern void adainit (void);
extern void adafinal (void);
extern void __gnat_initialize(void*);
extern void call_to_ada (void);

int main (int argc, char *argv[])
@{
  int SEH [2];

  /* Initialize the SEH context */
  __gnat_initialize (&SEH);

  adainit();

  /* Then call Ada services in the stand-alone library */

  call_to_ada();

  adafinal();
@}
@end smallexample

Note that this is not needed on x86_64-windows where the Windows
native SEH support is used.

@node Windows Calling Conventions
@section Windows Calling Conventions
@findex Stdcall
@findex APIENTRY

This section pertain only to Win32. On Win64 there is a single native
calling convention. All convention specifiers are ignored on this
platform.

@menu
* C Calling Convention::
* Stdcall Calling Convention::
* Win32 Calling Convention::
* DLL Calling Convention::
@end menu

@noindent
When a subprogram @code{F} (caller) calls a subprogram @code{G}
(callee), there are several ways to push @code{G}'s parameters on the
stack and there are several possible scenarios to clean up the stack
upon @code{G}'s return. A calling convention is an agreed upon software
protocol whereby the responsibilities between the caller (@code{F}) and
the callee (@code{G}) are clearly defined. Several calling conventions
are available for Windows:

@itemize @bullet
@item
@code{C} (Microsoft defined)

@item
@code{Stdcall} (Microsoft defined)

@item
@code{Win32} (GNAT specific)

@item
@code{DLL} (GNAT specific)
@end itemize

@node C Calling Convention
@subsection @code{C} Calling Convention

@noindent
This is the default calling convention used when interfacing to C/C++
routines compiled with either @command{gcc} or Microsoft Visual C++.

In the @code{C} calling convention subprogram parameters are pushed on the
stack by the caller from right to left. The caller itself is in charge of
cleaning up the stack after the call. In addition, the name of a routine
with @code{C} calling convention is mangled by adding a leading underscore.

The name to use on the Ada side when importing (or exporting) a routine
with @code{C} calling convention is the name of the routine. For
instance the C function:

@smallexample
int get_val (long);
@end smallexample

@noindent
should be imported from Ada as follows:

@smallexample @c ada
@group
@b{function} Get_Val (V : Interfaces.C.long) @b{return} Interfaces.C.int;
@b{pragma} Import (C, Get_Val, External_Name => "get_val");
@end group
@end smallexample

@noindent
Note that in this particular case the @code{External_Name} parameter could
have been omitted since, when missing, this parameter is taken to be the
name of the Ada entity in lower case. When the @code{Link_Name} parameter
is missing, as in the above example, this parameter is set to be the
@code{External_Name} with a leading underscore.

When importing a variable defined in C, you should always use the @code{C}
calling convention unless the object containing the variable is part of a
DLL (in which case you should use the @code{Stdcall} calling
convention, @pxref{Stdcall Calling Convention}).

@node Stdcall Calling Convention
@subsection @code{Stdcall} Calling Convention

@noindent
This convention, which was the calling convention used for Pascal
programs, is used by Microsoft for all the routines in the Win32 API for
efficiency reasons. It must be used to import any routine for which this
convention was specified.

In the @code{Stdcall} calling convention subprogram parameters are pushed
on the stack by the caller from right to left. The callee (and not the
caller) is in charge of cleaning the stack on routine exit. In addition,
the name of a routine with @code{Stdcall} calling convention is mangled by
adding a leading underscore (as for the @code{C} calling convention) and a
trailing @code{@@}@code{@var{nn}}, where @var{nn} is the overall size (in
bytes) of the parameters passed to the routine.

The name to use on the Ada side when importing a C routine with a
@code{Stdcall} calling convention is the name of the C routine. The leading
underscore and trailing @code{@@}@code{@var{nn}} are added automatically by
the compiler. For instance the Win32 function:

@smallexample
@b{APIENTRY} int get_val (long);
@end smallexample

@noindent
should be imported from Ada as follows:

@smallexample @c ada
@group
@b{function} Get_Val (V : Interfaces.C.long) @b{return} Interfaces.C.int;
@b{pragma} Import (Stdcall, Get_Val);
--@i{  On the x86 a long is 4 bytes, so the Link_Name is "_get_val@@4"}
@end group
@end smallexample

@noindent
As for the @code{C} calling convention, when the @code{External_Name}
parameter is missing, it is taken to be the name of the Ada entity in lower
case. If instead of writing the above import pragma you write:

@smallexample @c ada
@group
@b{function} Get_Val (V : Interfaces.C.long) @b{return} Interfaces.C.int;
@b{pragma} Import (Stdcall, Get_Val, External_Name => "retrieve_val");
@end group
@end smallexample

@noindent
then the imported routine is @code{_retrieve_val@@4}. However, if instead
of specifying the @code{External_Name} parameter you specify the
@code{Link_Name} as in the following example:

@smallexample @c ada
@group
@b{function} Get_Val (V : Interfaces.C.long) @b{return} Interfaces.C.int;
@b{pragma} Import (Stdcall, Get_Val, Link_Name => "retrieve_val");
@end group
@end smallexample

@noindent
then the imported routine is @code{retrieve_val}, that is, there is no
decoration at all. No leading underscore and no Stdcall suffix
@code{@@}@code{@var{nn}}.

@noindent
This is especially important as in some special cases a DLL's entry
point name lacks a trailing @code{@@}@code{@var{nn}} while the exported
name generated for a call has it.

@noindent
It is also possible to import variables defined in a DLL by using an
import pragma for a variable. As an example, if a DLL contains a
variable defined as:

@smallexample
int my_var;
@end smallexample

@noindent
then, to access this variable from Ada you should write:

@smallexample @c ada
@group
My_Var : Interfaces.C.int;
@b{pragma} Import (Stdcall, My_Var);
@end group
@end smallexample

@noindent
Note that to ease building cross-platform bindings this convention
will be handled as a @code{C} calling convention on non-Windows platforms.

@node Win32 Calling Convention
@subsection @code{Win32} Calling Convention

@noindent
This convention, which is GNAT-specific is fully equivalent to the
@code{Stdcall} calling convention described above.

@node DLL Calling Convention
@subsection @code{DLL} Calling Convention

@noindent
This convention, which is GNAT-specific is fully equivalent to the
@code{Stdcall} calling convention described above.

@node Introduction to Dynamic Link Libraries (DLLs)
@section Introduction to Dynamic Link Libraries (DLLs)
@findex DLL

@noindent
A Dynamically Linked Library (DLL) is a library that can be shared by
several applications running under Windows. A DLL can contain any number of
routines and variables.

One advantage of DLLs is that you can change and enhance them without
forcing all the applications that depend on them to be relinked or
recompiled. However, you should be aware than all calls to DLL routines are
slower since, as you will understand below, such calls are indirect.

