* gcc_release (build_sources): Use two new variables EXPORTTAG and

[official-gcc.git] / gcc / doc / cpp.texi

\input texinfo
@setfilename cpp.info
@settitle The C Preprocessor
@setchapternewpage off
@c @smallbook
@c @cropmarks
@c @finalout

@copying
@c man begin COPYRIGHT
Copyright @copyright{} 1987, 1989, 1991, 1992, 1993, 1994, 1995, 1996,
1997, 1998, 1999, 2000, 2001, 2002, 2003
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 1.1 or
any later version published by the Free Software Foundation.  A copy of
the license is included in the
@c man end
section entitled ``GNU Free Documentation License''.
@ignore
@c man begin COPYRIGHT
man page gfdl(7).
@c man end
@end ignore

@c man begin COPYRIGHT
This manual contains no Invariant Sections.  The Front-Cover Texts are
(a) (see below), and the Back-Cover Texts are (b) (see below).

(a) The FSF's Front-Cover Text is:

     A GNU Manual

(b) The FSF's Back-Cover Text is:

     You have freedom to copy and modify this GNU Manual, like GNU
     software.  Copies published by the Free Software Foundation raise
     funds for GNU development.
@c man end
@end copying

@macro gcctabopt{body}
@code{\body\}
@end macro

@c Create a separate index for command line options.
@defcodeindex op
@syncodeindex vr op

@c Used in cppopts.texi and cppenv.texi.
@set cppmanual

@ifinfo
@dircategory Programming
@direntry
* Cpp: (cpp).                  The GNU C preprocessor.
@end direntry
@end ifinfo

@titlepage
@title The C Preprocessor
@subtitle Last revised April 2001
@subtitle for GCC version 3
@author Richard M. Stallman
@author Zachary Weinberg
@page
@c There is a fill at the bottom of the page, so we need a filll to
@c override it.
@vskip 0pt plus 1filll
@insertcopying
@end titlepage
@contents
@page

@ifnottex
@node Top
@top
The C preprocessor implements the macro language used to transform C,
C++, and Objective-C programs before they are compiled.  It can also be
useful on its own.

@menu
* Overview::
* Header Files::
* Macros::
* Conditionals::
* Diagnostics::
* Line Control::
* Pragmas::
* Other Directives::
* Preprocessor Output::
* Traditional Mode::
* Implementation Details::
* Invocation::
* Environment Variables::
* GNU Free Documentation License::
* Index of Directives::
* Option Index::
* Concept Index::

@detailmenu
 --- The Detailed Node Listing ---

Overview

* Character sets::
* Initial processing::
* Tokenization::
* The preprocessing language::

Header Files

* Include Syntax::
* Include Operation::
* Search Path::
* Once-Only Headers::
* Computed Includes::
* Wrapper Headers::
* System Headers::

Macros

* Object-like Macros::
* Function-like Macros::
* Macro Arguments::
* Stringification::
* Concatenation::
* Variadic Macros::
* Predefined Macros::
* Undefining and Redefining Macros::
* Directives Within Macro Arguments::
* Macro Pitfalls::

Predefined Macros

* Standard Predefined Macros::
* Common Predefined Macros::
* System-specific Predefined Macros::
* C++ Named Operators::

Macro Pitfalls

* Misnesting::
* Operator Precedence Problems::
* Swallowing the Semicolon::
* Duplication of Side Effects::
* Self-Referential Macros::
* Argument Prescan::
* Newlines in Arguments::

Conditionals

* Conditional Uses::
* Conditional Syntax::
* Deleted Code::

Conditional Syntax

* Ifdef::
* If::
* Defined::
* Else::
* Elif::

Implementation Details

* Implementation-defined behavior::
* Implementation limits::
* Obsolete Features::
* Differences from previous versions::

Obsolete Features

* Assertions::
* Obsolete once-only headers::

@end detailmenu
@end menu

@insertcopying
@end ifnottex

@node Overview
@chapter Overview
@c man begin DESCRIPTION
The C preprocessor, often known as @dfn{cpp}, is a @dfn{macro processor}
that is used automatically by the C compiler to transform your program
before compilation.  It is called a macro processor because it allows
you to define @dfn{macros}, which are brief abbreviations for longer
constructs.

The C preprocessor is intended to be used only with C, C++, and
Objective-C source code.  In the past, it has been abused as a general
text processor.  It will choke on input which does not obey C's lexical
rules.  For example, apostrophes will be interpreted as the beginning of
character constants, and cause errors.  Also, you cannot rely on it
preserving characteristics of the input which are not significant to
C-family languages.  If a Makefile is preprocessed, all the hard tabs
will be removed, and the Makefile will not work.

Having said that, you can often get away with using cpp on things which
are not C@.  Other Algol-ish programming languages are often safe
(Pascal, Ada, etc.) So is assembly, with caution.  @option{-traditional-cpp}
mode preserves more white space, and is otherwise more permissive.  Many
of the problems can be avoided by writing C or C++ style comments
instead of native language comments, and keeping macros simple.

Wherever possible, you should use a preprocessor geared to the language
you are writing in.  Modern versions of the GNU assembler have macro
facilities.  Most high level programming languages have their own
conditional compilation and inclusion mechanism.  If all else fails,
try a true general text processor, such as GNU M4.

C preprocessors vary in some details.  This manual discusses the GNU C
preprocessor, which provides a small superset of the features of ISO
Standard C@.  In its default mode, the GNU C preprocessor does not do a
few things required by the standard.  These are features which are
rarely, if ever, used, and may cause surprising changes to the meaning
of a program which does not expect them.  To get strict ISO Standard C,
you should use the @option{-std=c89} or @option{-std=c99} options, depending
on which version of the standard you want.  To get all the mandatory
diagnostics, you must also use @option{-pedantic}.  @xref{Invocation}.

This manual describes the behavior of the ISO preprocessor.  To
minimize gratuitous differences, where the ISO preprocessor's
behavior does not conflict with traditional semantics, the
traditional preprocessor should behave the same way.  The various
differences that do exist are detailed in the section @ref{Traditional
Mode}.