To illustrate the remainder of this section, suppose that an application
wants to use the services of a DLL @file{API.dll}. To use the services
provided by @file{API.dll} you must statically link against the DLL or
an import library which contains a jump table with an entry for each
routine and variable exported by the DLL. In the Microsoft world this
import library is called @file{API.lib}. When using GNAT this import
library is called either @file{libAPI.dll.a}, @file{libapi.dll.a},
@file{libAPI.a} or @file{libapi.a} (names are case insensitive).

After you have linked your application with the DLL or the import library
and you run your application, here is what happens:

@enumerate
@item
Your application is loaded into memory.

@item
The DLL @file{API.dll} is mapped into the address space of your
application. This means that:

@itemize @bullet
@item
The DLL will use the stack of the calling thread.

@item
The DLL will use the virtual address space of the calling process.

@item
The DLL will allocate memory from the virtual address space of the calling
process.

@item
Handles (pointers) can be safely exchanged between routines in the DLL
routines and routines in the application using the DLL.
@end itemize

@item
The entries in the jump table (from the import library @file{libAPI.dll.a}
or @file{API.lib} or automatically created when linking against a DLL)
which is part of your application are initialized with the addresses
of the routines and variables in @file{API.dll}.

@item
If present in @file{API.dll}, routines @code{DllMain} or
@code{DllMainCRTStartup} are invoked. These routines typically contain
the initialization code needed for the well-being of the routines and
variables exported by the DLL.
@end enumerate

@noindent
There is an additional point which is worth mentioning. In the Windows
world there are two kind of DLLs: relocatable and non-relocatable
DLLs. Non-relocatable DLLs can only be loaded at a very specific address
in the target application address space. If the addresses of two
non-relocatable DLLs overlap and these happen to be used by the same
application, a conflict will occur and the application will run
incorrectly. Hence, when possible, it is always preferable to use and
build relocatable DLLs. Both relocatable and non-relocatable DLLs are
supported by GNAT. Note that the @option{-s} linker option (see GNU Linker
User's Guide) removes the debugging symbols from the DLL but the DLL can
still be relocated.

As a side note, an interesting difference between Microsoft DLLs and
Unix shared libraries, is the fact that on most Unix systems all public
routines are exported by default in a Unix shared library, while under
Windows it is possible (but not required) to list exported routines in
a definition file (@pxref{The Definition File}).

@node Using DLLs with GNAT
@section Using DLLs with GNAT

@menu
* Creating an Ada Spec for the DLL Services::
* Creating an Import Library::
@end menu

@noindent
To use the services of a DLL, say @file{API.dll}, in your Ada application
you must have:

@enumerate
@item
The Ada spec for the routines and/or variables you want to access in
@file{API.dll}. If not available this Ada spec must be built from the C/C++
header files provided with the DLL.

@item
The import library (@file{libAPI.dll.a} or @file{API.lib}). As previously
mentioned an import library is a statically linked library containing the
import table which will be filled at load time to point to the actual
@file{API.dll} routines. Sometimes you don't have an import library for the
DLL you want to use. The following sections will explain how to build
one. Note that this is optional.

@item
The actual DLL, @file{API.dll}.
@end enumerate

@noindent
Once you have all the above, to compile an Ada application that uses the
services of @file{API.dll} and whose main subprogram is @code{My_Ada_App},
you simply issue the command

@smallexample
$ gnatmake my_ada_app -largs -lAPI
@end smallexample

@noindent
The argument @option{-largs -lAPI} at the end of the @command{gnatmake} command
tells the GNAT linker to look for an import library. The linker will
look for a library name in this specific order:

@enumerate
@item @file{libAPI.dll.a}
@item @file{API.dll.a}
@item @file{libAPI.a}
@item @file{API.lib}
@item @file{libAPI.dll}
@item @file{API.dll}
@end enumerate

The first three are the GNU style import libraries. The third is the
Microsoft style import libraries. The last two are the actual DLL names.

Note that if the Ada package spec for @file{API.dll} contains the
following pragma

@smallexample @c ada
@b{pragma} Linker_Options ("-lAPI");
@end smallexample

@noindent
you do not have to add @option{-largs -lAPI} at the end of the
@command{gnatmake} command.

If any one of the items above is missing you will have to create it
yourself. The following sections explain how to do so using as an
example a fictitious DLL called @file{API.dll}.

@node Creating an Ada Spec for the DLL Services
@subsection Creating an Ada Spec for the DLL Services

@noindent
A DLL typically comes with a C/C++ header file which provides the
definitions of the routines and variables exported by the DLL. The Ada
equivalent of this header file is a package spec that contains definitions
for the imported entities. If the DLL you intend to use does not come with
an Ada spec you have to generate one such spec yourself. For example if
the header file of @file{API.dll} is a file @file{api.h} containing the
following two definitions:

@smallexample
@group
@cartouche
int some_var;
int get (char *);
@end cartouche
@end group
@end smallexample

@noindent
then the equivalent Ada spec could be:

@smallexample @c ada
@group
@cartouche
@b{with} Interfaces.C.Strings;
@b{package} API @b{is}
   @b{use} Interfaces;

   Some_Var : C.int;
   @b{function} Get (Str : C.Strings.Chars_Ptr) @b{return} C.int;

@b{private}
   @b{pragma} Import (C, Get);
   @b{pragma} Import (DLL, Some_Var);
@b{end} API;
@end cartouche
@end group
@end smallexample

@node Creating an Import Library
@subsection Creating an Import Library
@cindex Import library

@menu
* The Definition File::
* GNAT-Style Import Library::
* Microsoft-Style Import Library::
@end menu

@noindent
If a Microsoft-style import library @file{API.lib} or a GNAT-style
import library @file{libAPI.dll.a} or @file{libAPI.a} is available
with @file{API.dll} you can skip this section. You can also skip this
section if @file{API.dll} or @file{libAPI.dll} is built with GNU tools
as in this case it is possible to link directly against the
DLL. Otherwise read on.

@node The Definition File
@subsubsection The Definition File
@cindex Definition file
@findex .def

@noindent
As previously mentioned, and unlike Unix systems, the list of symbols
that are exported from a DLL must be provided explicitly in Windows.
The main goal of a definition file is precisely that: list the symbols
exported by a DLL. A definition file (usually a file with a @code{.def}
suffix) has the following structure:

@smallexample
@group
@cartouche
@r{[}LIBRARY @var{name}@r{]}
@r{[}DESCRIPTION @var{string}@r{]}
EXPORTS
   @var{symbol1}
   @var{symbol2}
   @dots{}
@end cartouche
@end group
@end smallexample

@table @code
@item LIBRARY @var{name}
This section, which is optional, gives the name of the DLL.

@item DESCRIPTION @var{string}
This section, which is optional, gives a description string that will be
embedded in the import library.