For clarity, unless noted otherwise, references to @samp{CPP} in this
manual refer to GNU CPP.
@c man end

@menu
* Character sets::
* Initial processing::
* Tokenization::
* The preprocessing language::
@end menu

@node Character sets
@section Character sets

Source code character set processing in C and related languages is
rather complicated.  The C standard discusses two character sets, but
there are really at least four.

The files input to CPP might be in any character set at all.  CPP's
very first action, before it even looks for line boundaries, is to
convert the file into the character set it uses for internal
processing.  That set is what the C standard calls the @dfn{source}
character set.  It must be isomorphic with ISO 10646, also known as
Unicode.  CPP uses the UTF-8 encoding of Unicode.

At present, GNU CPP does not implement conversion from arbitrary file
encodings to the source character set.  Use of any encoding other than
plain ASCII or UTF-8, except in comments, will cause errors.  Use of
encodings that are not strict supersets of ASCII, such as Shift JIS,
may cause errors even if non-ASCII characters appear only in comments. 
We plan to fix this in the near future.

All preprocessing work (the subject of the rest of this manual) is
carried out in the source character set.  If you request textual
output from the preprocessor with the @option{-E} option, it will be
in UTF-8.

After preprocessing is complete, string and character constants are
converted again, into the @dfn{execution} character set.  This
character set is under control of the user; the default is UTF-8,
matching the source character set.  Wide string and character
constants have their own character set, which is not called out
specifically in the standard.  Again, it is under control of the user.
The default is UTF-16 or UTF-32, whichever fits in the target's
@code{wchar_t} type, in the target machine's byte
order.@footnote{UTF-16 does not meet the requirements of the C
standard for a wide character set, but the choice of 16-bit
@code{wchar_t} is enshrined in some system ABIs so we cannot fix
this.}  Octal and hexadecimal escape sequences do not undergo
conversion; @t{'\x12'} has the value 0x12 regardless of the currently
selected execution character set.  All other escapes are replaced by
the character in the source character set that they represent, then
converted to the execution character set, just like unescaped
characters.

GCC does not permit the use of characters outside the ASCII range, nor
@samp{\u} and @samp{\U} escapes, in identifiers.  We hope this will
change eventually, but there are problems with the standard semantics
of such ``extended identifiers'' which must be resolved through the
ISO C and C++ committees first.

@node Initial processing
@section Initial processing

The preprocessor performs a series of textual transformations on its
input.  These happen before all other processing.  Conceptually, they
happen in a rigid order, and the entire file is run through each
transformation before the next one begins.  CPP actually does them
all at once, for performance reasons.  These transformations correspond
roughly to the first three ``phases of translation'' described in the C
standard.

@enumerate
@item
@cindex line endings
The input file is read into memory and broken into lines.

Different systems use different conventions to indicate the end of a
line.  GCC accepts the ASCII control sequences @kbd{LF}, @kbd{@w{CR
LF}} and @kbd{CR} as end-of-line markers.  These are the canonical
sequences used by Unix, DOS and VMS, and the classic Mac OS (before
OSX) respectively.  You may therefore safely copy source code written
on any of those systems to a different one and use it without
conversion.  (GCC may lose track of the current line number if a file
doesn't consistently use one convention, as sometimes happens when it
is edited on computers with different conventions that share a network
file system.)

If the last line of any input file lacks an end-of-line marker, the end
of the file is considered to implicitly supply one.  The C standard says
that this condition provokes undefined behavior, so GCC will emit a
warning message.

@item
@cindex trigraphs
@anchor{trigraphs}If trigraphs are enabled, they are replaced by their
corresponding single characters.  By default GCC ignores trigraphs,
but if you request a strictly conforming mode with the @option{-std}
option, or you specify the @option{-trigraphs} option, then it
converts them.

These are nine three-character sequences, all starting with @samp{??},
that are defined by ISO C to stand for single characters.  They permit
obsolete systems that lack some of C's punctuation to use C@.  For
example, @samp{??/} stands for @samp{\}, so @t{'??/n'} is a character
constant for a newline.

Trigraphs are not popular and many compilers implement them
incorrectly.  Portable code should not rely on trigraphs being either
converted or ignored.  With @option{-Wtrigraphs} GCC will warn you
when a trigraph may change the meaning of your program if it were
converted.  @xref{Wtrigraphs}.

In a string constant, you can prevent a sequence of question marks
from being confused with a trigraph by inserting a backslash between
the question marks, or by separating the string literal at the
trigraph and making use of string literal concatenation.  @t{"(??\?)"}
is the string @samp{(???)}, not @samp{(?]}.  Traditional C compilers
do not recognize these idioms.

The nine trigraphs and their replacements are

@example
Trigraph:       ??(  ??)  ??<  ??>  ??=  ??/  ??'  ??!  ??-
Replacement:      [    ]    @{    @}    #    \    ^    |    ~
@end example

@item
@cindex continued lines
@cindex backslash-newline
Continued lines are merged into one long line.

A continued line is a line which ends with a backslash, @samp{\}.  The
backslash is removed and the following line is joined with the current
one.  No space is inserted, so you may split a line anywhere, even in
the middle of a word.  (It is generally more readable to split lines
only at white space.)

The trailing backslash on a continued line is commonly referred to as a
@dfn{backslash-newline}.

If there is white space between a backslash and the end of a line, that
is still a continued line.  However, as this is usually the result of an
editing mistake, and many compilers will not accept it as a continued
line, GCC will warn you about it.

@item
@cindex comments
@cindex line comments
@cindex block comments
All comments are replaced with single spaces.

There are two kinds of comments.  @dfn{Block comments} begin with
@samp{/*} and continue until the next @samp{*/}.  Block comments do not
nest:

@example
/* @r{this is} /* @r{one comment} */ @r{text outside comment}
@end example

@dfn{Line comments} begin with @samp{//} and continue to the end of the
current line.  Line comments do not nest either, but it does not matter,
because they would end in the same place anyway.

@example
// @r{this is} // @r{one comment}
@r{text outside comment}
@end example
@end enumerate

It is safe to put line comments inside block comments, or vice versa.

@example
@group
/* @r{block comment}
   // @r{contains line comment}
   @r{yet more comment}
 */ @r{outside comment}

// @r{line comment} /* @r{contains block comment} */
@end group
@end example

But beware of commenting out one end of a block comment with a line
comment.

@example
@group
 // @r{l.c.}  /* @r{block comment begins}
    @r{oops! this isn't a comment anymore} */
@end group
@end example

Comments are not recognized within string literals.  
@t{@w{"/* blah */"}} is the string constant @samp{@w{/* blah */}}, not
an empty string.

Line comments are not in the 1989 edition of the C standard, but they
are recognized by GCC as an extension.  In C++ and in the 1999 edition
of the C standard, they are an official part of the language.

Since these transformations happen before all other processing, you can
split a line mechanically with backslash-newline anywhere.  You can
comment out the end of a line.  You can continue a line comment onto the
next line with backslash-newline.  You can even split @samp{/*},
@samp{*/}, and @samp{//} onto multiple lines with backslash-newline.
For example:

@example
@group
/\
*
*/ # /*
*/ defi\
ne FO\
O 10\
20
@end group
@end example

@noindent
is equivalent to @code{@w{#define FOO 1020}}.  All these tricks are
extremely confusing and should not be used in code intended to be
readable.

There is no way to prevent a backslash at the end of a line from being
interpreted as a backslash-newline.  This cannot affect any correct
program, however.

@node Tokenization
@section Tokenization

@cindex tokens
@cindex preprocessing tokens
After the textual transformations are finished, the input file is
converted into a sequence of @dfn{preprocessing tokens}.  These mostly
correspond to the syntactic tokens used by the C compiler, but there are
a few differences.  White space separates tokens; it is not itself a
token of any kind.  Tokens do not have to be separated by white space,
but it is often necessary to avoid ambiguities.

When faced with a sequence of characters that has more than one possible
tokenization, the preprocessor is greedy.  It always makes each token,
starting from the left, as big as possible before moving on to the next
token.  For instance, @code{a+++++b} is interpreted as
@code{@w{a ++ ++ + b}}, not as @code{@w{a ++ + ++ b}}, even though the
latter tokenization could be part of a valid C program and the former
could not.

Once the input file is broken into tokens, the token boundaries never
change, except when the @samp{##} preprocessing operator is used to paste
tokens together.  @xref{Concatenation}.  For example,

@example
@group
#define foo() bar
foo()baz
     @expansion{} bar baz
@emph{not}
     @expansion{} barbaz
@end group
@end example

The compiler does not re-tokenize the preprocessor's output.  Each
preprocessing token becomes one compiler token.

@cindex identifiers
Preprocessing tokens fall into five broad classes: identifiers,
preprocessing numbers, string literals, punctuators, and other.  An
@dfn{identifier} is the same as an identifier in C: any sequence of
letters, digits, or underscores, which begins with a letter or
underscore.  Keywords of C have no significance to the preprocessor;
they are ordinary identifiers.  You can define a macro whose name is a
keyword, for instance.  The only identifier which can be considered a
preprocessing keyword is @code{defined}.  @xref{Defined}.

This is mostly true of other languages which use the C preprocessor.
However, a few of the keywords of C++ are significant even in the
preprocessor.  @xref{C++ Named Operators}.

In the 1999 C standard, identifiers may contain letters which are not
part of the ``basic source character set,'' at the implementation's
discretion (such as accented Latin letters, Greek letters, or Chinese
ideograms).  This may be done with an extended character set, or the
@samp{\u} and @samp{\U} escape sequences.  GCC does not presently
implement either feature in the preprocessor or the compiler.

As an extension, GCC treats @samp{$} as a letter.  This is for
compatibility with some systems, such as VMS, where @samp{$} is commonly
used in system-defined function and object names.  @samp{$} is not a
letter in strictly conforming mode, or if you specify the @option{-$}
option.  @xref{Invocation}.

@cindex numbers
@cindex preprocessing numbers
A @dfn{preprocessing number} has a rather bizarre definition.  The
category includes all the normal integer and floating point constants
one expects of C, but also a number of other things one might not
initially recognize as a number.  Formally, preprocessing numbers begin
with an optional period, a required decimal digit, and then continue
with any sequence of letters, digits, underscores, periods, and
exponents.  Exponents are the two-character sequences @samp{e+},
@samp{e-}, @samp{E+}, @samp{E-}, @samp{p+}, @samp{p-}, @samp{P+}, and
@samp{P-}.  (The exponents that begin with @samp{p} or @samp{P} are new
to C99.  They are used for hexadecimal floating-point constants.)

The purpose of this unusual definition is to isolate the preprocessor
from the full complexity of numeric constants.  It does not have to
distinguish between lexically valid and invalid floating-point numbers,
which is complicated.  The definition also permits you to split an
identifier at any position and get exactly two tokens, which can then be
pasted back together with the @samp{##} operator.

It's possible for preprocessing numbers to cause programs to be
misinterpreted.  For example, @code{0xE+12} is a preprocessing number
which does not translate to any valid numeric constant, therefore a
syntax error.  It does not mean @code{@w{0xE + 12}}, which is what you
might have intended.

@cindex string literals
@cindex string constants
@cindex character constants
@cindex header file names
@c the @: prevents makeinfo from turning '' into ".
@dfn{String literals} are string constants, character constants, and
header file names (the argument of @samp{#include}).@footnote{The C
standard uses the term @dfn{string literal} to refer only to what we are
calling @dfn{string constants}.}  String constants and character
constants are straightforward: @t{"@dots{}"} or @t{'@dots{}'}.  In
either case embedded quotes should be escaped with a backslash:
@t{'\'@:'} is the character constant for @samp{'}.  There is no limit on
the length of a character constant, but the value of a character
constant that contains more than one character is
implementation-defined.  @xref{Implementation Details}.

Header file names either look like string constants, @t{"@dots{}"}, or are
written with angle brackets instead, @t{<@dots{}>}.  In either case,
backslash is an ordinary character.  There is no way to escape the
closing quote or angle bracket.  The preprocessor looks for the header
file in different places depending on which form you use.  @xref{Include
Operation}.

No string literal may extend past the end of a line.  Older versions
of GCC accepted multi-line string constants.  You may use continued
lines instead, or string constant concatenation.  @xref{Differences
from previous versions}.