@item EXPORTS
This section gives the list of exported symbols (procedures, functions or
variables). For instance in the case of @file{API.dll} the @code{EXPORTS}
section of @file{API.def} looks like:

@smallexample
@group
@cartouche
EXPORTS
   some_var
   get
@end cartouche
@end group
@end smallexample
@end table

@noindent
Note that you must specify the correct suffix (@code{@@}@code{@var{nn}})
(@pxref{Windows Calling Conventions}) for a Stdcall
calling convention function in the exported symbols list.

@noindent
There can actually be other sections in a definition file, but these
sections are not relevant to the discussion at hand.

@node GNAT-Style Import Library
@subsubsection GNAT-Style Import Library

@noindent
To create a static import library from @file{API.dll} with the GNAT tools
you should proceed as follows:

@enumerate
@item
Create the definition file @file{API.def} (@pxref{The Definition File}).
For that use the @code{dll2def} tool as follows:

@smallexample
$ dll2def API.dll > API.def
@end smallexample

@noindent
@code{dll2def} is a very simple tool: it takes as input a DLL and prints
to standard output the list of entry points in the DLL. Note that if
some routines in the DLL have the @code{Stdcall} convention
(@pxref{Windows Calling Conventions}) with stripped @code{@@}@var{nn}
suffix then you'll have to edit @file{api.def} to add it, and specify
@option{-k} to @command{gnatdll} when creating the import library.

@noindent
Here are some hints to find the right @code{@@}@var{nn} suffix.

@enumerate
@item
If you have the Microsoft import library (.lib), it is possible to get
the right symbols by using Microsoft @code{dumpbin} tool (see the
corresponding Microsoft documentation for further details).

@smallexample
$ dumpbin /exports api.lib
@end smallexample

@item
If you have a message about a missing symbol at link time the compiler
tells you what symbol is expected. You just have to go back to the
definition file and add the right suffix.
@end enumerate

@item
Build the import library @code{libAPI.dll.a}, using @code{gnatdll}
(@pxref{Using gnatdll}) as follows:

@smallexample
$ gnatdll -e API.def -d API.dll
@end smallexample

@noindent
@code{gnatdll} takes as input a definition file @file{API.def} and the
name of the DLL containing the services listed in the definition file
@file{API.dll}. The name of the static import library generated is
computed from the name of the definition file as follows: if the
definition file name is @var{xyz}@code{.def}, the import library name will
be @code{lib}@var{xyz}@code{.a}. Note that in the previous example option
@option{-e} could have been removed because the name of the definition
file (before the ``@code{.def}'' suffix) is the same as the name of the
DLL (@pxref{Using gnatdll} for more information about @code{gnatdll}).
@end enumerate

@node Microsoft-Style Import Library
@subsubsection Microsoft-Style Import Library

@noindent
With GNAT you can either use a GNAT-style or Microsoft-style import
library. A Microsoft import library is needed only if you plan to make an
Ada DLL available to applications developed with Microsoft
tools (@pxref{Mixed-Language Programming on Windows}).

To create a Microsoft-style import library for @file{API.dll} you
should proceed as follows:

@enumerate
@item
Create the definition file @file{API.def} from the DLL. For this use either
the @code{dll2def} tool as described above or the Microsoft @code{dumpbin}
tool (see the corresponding Microsoft documentation for further details).

@item
Build the actual import library using Microsoft's @code{lib} utility:

@smallexample
$ lib -machine:IX86 -def:API.def -out:API.lib
@end smallexample

@noindent
If you use the above command the definition file @file{API.def} must
contain a line giving the name of the DLL:

@smallexample
LIBRARY      "API"
@end smallexample

@noindent
See the Microsoft documentation for further details about the usage of
@code{lib}.
@end enumerate

@node Building DLLs with GNAT Project files
@section Building DLLs with GNAT Project files
@cindex DLLs, building

@noindent
There is nothing specific to Windows in the build process.
@pxref{Library Projects}.

@noindent
Due to a system limitation, it is not possible under Windows to create threads
when inside the @code{DllMain} routine which is used for auto-initialization
of shared libraries, so it is not possible to have library level tasks in SALs.

@node Building DLLs with GNAT
@section Building DLLs with GNAT
@cindex DLLs, building

@noindent
This section explain how to build DLLs using the GNAT built-in DLL
support. With the following procedure it is straight forward to build
and use DLLs with GNAT.

@enumerate

@item building object files

The first step is to build all objects files that are to be included
into the DLL. This is done by using the standard @command{gnatmake} tool.

@item building the DLL

To build the DLL you must use @command{gcc}'s @option{-shared} and
@option{-shared-libgcc} options. It is quite simple to use this method:

@smallexample
$ gcc -shared -shared-libgcc -o api.dll obj1.o obj2.o @dots{}
@end smallexample

It is important to note that in this case all symbols found in the
object files are automatically exported. It is possible to restrict
the set of symbols to export by passing to @command{gcc} a definition
file, @pxref{The Definition File}. For example:

@smallexample
$ gcc -shared -shared-libgcc -o api.dll api.def obj1.o obj2.o @dots{}
@end smallexample

If you use a definition file you must export the elaboration procedures
for every package that required one. Elaboration procedures are named
using the package name followed by "_E".

@item preparing DLL to be used

For the DLL to be used by client programs the bodies must be hidden
from it and the .ali set with read-only attribute. This is very important
otherwise GNAT will recompile all packages and will not actually use
the code in the DLL. For example:

@smallexample
$ mkdir apilib
$ copy *.ads *.ali api.dll apilib
$ attrib +R apilib\*.ali
@end smallexample

@end enumerate

At this point it is possible to use the DLL by directly linking
against it. Note that you must use the GNAT shared runtime when using
GNAT shared libraries. This is achieved by using @option{-shared} binder's
option.

@smallexample
$ gnatmake main -Iapilib -bargs -shared -largs -Lapilib -lAPI
@end smallexample

@node Building DLLs with gnatdll
@section Building DLLs with gnatdll
@cindex DLLs, building

@menu
* Limitations When Using Ada DLLs from Ada::
* Exporting Ada Entities::
* Ada DLLs and Elaboration::
* Ada DLLs and Finalization::
* Creating a Spec for Ada DLLs::
* Creating the Definition File::
* Using gnatdll::
@end menu

@noindent
Note that it is preferred to use GNAT Project files
(@pxref{Building DLLs with GNAT Project files}) or the built-in GNAT
DLL support (@pxref{Building DLLs with GNAT}) or to build DLLs.

This section explains how to build DLLs containing Ada code using
@code{gnatdll}. These DLLs will be referred to as Ada DLLs in the
remainder of this section.