@cindex punctuators
@cindex digraphs
@cindex alternative tokens
@dfn{Punctuators} are all the usual bits of punctuation which are
meaningful to C and C++.  All but three of the punctuation characters in
ASCII are C punctuators.  The exceptions are @samp{@@}, @samp{$}, and
@samp{`}.  In addition, all the two- and three-character operators are
punctuators.  There are also six @dfn{digraphs}, which the C++ standard
calls @dfn{alternative tokens}, which are merely alternate ways to spell
other punctuators.  This is a second attempt to work around missing
punctuation in obsolete systems.  It has no negative side effects,
unlike trigraphs, but does not cover as much ground.  The digraphs and
their corresponding normal punctuators are:

@example
Digraph:        <%  %>  <:  :>  %:  %:%:
Punctuator:      @{   @}   [   ]   #    ##
@end example

@cindex other tokens
Any other single character is considered ``other.'' It is passed on to
the preprocessor's output unmolested.  The C compiler will almost
certainly reject source code containing ``other'' tokens.  In ASCII, the
only other characters are @samp{@@}, @samp{$}, @samp{`}, and control
characters other than NUL (all bits zero).  (Note that @samp{$} is
normally considered a letter.)  All characters with the high bit set
(numeric range 0x7F--0xFF) are also ``other'' in the present
implementation.  This will change when proper support for international
character sets is added to GCC@.

NUL is a special case because of the high probability that its
appearance is accidental, and because it may be invisible to the user
(many terminals do not display NUL at all).  Within comments, NULs are
silently ignored, just as any other character would be.  In running
text, NUL is considered white space.  For example, these two directives
have the same meaning.

@example
#define X^@@1
#define X 1
@end example

@noindent
(where @samp{^@@} is ASCII NUL)@.  Within string or character constants,
NULs are preserved.  In the latter two cases the preprocessor emits a
warning message.

@node The preprocessing language
@section The preprocessing language
@cindex directives
@cindex preprocessing directives
@cindex directive line
@cindex directive name

After tokenization, the stream of tokens may simply be passed straight
to the compiler's parser.  However, if it contains any operations in the
@dfn{preprocessing language}, it will be transformed first.  This stage
corresponds roughly to the standard's ``translation phase 4'' and is
what most people think of as the preprocessor's job.

The preprocessing language consists of @dfn{directives} to be executed
and @dfn{macros} to be expanded.  Its primary capabilities are:

@itemize @bullet
@item
Inclusion of header files.  These are files of declarations that can be
substituted into your program.

@item
Macro expansion.  You can define @dfn{macros}, which are abbreviations
for arbitrary fragments of C code.  The preprocessor will replace the
macros with their definitions throughout the program.  Some macros are
automatically defined for you.

@item
Conditional compilation.  You can include or exclude parts of the
program according to various conditions.

@item
Line control.  If you use a program to combine or rearrange source files
into an intermediate file which is then compiled, you can use line
control to inform the compiler where each source line originally came
from.

@item
Diagnostics.  You can detect problems at compile time and issue errors
or warnings.
@end itemize

There are a few more, less useful, features.

Except for expansion of predefined macros, all these operations are
triggered with @dfn{preprocessing directives}.  Preprocessing directives
are lines in your program that start with @samp{#}.  Whitespace is
allowed before and after the @samp{#}.  The @samp{#} is followed by an
identifier, the @dfn{directive name}.  It specifies the operation to
perform.  Directives are commonly referred to as @samp{#@var{name}}
where @var{name} is the directive name.  For example, @samp{#define} is
the directive that defines a macro.

The @samp{#} which begins a directive cannot come from a macro
expansion.  Also, the directive name is not macro expanded.  Thus, if
@code{foo} is defined as a macro expanding to @code{define}, that does
not make @samp{#foo} a valid preprocessing directive.

The set of valid directive names is fixed.  Programs cannot define new
preprocessing directives.

Some directives require arguments; these make up the rest of the
directive line and must be separated from the directive name by
whitespace.  For example, @samp{#define} must be followed by a macro
name and the intended expansion of the macro.

A preprocessing directive cannot cover more than one line.  The line
may, however, be continued with backslash-newline, or by a block comment
which extends past the end of the line.  In either case, when the
directive is processed, the continuations have already been merged with
the first line to make one long line.

@node Header Files
@chapter Header Files

@cindex header file
A header file is a file containing C declarations and macro definitions
(@pxref{Macros}) to be shared between several source files.  You request
the use of a header file in your program by @dfn{including} it, with the
C preprocessing directive @samp{#include}.

Header files serve two purposes.

@itemize @bullet
@item
@cindex system header files
System header files declare the interfaces to parts of the operating
system.  You include them in your program to supply the definitions and
declarations you need to invoke system calls and libraries.

@item
Your own header files contain declarations for interfaces between the
source files of your program.  Each time you have a group of related
declarations and macro definitions all or most of which are needed in
several different source files, it is a good idea to create a header
file for them.
@end itemize

Including a header file produces the same results as copying the header
file into each source file that needs it.  Such copying would be
time-consuming and error-prone.  With a header file, the related
declarations appear in only one place.  If they need to be changed, they
can be changed in one place, and programs that include the header file
will automatically use the new version when next recompiled.  The header
file eliminates the labor of finding and changing all the copies as well
as the risk that a failure to find one copy will result in
inconsistencies within a program.

In C, the usual convention is to give header files names that end with
@file{.h}.  It is most portable to use only letters, digits, dashes, and
underscores in header file names, and at most one dot.

@menu
* Include Syntax::
* Include Operation::
* Search Path::
* Once-Only Headers::
* Computed Includes::
* Wrapper Headers::
* System Headers::
@end menu

@node Include Syntax
@section Include Syntax

@findex #include
Both user and system header files are included using the preprocessing
directive @samp{#include}.  It has two variants:

@table @code
@item #include <@var{file}>
This variant is used for system header files.  It searches for a file
named @var{file} in a standard list of system directories.  You can prepend
directories to this list with the @option{-I} option (@pxref{Invocation}).

@item #include "@var{file}"
This variant is used for header files of your own program.  It searches
for a file named @var{file} first in the directory containing the
current file, then in the same directories used for @code{<@var{file}>}.
@end table

The argument of @samp{#include}, whether delimited with quote marks or
angle brackets, behaves like a string constant in that comments are not
recognized, and macro names are not expanded.  Thus, @code{@w{#include
<x/*y>}} specifies inclusion of a system header file named @file{x/*y}.

However, if backslashes occur within @var{file}, they are considered
ordinary text characters, not escape characters.  None of the character
escape sequences appropriate to string constants in C are processed.
Thus, @code{@w{#include "x\n\\y"}} specifies a filename containing three
backslashes.  (Some systems interpret @samp{\} as a pathname separator.
All of these also interpret @samp{/} the same way.  It is most portable
to use only @samp{/}.)

It is an error if there is anything (other than comments) on the line
after the file name.

@node Include Operation
@section Include Operation

The @samp{#include} directive works by directing the C preprocessor to
scan the specified file as input before continuing with the rest of the
current file.  The output from the preprocessor contains the output
already generated, followed by the output resulting from the included
file, followed by the output that comes from the text after the
@samp{#include} directive.  For example, if you have a header file
@file{header.h} as follows,

@example
char *test (void);
@end example

@noindent
and a main program called @file{program.c} that uses the header file,
like this,

@example
int x;
#include "header.h"

int
main (void)
@{
  puts (test ());
@}
@end example

@noindent
the compiler will see the same token stream as it would if
@file{program.c} read

@example
int x;
char *test (void);

int
main (void)
@{
  puts (test ());
@}
@end example

Included files are not limited to declarations and macro definitions;
those are merely the typical uses.  Any fragment of a C program can be
included from another file.  The include file could even contain the
beginning of a statement that is concluded in the containing file, or
the end of a statement that was started in the including file.  However,
an included file must consist of complete tokens.  Comments and string
literals which have not been closed by the end of an included file are
invalid.  For error recovery, they are considered to end at the end of
the file.

To avoid confusion, it is best if header files contain only complete
syntactic units---function declarations or definitions, type
declarations, etc.

The line following the @samp{#include} directive is always treated as a
separate line by the C preprocessor, even if the included file lacks a
final newline.

@node Search Path
@section Search Path

GCC looks in several different places for headers.  On a normal Unix
system, if you do not instruct it otherwise, it will look for headers
requested with @code{@w{#include <@var{file}>}} in:

@example
/usr/local/include
@var{libdir}/gcc/@var{target}/@var{version}/include
/usr/@var{target}/include
/usr/include
@end example

For C++ programs, it will also look in @file{/usr/include/g++-v3},
first.  In the above, @var{target} is the canonical name of the system
GCC was configured to compile code for; often but not always the same as
the canonical name of the system it runs on.  @var{version} is the
version of GCC in use.

You can add to this list with the @option{-I@var{dir}} command line
option.  All the directories named by @option{-I} are searched, in
left-to-right order, @emph{before} the default directories.  The only
exception is when @file{dir} is already searched by default.  In
this case, the option is ignored and the search order for system
directories remains unchanged.

Duplicate directories are removed from the quote and bracket search
chains before the two chains are merged to make the final search chain.
Thus, it is possible for a directory to occur twice in the final search
chain if it was specified in both the quote and bracket chains.

You can prevent GCC from searching any of the default directories with
the @option{-nostdinc} option.  This is useful when you are compiling an
operating system kernel or some other program that does not use the
standard C library facilities, or the standard C library itself.
@option{-I} options are not ignored as described above when
@option{-nostdinc} is in effect.

GCC looks for headers requested with @code{@w{#include "@var{file}"}}
first in the directory containing the current file, then in the same
places it would have looked for a header requested with angle brackets.
For example, if @file{/usr/include/sys/stat.h} contains
@code{@w{#include "types.h"}}, GCC looks for @file{types.h} first in
@file{/usr/include/sys}, then in its usual search path.

@samp{#line} (@pxref{Line Control}) does not change GCC's idea of the
directory containing the current file.

You may put @option{-I-} at any point in your list of @option{-I} options.
This has two effects.  First, directories appearing before the
@option{-I-} in the list are searched only for headers requested with
quote marks.  Directories after @option{-I-} are searched for all
headers.  Second, the directory containing the current file is not
searched for anything, unless it happens to be one of the directories
named by an @option{-I} switch.

@option{-I. -I-} is not the same as no @option{-I} options at all, and does
not cause the same behavior for @samp{<>} includes that @samp{""}
includes get with no special options.  @option{-I.} searches the
compiler's current working directory for header files.  That may or may
not be the same as the directory containing the current file.

If you need to look for headers in a directory named @file{-}, write
@option{-I./-}.

There are several more ways to adjust the header search path.  They are
generally less useful.  @xref{Invocation}.

@node Once-Only Headers
@section Once-Only Headers
@cindex repeated inclusion
@cindex including just once
@cindex wrapper @code{#ifndef}

If a header file happens to be included twice, the compiler will process
its contents twice.  This is very likely to cause an error, e.g.@: when the
compiler sees the same structure definition twice.  Even if it does not,
it will certainly waste time.

The standard way to prevent this is to enclose the entire real contents
of the file in a conditional, like this:

@example
@group
/* File foo.  */
#ifndef FILE_FOO_SEEN
#define FILE_FOO_SEEN

@var{the entire file}

#endif /* !FILE_FOO_SEEN */
@end group
@end example

This construct is commonly known as a @dfn{wrapper #ifndef}.
When the header is included again, the conditional will be false,
because @code{FILE_FOO_SEEN} is defined.  The preprocessor will skip
over the entire contents of the file, and the compiler will not see it
twice.

CPP optimizes even further.  It remembers when a header file has a
wrapper @samp{#ifndef}.  If a subsequent @samp{#include} specifies that
header, and the macro in the @samp{#ifndef} is still defined, it does
not bother to rescan the file at all.

You can put comments outside the wrapper.  They will not interfere with
this optimization.

@cindex controlling macro
@cindex guard macro
The macro @code{FILE_FOO_SEEN} is called the @dfn{controlling macro} or
@dfn{guard macro}.  In a user header file, the macro name should not
begin with @samp{_}.  In a system header file, it should begin with
@samp{__} to avoid conflicts with user programs.  In any kind of header
file, the macro name should contain the name of the file and some
additional text, to avoid conflicts with other header files.