The steps required to build an Ada DLL that is to be used by Ada as well as
non-Ada applications are as follows:

@enumerate
@item
You need to mark each Ada @i{entity} exported by the DLL with a @code{C} or
@code{Stdcall} calling convention to avoid any Ada name mangling for the
entities exported by the DLL (@pxref{Exporting Ada Entities}). You can
skip this step if you plan to use the Ada DLL only from Ada applications.

@item
Your Ada code must export an initialization routine which calls the routine
@code{adainit} generated by @command{gnatbind} to perform the elaboration of
the Ada code in the DLL (@pxref{Ada DLLs and Elaboration}). The initialization
routine exported by the Ada DLL must be invoked by the clients of the DLL
to initialize the DLL.

@item
When useful, the DLL should also export a finalization routine which calls
routine @code{adafinal} generated by @command{gnatbind} to perform the
finalization of the Ada code in the DLL (@pxref{Ada DLLs and Finalization}).
The finalization routine exported by the Ada DLL must be invoked by the
clients of the DLL when the DLL services are no further needed.

@item
You must provide a spec for the services exported by the Ada DLL in each
of the programming languages to which you plan to make the DLL available.

@item
You must provide a definition file listing the exported entities
(@pxref{The Definition File}).

@item
Finally you must use @code{gnatdll} to produce the DLL and the import
library (@pxref{Using gnatdll}).
@end enumerate

@noindent
Note that a relocatable DLL stripped using the @code{strip}
binutils tool will not be relocatable anymore. To build a DLL without
debug information pass @code{-largs -s} to @code{gnatdll}. This
restriction does not apply to a DLL built using a Library Project.
@pxref{Library Projects}.

@node Limitations When Using Ada DLLs from Ada
@subsection Limitations When Using Ada DLLs from Ada

@noindent
When using Ada DLLs from Ada applications there is a limitation users
should be aware of. Because on Windows the GNAT run time is not in a DLL of
its own, each Ada DLL includes a part of the GNAT run time. Specifically,
each Ada DLL includes the services of the GNAT run time that are necessary
to the Ada code inside the DLL. As a result, when an Ada program uses an
Ada DLL there are two independent GNAT run times: one in the Ada DLL and
one in the main program.

It is therefore not possible to exchange GNAT run-time objects between the
Ada DLL and the main Ada program. Example of GNAT run-time objects are file
handles (e.g.@: @code{Text_IO.File_Type}), tasks types, protected objects
types, etc.

It is completely safe to exchange plain elementary, array or record types,
Windows object handles, etc.

@node Exporting Ada Entities
@subsection Exporting Ada Entities
@cindex Export table

@noindent
Building a DLL is a way to encapsulate a set of services usable from any
application. As a result, the Ada entities exported by a DLL should be
exported with the @code{C} or @code{Stdcall} calling conventions to avoid
any Ada name mangling. As an example here is an Ada package
@code{API}, spec and body, exporting two procedures, a function, and a
variable:

@smallexample @c ada
@group
@cartouche
@b{with} Interfaces.C; @b{use} Interfaces;
@b{package} API @b{is}
   Count : C.int := 0;
   @b{function} Factorial (Val : C.int) @b{return} C.int;

   @b{procedure} Initialize_API;
   @b{procedure} Finalize_API;
   --@i{  Initialization & Finalization routines. More in the next section.}
@b{private}
   @b{pragma} Export (C, Initialize_API);
   @b{pragma} Export (C, Finalize_API);
   @b{pragma} Export (C, Count);
   @b{pragma} Export (C, Factorial);
@b{end} API;
@end cartouche
@end group
@end smallexample

@smallexample @c ada
@group
@cartouche
@b{package} @b{body} API @b{is}
   @b{function} Factorial (Val : C.int) @b{return} C.int @b{is}
      Fact : C.int := 1;
   @b{begin}
      Count := Count + 1;
      @b{for} K @b{in} 1 .. Val @b{loop}
         Fact := Fact * K;
      @b{end} @b{loop};
      @b{return} Fact;
   @b{end} Factorial;

   @b{procedure} Initialize_API @b{is}
      @b{procedure} Adainit;
      @b{pragma} Import (C, Adainit);
   @b{begin}
      Adainit;
   @b{end} Initialize_API;

   @b{procedure} Finalize_API @b{is}
      @b{procedure} Adafinal;
      @b{pragma} Import (C, Adafinal);
   @b{begin}
      Adafinal;
   @b{end} Finalize_API;
@b{end} API;
@end cartouche
@end group
@end smallexample

@noindent
If the Ada DLL you are building will only be used by Ada applications
you do not have to export Ada entities with a @code{C} or @code{Stdcall}
convention. As an example, the previous package could be written as
follows:

@smallexample @c ada
@group
@cartouche
@b{package} API @b{is}
   Count : Integer := 0;
   @b{function} Factorial (Val : Integer) @b{return} Integer;

   @b{procedure} Initialize_API;
   @b{procedure} Finalize_API;
   --@i{  Initialization and Finalization routines.}
@b{end} API;
@end cartouche
@end group
@end smallexample

@smallexample @c ada
@group
@cartouche
@b{package} @b{body} API @b{is}
   @b{function} Factorial (Val : Integer) @b{return} Integer @b{is}
      Fact : Integer := 1;
   @b{begin}
      Count := Count + 1;
      @b{for} K @b{in} 1 .. Val @b{loop}
         Fact := Fact * K;
      @b{end} @b{loop};
      @b{return} Fact;
   @b{end} Factorial;

   @dots{}
   --@i{  The remainder of this package body is unchanged.}
@b{end} API;
@end cartouche
@end group
@end smallexample

@noindent
Note that if you do not export the Ada entities with a @code{C} or
@code{Stdcall} convention you will have to provide the mangled Ada names
in the definition file of the Ada DLL
(@pxref{Creating the Definition File}).

@node Ada DLLs and Elaboration
@subsection Ada DLLs and Elaboration
@cindex DLLs and elaboration

@noindent
The DLL that you are building contains your Ada code as well as all the
routines in the Ada library that are needed by it. The first thing a
user of your DLL must do is elaborate the Ada code
(@pxref{Elaboration Order Handling in GNAT}).

To achieve this you must export an initialization routine
(@code{Initialize_API} in the previous example), which must be invoked
before using any of the DLL services. This elaboration routine must call
the Ada elaboration routine @code{adainit} generated by the GNAT binder
(@pxref{Binding with Non-Ada Main Programs}). See the body of
@code{Initialize_Api} for an example. Note that the GNAT binder is
automatically invoked during the DLL build process by the @code{gnatdll}
tool (@pxref{Using gnatdll}).

When a DLL is loaded, Windows systematically invokes a routine called
@code{DllMain}. It would therefore be possible to call @code{adainit}
directly from @code{DllMain} without having to provide an explicit
initialization routine. Unfortunately, it is not possible to call
@code{adainit} from the @code{DllMain} if your program has library level
tasks because access to the @code{DllMain} entry point is serialized by
the system (that is, only a single thread can execute ``through'' it at a
time), which means that the GNAT run time will deadlock waiting for the
newly created task to complete its initialization.