@node Computed Includes
@section Computed Includes
@cindex computed includes
@cindex macros in include

Sometimes it is necessary to select one of several different header
files to be included into your program.  They might specify
configuration parameters to be used on different sorts of operating
systems, for instance.  You could do this with a series of conditionals,

@example
#if SYSTEM_1
# include "system_1.h"
#elif SYSTEM_2
# include "system_2.h"
#elif SYSTEM_3
@dots{}
#endif
@end example

That rapidly becomes tedious.  Instead, the preprocessor offers the
ability to use a macro for the header name.  This is called a
@dfn{computed include}.  Instead of writing a header name as the direct
argument of @samp{#include}, you simply put a macro name there instead:

@example
#define SYSTEM_H "system_1.h"
@dots{}
#include SYSTEM_H
@end example

@noindent
@code{SYSTEM_H} will be expanded, and the preprocessor will look for
@file{system_1.h} as if the @samp{#include} had been written that way
originally.  @code{SYSTEM_H} could be defined by your Makefile with a
@option{-D} option.

You must be careful when you define the macro.  @samp{#define} saves
tokens, not text.  The preprocessor has no way of knowing that the macro
will be used as the argument of @samp{#include}, so it generates
ordinary tokens, not a header name.  This is unlikely to cause problems
if you use double-quote includes, which are close enough to string
constants.  If you use angle brackets, however, you may have trouble.

The syntax of a computed include is actually a bit more general than the
above.  If the first non-whitespace character after @samp{#include} is
not @samp{"} or @samp{<}, then the entire line is macro-expanded
like running text would be.

If the line expands to a single string constant, the contents of that
string constant are the file to be included.  CPP does not re-examine the
string for embedded quotes, but neither does it process backslash
escapes in the string.  Therefore

@example
#define HEADER "a\"b"
#include HEADER
@end example

@noindent
looks for a file named @file{a\"b}.  CPP searches for the file according
to the rules for double-quoted includes.

If the line expands to a token stream beginning with a @samp{<} token
and including a @samp{>} token, then the tokens between the @samp{<} and
the first @samp{>} are combined to form the filename to be included.
Any whitespace between tokens is reduced to a single space; then any
space after the initial @samp{<} is retained, but a trailing space
before the closing @samp{>} is ignored.  CPP searches for the file
according to the rules for angle-bracket includes.

In either case, if there are any tokens on the line after the file name,
an error occurs and the directive is not processed.  It is also an error
if the result of expansion does not match either of the two expected
forms.

These rules are implementation-defined behavior according to the C
standard.  To minimize the risk of different compilers interpreting your
computed includes differently, we recommend you use only a single
object-like macro which expands to a string constant.  This will also
minimize confusion for people reading your program.

@node Wrapper Headers
@section Wrapper Headers
@cindex wrapper headers
@cindex overriding a header file
@findex #include_next

Sometimes it is necessary to adjust the contents of a system-provided
header file without editing it directly.  GCC's @command{fixincludes}
operation does this, for example.  One way to do that would be to create
a new header file with the same name and insert it in the search path
before the original header.  That works fine as long as you're willing
to replace the old header entirely.  But what if you want to refer to
the old header from the new one?

You cannot simply include the old header with @samp{#include}.  That
will start from the beginning, and find your new header again.  If your
header is not protected from multiple inclusion (@pxref{Once-Only
Headers}), it will recurse infinitely and cause a fatal error.

You could include the old header with an absolute pathname:
@example
#include "/usr/include/old-header.h"
@end example
@noindent
This works, but is not clean; should the system headers ever move, you
would have to edit the new headers to match.

There is no way to solve this problem within the C standard, but you can
use the GNU extension @samp{#include_next}.  It means, ``Include the
@emph{next} file with this name.''  This directive works like
@samp{#include} except in searching for the specified file: it starts
searching the list of header file directories @emph{after} the directory
in which the current file was found.

Suppose you specify @option{-I /usr/local/include}, and the list of
directories to search also includes @file{/usr/include}; and suppose
both directories contain @file{signal.h}.  Ordinary @code{@w{#include
<signal.h>}} finds the file under @file{/usr/local/include}.  If that
file contains @code{@w{#include_next <signal.h>}}, it starts searching
after that directory, and finds the file in @file{/usr/include}.

@samp{#include_next} does not distinguish between @code{<@var{file}>}
and @code{"@var{file}"} inclusion, nor does it check that the file you
specify has the same name as the current file.  It simply looks for the
file named, starting with the directory in the search path after the one
where the current file was found.

The use of @samp{#include_next} can lead to great confusion.  We
recommend it be used only when there is no other alternative.  In
particular, it should not be used in the headers belonging to a specific
program; it should be used only to make global corrections along the
lines of @command{fixincludes}.

@node System Headers
@section System Headers
@cindex system header files

The header files declaring interfaces to the operating system and
runtime libraries often cannot be written in strictly conforming C@.
Therefore, GCC gives code found in @dfn{system headers} special
treatment.  All warnings, other than those generated by @samp{#warning}
(@pxref{Diagnostics}), are suppressed while GCC is processing a system
header.  Macros defined in a system header are immune to a few warnings
wherever they are expanded.  This immunity is granted on an ad-hoc
basis, when we find that a warning generates lots of false positives
because of code in macros defined in system headers.

Normally, only the headers found in specific directories are considered
system headers.  These directories are determined when GCC is compiled.
There are, however, two ways to make normal headers into system headers.

The @option{-isystem} command line option adds its argument to the list of
directories to search for headers, just like @option{-I}.  Any headers
found in that directory will be considered system headers.

All directories named by @option{-isystem} are searched @emph{after} all
directories named by @option{-I}, no matter what their order was on the
command line.  If the same directory is named by both @option{-I} and
@option{-isystem}, the @option{-I} option is ignored.  GCC provides an
informative message when this occurs if @option{-v} is used.

@findex #pragma GCC system_header
There is also a directive, @code{@w{#pragma GCC system_header}}, which
tells GCC to consider the rest of the current include file a system
header, no matter where it was found.  Code that comes before the
@samp{#pragma} in the file will not be affected.  @code{@w{#pragma GCC
system_header}} has no effect in the primary source file.