@node Ada DLLs and Finalization
@subsection Ada DLLs and Finalization
@cindex DLLs and finalization

@noindent
When the services of an Ada DLL are no longer needed, the client code should
invoke the DLL finalization routine, if available. The DLL finalization
routine is in charge of releasing all resources acquired by the DLL. In the
case of the Ada code contained in the DLL, this is achieved by calling
routine @code{adafinal} generated by the GNAT binder
(@pxref{Binding with Non-Ada Main Programs}).
See the body of @code{Finalize_Api} for an
example. As already pointed out the GNAT binder is automatically invoked
during the DLL build process by the @code{gnatdll} tool
(@pxref{Using gnatdll}).

@node Creating a Spec for Ada DLLs
@subsection Creating a Spec for Ada DLLs

@noindent
To use the services exported by the Ada DLL from another programming
language (e.g.@: C), you have to translate the specs of the exported Ada
entities in that language. For instance in the case of @code{API.dll},
the corresponding C header file could look like:

@smallexample
@group
@cartouche
extern int *_imp__count;
#define count (*_imp__count)
int factorial (int);
@end cartouche
@end group
@end smallexample

@noindent
It is important to understand that when building an Ada DLL to be used by
other Ada applications, you need two different specs for the packages
contained in the DLL: one for building the DLL and the other for using
the DLL. This is because the @code{DLL} calling convention is needed to
use a variable defined in a DLL, but when building the DLL, the variable
must have either the @code{Ada} or @code{C} calling convention. As an
example consider a DLL comprising the following package @code{API}:

@smallexample @c ada
@group
@cartouche
@b{package} API @b{is}
   Count : Integer := 0;
   @dots{}
   --@i{  Remainder of the package omitted.}
@b{end} API;
@end cartouche
@end group
@end smallexample

@noindent
After producing a DLL containing package @code{API}, the spec that
must be used to import @code{API.Count} from Ada code outside of the
DLL is:

@smallexample @c ada
@group
@cartouche
@b{package} API @b{is}
   Count : Integer;
   @b{pragma} Import (DLL, Count);
@b{end} API;
@end cartouche
@end group
@end smallexample

@node Creating the Definition File
@subsection Creating the Definition File

@noindent
The definition file is the last file needed to build the DLL. It lists
the exported symbols. As an example, the definition file for a DLL
containing only package @code{API} (where all the entities are exported
with a @code{C} calling convention) is:

@smallexample
@group
@cartouche
EXPORTS
    count
    factorial
    finalize_api
    initialize_api
@end cartouche
@end group
@end smallexample

@noindent
If the @code{C} calling convention is missing from package @code{API},
then the definition file contains the mangled Ada names of the above
entities, which in this case are:

@smallexample
@group
@cartouche
EXPORTS
    api__count
    api__factorial
    api__finalize_api
    api__initialize_api
@end cartouche
@end group
@end smallexample

@node Using gnatdll
@subsection Using @code{gnatdll}
@findex gnatdll

@menu
* gnatdll Example::
* gnatdll behind the Scenes::
* Using dlltool::
@end menu

@noindent
@code{gnatdll} is a tool to automate the DLL build process once all the Ada
and non-Ada sources that make up your DLL have been compiled.
@code{gnatdll} is actually in charge of two distinct tasks: build the
static import library for the DLL and the actual DLL. The form of the
@code{gnatdll} command is

@smallexample
@cartouche
@c $ gnatdll @ovar{switches} @var{list-of-files} @r{[}-largs @var{opts}@r{]}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ gnatdll @r{[}@var{switches}@r{]} @var{list-of-files} @r{[}-largs @var{opts}@r{]}
@end cartouche
@end smallexample

@noindent
where @var{list-of-files} is a list of ALI and object files. The object
file list must be the exact list of objects corresponding to the non-Ada
sources whose services are to be included in the DLL. The ALI file list
must be the exact list of ALI files for the corresponding Ada sources
whose services are to be included in the DLL. If @var{list-of-files} is
missing, only the static import library is generated.

@noindent
You may specify any of the following switches to @code{gnatdll}:

@table @code
@c @item -a@ovar{address}
@c Expanding @ovar macro inline (explanation in macro def comments)
@item -a@r{[}@var{address}@r{]}
@cindex @option{-a} (@code{gnatdll})
Build a non-relocatable DLL at @var{address}. If @var{address} is not
specified the default address @var{0x11000000} will be used. By default,
when this switch is missing, @code{gnatdll} builds relocatable DLL. We
advise the reader to build relocatable DLL.

@item -b @var{address}
@cindex @option{-b} (@code{gnatdll})
Set the relocatable DLL base address. By default the address is
@code{0x11000000}.

@item -bargs @var{opts}
@cindex @option{-bargs} (@code{gnatdll})
Binder options. Pass @var{opts} to the binder.

@item -d @var{dllfile}
@cindex @option{-d} (@code{gnatdll})
@var{dllfile} is the name of the DLL. This switch must be present for
@code{gnatdll} to do anything. The name of the generated import library is
obtained algorithmically from @var{dllfile} as shown in the following
example: if @var{dllfile} is @code{xyz.dll}, the import library name is
@code{libxyz.dll.a}. The name of the definition file to use (if not specified
by option @option{-e}) is obtained algorithmically from @var{dllfile}
as shown in the following example:
if @var{dllfile} is @code{xyz.dll}, the definition
file used is @code{xyz.def}.

@item -e @var{deffile}
@cindex @option{-e} (@code{gnatdll})
@var{deffile} is the name of the definition file.

@item -g
@cindex @option{-g} (@code{gnatdll})
Generate debugging information. This information is stored in the object
file and copied from there to the final DLL file by the linker,
where it can be read by the debugger. You must use the
@option{-g} switch if you plan on using the debugger or the symbolic
stack traceback.

@item -h
@cindex @option{-h} (@code{gnatdll})
Help mode. Displays @code{gnatdll} switch usage information.

@item -Idir
@cindex @option{-I} (@code{gnatdll})
Direct @code{gnatdll} to search the @var{dir} directory for source and
object files needed to build the DLL.
(@pxref{Search Paths and the Run-Time Library (RTL)}).

@item -k
@cindex @option{-k} (@code{gnatdll})
Removes the @code{@@}@var{nn} suffix from the import library's exported
names, but keeps them for the link names. You must specify this
option if you want to use a @code{Stdcall} function in a DLL for which
the @code{@@}@var{nn} suffix has been removed. This is the case for most
of the Windows NT DLL for example. This option has no effect when
@option{-n} option is specified.

@item -l @var{file}
@cindex @option{-l} (@code{gnatdll})
The list of ALI and object files used to build the DLL are listed in
@var{file}, instead of being given in the command line. Each line in
@var{file} contains the name of an ALI or object file.

@item -n
@cindex @option{-n} (@code{gnatdll})
No Import. Do not create the import library.

@item -q
@cindex @option{-q} (@code{gnatdll})
Quiet mode. Do not display unnecessary messages.

@item -v
@cindex @option{-v} (@code{gnatdll})
Verbose mode. Display extra information.

@item -largs @var{opts}
@cindex @option{-largs} (@code{gnatdll})
Linker options. Pass @var{opts} to the linker.
@end table

@node gnatdll Example
@subsubsection @code{gnatdll} Example

@noindent
As an example the command to build a relocatable DLL from @file{api.adb}
once @file{api.adb} has been compiled and @file{api.def} created is

@smallexample
$ gnatdll -d api.dll api.ali
@end smallexample

@noindent
The above command creates two files: @file{libapi.dll.a} (the import
library) and @file{api.dll} (the actual DLL). If you want to create
only the DLL, just type:

@smallexample
$ gnatdll -d api.dll -n api.ali
@end smallexample

@noindent
Alternatively if you want to create just the import library, type:

@smallexample
$ gnatdll -d api.dll
@end smallexample

@node gnatdll behind the Scenes
@subsubsection @code{gnatdll} behind the Scenes

@noindent
This section details the steps involved in creating a DLL. @code{gnatdll}
does these steps for you. Unless you are interested in understanding what
goes on behind the scenes, you should skip this section.

We use the previous example of a DLL containing the Ada package @code{API},
to illustrate the steps necessary to build a DLL. The starting point is a
set of objects that will make up the DLL and the corresponding ALI
files. In the case of this example this means that @file{api.o} and
@file{api.ali} are available. To build a relocatable DLL, @code{gnatdll} does
the following:

@enumerate
@item
@code{gnatdll} builds the base file (@file{api.base}). A base file gives
the information necessary to generate relocation information for the
DLL.

@smallexample
@group
$ gnatbind -n api
$ gnatlink api -o api.jnk -mdll -Wl,--base-file,api.base
@end group
@end smallexample

@noindent
In addition to the base file, the @command{gnatlink} command generates an
output file @file{api.jnk} which can be discarded. The @option{-mdll} switch
asks @command{gnatlink} to generate the routines @code{DllMain} and
@code{DllMainCRTStartup} that are called by the Windows loader when the DLL
is loaded into memory.

@item
@code{gnatdll} uses @code{dlltool} (@pxref{Using dlltool}) to build the
export table (@file{api.exp}). The export table contains the relocation
information in a form which can be used during the final link to ensure
that the Windows loader is able to place the DLL anywhere in memory.

@smallexample
@group
$ dlltool --dllname api.dll --def api.def --base-file api.base \
          --output-exp api.exp
@end group
@end smallexample

@item
@code{gnatdll} builds the base file using the new export table. Note that
@command{gnatbind} must be called once again since the binder generated file
has been deleted during the previous call to @command{gnatlink}.

@smallexample
@group
$ gnatbind -n api
$ gnatlink api -o api.jnk api.exp -mdll
      -Wl,--base-file,api.base
@end group
@end smallexample

@item
@code{gnatdll} builds the new export table using the new base file and
generates the DLL import library @file{libAPI.dll.a}.

@smallexample
@group
$ dlltool --dllname api.dll --def api.def --base-file api.base \
          --output-exp api.exp --output-lib libAPI.a
@end group
@end smallexample

@item
Finally @code{gnatdll} builds the relocatable DLL using the final export
table.

@smallexample
@group
$ gnatbind -n api
$ gnatlink api api.exp -o api.dll -mdll
@end group
@end smallexample
@end enumerate

@node Using dlltool
@subsubsection Using @code{dlltool}

@noindent
@code{dlltool} is the low-level tool used by @code{gnatdll} to build
DLLs and static import libraries. This section summarizes the most
common @code{dlltool} switches. The form of the @code{dlltool} command
is

@smallexample
@c $ dlltool @ovar{switches}
@c Expanding @ovar macro inline (explanation in macro def comments)
$ dlltool @r{[}@var{switches}@r{]}
@end smallexample

@noindent
@code{dlltool} switches include:

@table @option
@item --base-file @var{basefile}
@cindex @option{--base-file} (@command{dlltool})
Read the base file @var{basefile} generated by the linker. This switch
is used to create a relocatable DLL.

@item --def @var{deffile}
@cindex @option{--def} (@command{dlltool})
Read the definition file.

@item --dllname @var{name}
@cindex @option{--dllname} (@command{dlltool})
Gives the name of the DLL. This switch is used to embed the name of the
DLL in the static import library generated by @code{dlltool} with switch
@option{--output-lib}.

@item -k
@cindex @option{-k} (@command{dlltool})
Kill @code{@@}@var{nn} from exported names
(@pxref{Windows Calling Conventions}
for a discussion about @code{Stdcall}-style symbols.

@item --help
@cindex @option{--help} (@command{dlltool})
Prints the @code{dlltool} switches with a concise description.

@item --output-exp @var{exportfile}
@cindex @option{--output-exp} (@command{dlltool})
Generate an export file @var{exportfile}. The export file contains the
export table (list of symbols in the DLL) and is used to create the DLL.

@item --output-lib @var{libfile}
@cindex @option{--output-lib} (@command{dlltool})
Generate a static import library @var{libfile}.

@item -v
@cindex @option{-v} (@command{dlltool})
Verbose mode.

@item --as @var{assembler-name}
@cindex @option{--as} (@command{dlltool})
Use @var{assembler-name} as the assembler. The default is @code{as}.
@end table

@node GNAT and Windows Resources
@section GNAT and Windows Resources
@cindex Resources, windows

@menu
* Building Resources::
* Compiling Resources::
* Using Resources::
@end menu

@noindent
Resources are an easy way to add Windows specific objects to your
application. The objects that can be added as resources include:

@itemize @bullet
@item menus

@item accelerators

@item dialog boxes

@item string tables

@item bitmaps

@item cursors

@item icons

@item fonts

@item version information
@end itemize

For example, a version information resource can be defined as follow and
embedded into an executable or DLL:

A version information resource can be used to embed information into an
executable or a DLL. These information can be viewed using the file properties
from the Windows Explorer. Here is an example of a version information
resource:

@smallexample
@group
1 VERSIONINFO
FILEVERSION     1,0,0,0
PRODUCTVERSION  1,0,0,0
BEGIN
  BLOCK "StringFileInfo"
  BEGIN
    BLOCK "080904E4"
    BEGIN
      VALUE "CompanyName", "My Company Name"
      VALUE "FileDescription", "My application"
      VALUE "FileVersion", "1.0"
      VALUE "InternalName", "my_app"
      VALUE "LegalCopyright", "My Name"
      VALUE "OriginalFilename", "my_app.exe"
      VALUE "ProductName", "My App"
      VALUE "ProductVersion", "1.0"
    END
  END

  BLOCK "VarFileInfo"
  BEGIN
    VALUE "Translation", 0x809, 1252
  END
END
@end group
@end smallexample

The value @code{0809} (langID) is for the U.K English language and
@code{04E4} (charsetID), which is equal to @code{1252} decimal, for
multilingual.