On very old systems, some of the pre-defined system header directories
get even more special treatment.  GNU C++ considers code in headers
found in those directories to be surrounded by an @code{@w{extern "C"}}
block.  There is no way to request this behavior with a @samp{#pragma},
or from the command line.

@node Macros
@chapter Macros

A @dfn{macro} is a fragment of code which has been given a name.
Whenever the name is used, it is replaced by the contents of the macro.
There are two kinds of macros.  They differ mostly in what they look
like when they are used.  @dfn{Object-like} macros resemble data objects
when used, @dfn{function-like} macros resemble function calls.

You may define any valid identifier as a macro, even if it is a C
keyword.  The preprocessor does not know anything about keywords.  This
can be useful if you wish to hide a keyword such as @code{const} from an
older compiler that does not understand it.  However, the preprocessor
operator @code{defined} (@pxref{Defined}) can never be defined as a
macro, and C++'s named operators (@pxref{C++ Named Operators}) cannot be
macros when you are compiling C++.

@menu
* Object-like Macros::
* Function-like Macros::
* Macro Arguments::
* Stringification::
* Concatenation::
* Variadic Macros::
* Predefined Macros::
* Undefining and Redefining Macros::
* Directives Within Macro Arguments::
* Macro Pitfalls::
@end menu

@node Object-like Macros
@section Object-like Macros
@cindex object-like macro
@cindex symbolic constants
@cindex manifest constants

An @dfn{object-like macro} is a simple identifier which will be replaced
by a code fragment.  It is called object-like because it looks like a
data object in code that uses it.  They are most commonly used to give
symbolic names to numeric constants.

@findex #define
You create macros with the @samp{#define} directive.  @samp{#define} is
followed by the name of the macro and then the token sequence it should
be an abbreviation for, which is variously referred to as the macro's
@dfn{body}, @dfn{expansion} or @dfn{replacement list}.  For example,

@example
#define BUFFER_SIZE 1024
@end example

@noindent
defines a macro named @code{BUFFER_SIZE} as an abbreviation for the
token @code{1024}.  If somewhere after this @samp{#define} directive
there comes a C statement of the form

@example
foo = (char *) malloc (BUFFER_SIZE);
@end example

@noindent
then the C preprocessor will recognize and @dfn{expand} the macro
@code{BUFFER_SIZE}.  The C compiler will see the same tokens as it would
if you had written

@example
foo = (char *) malloc (1024);
@end example

By convention, macro names are written in uppercase.  Programs are
easier to read when it is possible to tell at a glance which names are
macros.

The macro's body ends at the end of the @samp{#define} line.  You may
continue the definition onto multiple lines, if necessary, using
backslash-newline.  When the macro is expanded, however, it will all
come out on one line.  For example,

@example
#define NUMBERS 1, \
                2, \
                3
int x[] = @{ NUMBERS @};
     @expansion{} int x[] = @{ 1, 2, 3 @};
@end example

@noindent
The most common visible consequence of this is surprising line numbers
in error messages.

There is no restriction on what can go in a macro body provided it
decomposes into valid preprocessing tokens.  Parentheses need not
balance, and the body need not resemble valid C code.  (If it does not,
you may get error messages from the C compiler when you use the macro.)

The C preprocessor scans your program sequentially.  Macro definitions
take effect at the place you write them.  Therefore, the following input
to the C preprocessor

@example
foo = X;
#define X 4
bar = X;
@end example

@noindent
produces

@example
foo = X;
bar = 4;
@end example

When the preprocessor expands a macro name, the macro's expansion
replaces the macro invocation, then the expansion is examined for more
macros to expand.  For example,

@example
@group
#define TABLESIZE BUFSIZE
#define BUFSIZE 1024
TABLESIZE
     @expansion{} BUFSIZE
     @expansion{} 1024
@end group
@end example

@noindent
@code{TABLESIZE} is expanded first to produce @code{BUFSIZE}, then that
macro is expanded to produce the final result, @code{1024}.

Notice that @code{BUFSIZE} was not defined when @code{TABLESIZE} was
defined.  The @samp{#define} for @code{TABLESIZE} uses exactly the
expansion you specify---in this case, @code{BUFSIZE}---and does not
check to see whether it too contains macro names.  Only when you
@emph{use} @code{TABLESIZE} is the result of its expansion scanned for
more macro names.

This makes a difference if you change the definition of @code{BUFSIZE}
at some point in the source file.  @code{TABLESIZE}, defined as shown,
will always expand using the definition of @code{BUFSIZE} that is
currently in effect:

@example
#define BUFSIZE 1020
#define TABLESIZE BUFSIZE
#undef BUFSIZE
#define BUFSIZE 37
@end example

@noindent
Now @code{TABLESIZE} expands (in two stages) to @code{37}.

If the expansion of a macro contains its own name, either directly or
via intermediate macros, it is not expanded again when the expansion is
examined for more macros.  This prevents infinite recursion.
@xref{Self-Referential Macros}, for the precise details.

@node Function-like Macros
@section Function-like Macros
@cindex function-like macros

You can also define macros whose use looks like a function call.  These
are called @dfn{function-like macros}.  To define a function-like macro,
you use the same @samp{#define} directive, but you put a pair of
parentheses immediately after the macro name.  For example,

@example
#define lang_init()  c_init()
lang_init()
     @expansion{} c_init()
@end example

A function-like macro is only expanded if its name appears with a pair
of parentheses after it.  If you write just the name, it is left alone.
This can be useful when you have a function and a macro of the same
name, and you wish to use the function sometimes.

@example
extern void foo(void);
#define foo() /* optimized inline version */
@dots{}
  foo();
  funcptr = foo;
@end example

Here the call to @code{foo()} will use the macro, but the function
pointer will get the address of the real function.  If the macro were to
be expanded, it would cause a syntax error.

If you put spaces between the macro name and the parentheses in the
macro definition, that does not define a function-like macro, it defines
an object-like macro whose expansion happens to begin with a pair of
parentheses.

@example
#define lang_init ()    c_init()
lang_init()
     @expansion{} () c_init()()
@end example

The first two pairs of parentheses in this expansion come from the
macro.  The third is the pair that was originally after the macro
invocation.  Since @code{lang_init} is an object-like macro, it does not
consume those parentheses.