@noindent
This section explains how to build, compile and use resources. Note that this
section does not cover all resource objects, for a complete description see
the corresponding Microsoft documentation.

@node Building Resources
@subsection Building Resources
@cindex Resources, building

@noindent
A resource file is an ASCII file. By convention resource files have an
@file{.rc} extension.
The easiest way to build a resource file is to use Microsoft tools
such as @code{imagedit.exe} to build bitmaps, icons and cursors and
@code{dlgedit.exe} to build dialogs.
It is always possible to build an @file{.rc} file yourself by writing a
resource script.

It is not our objective to explain how to write a resource file. A
complete description of the resource script language can be found in the
Microsoft documentation.

@node Compiling Resources
@subsection Compiling Resources
@findex rc
@findex windres
@cindex Resources, compiling

@noindent
This section describes how to build a GNAT-compatible (COFF) object file
containing the resources. This is done using the Resource Compiler
@code{windres} as follows:

@smallexample
$ windres -i myres.rc -o myres.o
@end smallexample

@noindent
By default @code{windres} will run @command{gcc} to preprocess the @file{.rc}
file. You can specify an alternate preprocessor (usually named
@file{cpp.exe}) using the @code{windres} @option{--preprocessor}
parameter. A list of all possible options may be obtained by entering
the command @code{windres} @option{--help}.

It is also possible to use the Microsoft resource compiler @code{rc.exe}
to produce a @file{.res} file (binary resource file). See the
corresponding Microsoft documentation for further details. In this case
you need to use @code{windres} to translate the @file{.res} file to a
GNAT-compatible object file as follows:

@smallexample
$ windres -i myres.res -o myres.o
@end smallexample

@node Using Resources
@subsection Using Resources
@cindex Resources, using

@noindent
To include the resource file in your program just add the
GNAT-compatible object file for the resource(s) to the linker
arguments. With @command{gnatmake} this is done by using the @option{-largs}
option:

@smallexample
$ gnatmake myprog -largs myres.o
@end smallexample

@node Debugging a DLL
@section Debugging a DLL
@cindex DLL debugging

@menu
* Program and DLL Both Built with GCC/GNAT::
* Program Built with Foreign Tools and DLL Built with GCC/GNAT::
@end menu

@noindent
Debugging a DLL is similar to debugging a standard program. But
we have to deal with two different executable parts: the DLL and the
program that uses it. We have the following four possibilities:

@enumerate 1
@item
The program and the DLL are built with @code{GCC/GNAT}.
@item
The program is built with foreign tools and the DLL is built with
@code{GCC/GNAT}.
@item
The program is built with @code{GCC/GNAT} and the DLL is built with
foreign tools.
@end enumerate

@noindent
In this section we address only cases one and two above.
There is no point in trying to debug
a DLL with @code{GNU/GDB}, if there is no GDB-compatible debugging
information in it. To do so you must use a debugger compatible with the
tools suite used to build the DLL.

@node Program and DLL Both Built with GCC/GNAT
@subsection Program and DLL Both Built with GCC/GNAT

@noindent
This is the simplest case. Both the DLL and the program have @code{GDB}
compatible debugging information. It is then possible to break anywhere in
the process. Let's suppose here that the main procedure is named
@code{ada_main} and that in the DLL there is an entry point named
@code{ada_dll}.

@noindent
The DLL (@pxref{Introduction to Dynamic Link Libraries (DLLs)}) and
program must have been built with the debugging information (see GNAT -g
switch). Here are the step-by-step instructions for debugging it:

@enumerate 1
@item Launch @code{GDB} on the main program.

@smallexample
$ gdb -nw ada_main
@end smallexample

@item Start the program and stop at the beginning of the main procedure

@smallexample
(gdb) start
@end smallexample

@noindent
This step is required to be able to set a breakpoint inside the DLL. As long
as the program is not run, the DLL is not loaded. This has the
consequence that the DLL debugging information is also not loaded, so it is not
possible to set a breakpoint in the DLL.

@item Set a breakpoint inside the DLL

@smallexample
(gdb) break ada_dll
(gdb) cont
@end smallexample

@end enumerate

@noindent
At this stage a breakpoint is set inside the DLL. From there on
you can use the standard approach to debug the whole program
(@pxref{Running and Debugging Ada Programs}).

@ignore
@c This used to work, probably because the DLLs were non-relocatable
@c keep this section around until the problem is sorted out.

To break on the @code{DllMain} routine it is not possible to follow
the procedure above. At the time the program stop on @code{ada_main}
the @code{DllMain} routine as already been called. Either you can use
the procedure below @pxref{Debugging the DLL Directly} or this procedure:

@enumerate 1
@item Launch @code{GDB} on the main program.

@smallexample
$ gdb ada_main
@end smallexample

@item Load DLL symbols

@smallexample
(gdb) add-sym api.dll
@end smallexample

@item Set a breakpoint inside the DLL

@smallexample
(gdb) break ada_dll.adb:45
@end smallexample

Note that at this point it is not possible to break using the routine symbol
directly as the program is not yet running. The solution is to break
on the proper line (break in @file{ada_dll.adb} line 45).

@item Start the program

@smallexample
(gdb) run
@end smallexample

@end enumerate
@end ignore

@node Program Built with Foreign Tools and DLL Built with GCC/GNAT
@subsection Program Built with Foreign Tools and DLL Built with GCC/GNAT

@menu
* Debugging the DLL Directly::
* Attaching to a Running Process::
@end menu

@noindent
In this case things are slightly more complex because it is not possible to
start the main program and then break at the beginning to load the DLL and the
associated DLL debugging information. It is not possible to break at the
beginning of the program because there is no @code{GDB} debugging information,
and therefore there is no direct way of getting initial control. This
section addresses this issue by describing some methods that can be used
to break somewhere in the DLL to debug it.

@noindent
First suppose that the main procedure is named @code{main} (this is for
example some C code built with Microsoft Visual C) and that there is a
DLL named @code{test.dll} containing an Ada entry point named
@code{ada_dll}.

@noindent
The DLL (@pxref{Introduction to Dynamic Link Libraries (DLLs)}) must have
been built with debugging information (see GNAT -g option).

@node Debugging the DLL Directly
@subsubsection Debugging the DLL Directly

@enumerate 1
@item
Find out the executable starting address

@smallexample
$ objdump --file-header main.exe
@end smallexample

The starting address is reported on the last line. For example:

@smallexample
main.exe:     file format pei-i386
architecture: i386, flags 0x0000010a:
EXEC_P, HAS_DEBUG, D_PAGED
start address 0x00401010
@end smallexample

@item
Launch the debugger on the executable.

@smallexample
$ gdb main.exe
@end smallexample

@item
Set a breakpoint at the starting address, and launch the program.

@smallexample
$ (gdb) break *0x00401010
$ (gdb) run
@end smallexample

The program will stop at the given address.

@item
Set a breakpoint on a DLL subroutine.

@smallexample
(gdb) break ada_dll.adb:45
@end smallexample

Or if you want to break using a symbol on the DLL, you need first to
select the Ada language (language used by the DLL).

@smallexample
(gdb) set language ada
(gdb) break ada_dll
@end smallexample

@item
Continue the program.

@smallexample
(gdb) cont
@end smallexample

@noindent
This will run the program until it reaches the breakpoint that has been
set. From that point you can use the standard way to debug a program
as described in (@pxref{Running and Debugging Ada Programs}).

@end enumerate

@noindent
It is also possible to debug the DLL by attaching to a running process.

@node Attaching to a Running Process
@subsubsection Attaching to a Running Process
@cindex DLL debugging, attach to process

@noindent
With @code{GDB} it is always possible to debug a running process by
attaching to it. It is possible to debug a DLL this way. The limitation
of this approach is that the DLL must run long enough to perform the
attach operation. It may be useful for instance to insert a time wasting
loop in the code of the DLL to meet this criterion.

@enumerate 1

@item Launch the main program @file{main.exe}.

@smallexample
$ main
@end smallexample

@item Use the Windows @i{Task Manager} to find the process ID. Let's say
that the process PID for @file{main.exe} is 208.

@item Launch gdb.

@smallexample
$ gdb
@end smallexample

@item Attach to the running process to be debugged.

@smallexample
(gdb) attach 208
@end smallexample

@item Load the process debugging information.

@smallexample
(gdb) symbol-file main.exe
@end smallexample

@item Break somewhere in the DLL.

@smallexample
(gdb) break ada_dll
@end smallexample

@item Continue process execution.

@smallexample
(gdb) cont
@end smallexample

@end enumerate

@noindent
This last step will resume the process execution, and stop at
the breakpoint we have set. From there you can use the standard
approach to debug a program as described in
(@pxref{Running and Debugging Ada Programs}).

@node Setting Stack Size from gnatlink
@section Setting Stack Size from @command{gnatlink}

@noindent
It is possible to specify the program stack size at link time. On modern
versions of Windows, starting with XP, this is mostly useful to set the size of
the main stack (environment task). The other task stacks are set with pragma
Storage_Size or with the @command{gnatbind -d} command.

Since older versions of Windows (2000, NT4, etc.) do not allow setting the
reserve size of individual tasks, the link-time stack size applies to all
tasks, and pragma Storage_Size has no effect.
In particular, Stack Overflow checks are made against this
link-time specified size.

This setting can be done with
@command{gnatlink} using either:

@itemize @bullet

@item using @option{-Xlinker} linker option

@smallexample
$ gnatlink hello -Xlinker --stack=0x10000,0x1000
@end smallexample

This sets the stack reserve size to 0x10000 bytes and the stack commit
size to 0x1000 bytes.

@item using @option{-Wl} linker option

@smallexample
$ gnatlink hello -Wl,--stack=0x1000000
@end smallexample

This sets the stack reserve size to 0x1000000 bytes. Note that with
@option{-Wl} option it is not possible to set the stack commit size
because the coma is a separator for this option.

@end itemize

@node Setting Heap Size from gnatlink
@section Setting Heap Size from @command{gnatlink}

@noindent
Under Windows systems, it is possible to specify the program heap size from
@command{gnatlink} using either:

@itemize @bullet

@item using @option{-Xlinker} linker option

@smallexample
$ gnatlink hello -Xlinker --heap=0x10000,0x1000
@end smallexample

This sets the heap reserve size to 0x10000 bytes and the heap commit
size to 0x1000 bytes.

@item using @option{-Wl} linker option

@smallexample
$ gnatlink hello -Wl,--heap=0x1000000
@end smallexample

This sets the heap reserve size to 0x1000000 bytes. Note that with
@option{-Wl} option it is not possible to set the heap commit size
because the coma is a separator for this option.

@end itemize

@node Mac OS Topics
@appendix Mac OS Topics
@cindex OS X

@noindent
This chapter describes topics that are specific to Apple's OS X
platform.

@menu
* Codesigning the Debugger::
@end menu

@node Codesigning the Debugger
@section Codesigning the Debugger

@noindent
The Darwin Kernel requires the debugger to have special permissions
before it is allowed to control other processes. These permissions
are granted by codesigning the GDB executable. Without these
permissions, the debugger will report error messages such as:

@smallexample
Starting program: /x/y/foo
Unable to find Mach task port for process-id 28885: (os/kern) failure (0x5).
 (please check gdb is codesigned - see taskgated(8))
@end smallexample

Codesigning requires a certificate.  The following procedure explains
how to create one:

@itemize @bullet
@item Start the Keychain Access application (in
/Applications/Utilities/Keychain Access.app)

@item Select the Keychain Access -> Certificate Assistant ->
Create a Certificate... menu

@item Then:

@itemize @bullet
@item Choose a name for the new certificate (this procedure will use
"gdb-cert" as an example)

@item Set "Identity Type" to "Self Signed Root"

@item Set "Certificate Type" to "Code Signing"

@item Activate the "Let me override defaults" option

@end itemize

@item Click several times on "Continue" until the "Specify a Location
For The Certificate" screen appears, then set "Keychain" to "System"

@item Click on "Continue" until the certificate is created

@item Finally, in the view, double-click on the new certificate,
and set "When using this certificate" to "Always Trust"

@item Exit the Keychain Access application and restart the computer
(this is unfortunately required)

@end itemize

Once a certificate has been created, the debugger can be codesigned
as follow. In a Terminal, run the following command...

@smallexample
codesign -f -s  "gdb-cert"  <gnat_install_prefix>/bin/gdb
@end smallexample

... where "gdb-cert" should be replaced by the actual certificate
name chosen above, and <gnat_install_prefix> should be replaced by
the location where you installed GNAT.  Also, be sure that users are
in the Unix group @samp{_developer}.

@c **********************************
@c * GNU Free Documentation License *
@c **********************************
@include fdl.texi
@c GNU Free Documentation License

@node Index
@unnumbered Index

@printindex cp

@contents
@c Put table of contents at end, otherwise it precedes the "title page" in
@c the .txt version
@c Edit the pdf file to move the contents to the beginning, after the title
@c page

@bye