@node Macro Arguments
@section Macro Arguments
@cindex arguments
@cindex macros with arguments
@cindex arguments in macro definitions

Function-like macros can take @dfn{arguments}, just like true functions.
To define a macro that uses arguments, you insert @dfn{parameters}
between the pair of parentheses in the macro definition that make the
macro function-like.  The parameters must be valid C identifiers,
separated by commas and optionally whitespace.

To invoke a macro that takes arguments, you write the name of the macro
followed by a list of @dfn{actual arguments} in parentheses, separated
by commas.  The invocation of the macro need not be restricted to a
single logical line---it can cross as many lines in the source file as
you wish.  The number of arguments you give must match the number of
parameters in the macro definition.  When the macro is expanded, each
use of a parameter in its body is replaced by the tokens of the
corresponding argument.  (You need not use all of the parameters in the
macro body.)

As an example, here is a macro that computes the minimum of two numeric
values, as it is defined in many C programs, and some uses.

@example
#define min(X, Y)  ((X) < (Y) ? (X) : (Y))
  x = min(a, b);          @expansion{}  x = ((a) < (b) ? (a) : (b));
  y = min(1, 2);          @expansion{}  y = ((1) < (2) ? (1) : (2));
  z = min(a + 28, *p);    @expansion{}  z = ((a + 28) < (*p) ? (a + 28) : (*p));
@end example

@noindent
(In this small example you can already see several of the dangers of
macro arguments.  @xref{Macro Pitfalls}, for detailed explanations.)

Leading and trailing whitespace in each argument is dropped, and all
whitespace between the tokens of an argument is reduced to a single
space.  Parentheses within each argument must balance; a comma within
such parentheses does not end the argument.  However, there is no
requirement for square brackets or braces to balance, and they do not
prevent a comma from separating arguments.  Thus,

@example
macro (array[x = y, x + 1])
@end example

@noindent
passes two arguments to @code{macro}: @code{array[x = y} and @code{x +
1]}.  If you want to supply @code{array[x = y, x + 1]} as an argument,
you can write it as @code{array[(x = y, x + 1)]}, which is equivalent C
code.

All arguments to a macro are completely macro-expanded before they are
substituted into the macro body.  After substitution, the complete text
is scanned again for macros to expand, including the arguments.  This rule
may seem strange, but it is carefully designed so you need not worry
about whether any function call is actually a macro invocation.  You can
run into trouble if you try to be too clever, though.  @xref{Argument
Prescan}, for detailed discussion.

For example, @code{min (min (a, b), c)} is first expanded to

@example
  min (((a) < (b) ? (a) : (b)), (c))
@end example

@noindent
and then to

@example
@group
((((a) < (b) ? (a) : (b))) < (c)
 ? (((a) < (b) ? (a) : (b)))
 : (c))
@end group
@end example

@noindent
(Line breaks shown here for clarity would not actually be generated.)

@cindex empty macro arguments
You can leave macro arguments empty; this is not an error to the
preprocessor (but many macros will then expand to invalid code).
You cannot leave out arguments entirely; if a macro takes two arguments,
there must be exactly one comma at the top level of its argument list.
Here are some silly examples using @code{min}:

@example
min(, b)        @expansion{} ((   ) < (b) ? (   ) : (b))
min(a, )        @expansion{} ((a  ) < ( ) ? (a  ) : ( ))
min(,)          @expansion{} ((   ) < ( ) ? (   ) : ( ))
min((,),)       @expansion{} (((,)) < ( ) ? ((,)) : ( ))

min()      @error{} macro "min" requires 2 arguments, but only 1 given
min(,,)    @error{} macro "min" passed 3 arguments, but takes just 2
@end example

Whitespace is not a preprocessing token, so if a macro @code{foo} takes
one argument, @code{@w{foo ()}} and @code{@w{foo ( )}} both supply it an
empty argument.  Previous GNU preprocessor implementations and
documentation were incorrect on this point, insisting that a
function-like macro that takes a single argument be passed a space if an
empty argument was required.

Macro parameters appearing inside string literals are not replaced by
their corresponding actual arguments.

@example
#define foo(x) x, "x"
foo(bar)        @expansion{} bar, "x"
@end example

@node Stringification
@section Stringification
@cindex stringification
@cindex @samp{#} operator

Sometimes you may want to convert a macro argument into a string
constant.  Parameters are not replaced inside string constants, but you
can use the @samp{#} preprocessing operator instead.  When a macro
parameter is used with a leading @samp{#}, the preprocessor replaces it
with the literal text of the actual argument, converted to a string
constant.  Unlike normal parameter replacement, the argument is not
macro-expanded first.  This is called @dfn{stringification}.

There is no way to combine an argument with surrounding text and
stringify it all together.  Instead, you can write a series of adjacent
string constants and stringified arguments.  The preprocessor will
replace the stringified arguments with string constants.  The C
compiler will then combine all the adjacent string constants into one
long string.

Here is an example of a macro definition that uses stringification:

@example
@group
#define WARN_IF(EXP) \
do @{ if (EXP) \
        fprintf (stderr, "Warning: " #EXP "\n"); @} \
while (0)
WARN_IF (x == 0);
     @expansion{} do @{ if (x == 0)
           fprintf (stderr, "Warning: " "x == 0" "\n"); @} while (0);
@end group
@end example

@noindent
The argument for @code{EXP} is substituted once, as-is, into the
@code{if} statement, and once, stringified, into the argument to
@code{fprintf}.  If @code{x} were a macro, it would be expanded in the
@code{if} statement, but not in the string.

The @code{do} and @code{while (0)} are a kludge to make it possible to
write @code{WARN_IF (@var{arg});}, which the resemblance of
@code{WARN_IF} to a function would make C programmers want to do; see
@ref{Swallowing the Semicolon}.

Stringification in C involves more than putting double-quote characters
around the fragment.  The preprocessor backslash-escapes the quotes
surrounding embedded string constants, and all backslashes within string and
character constants, in order to get a valid C string constant with the
proper contents.  Thus, stringifying @code{@w{p = "foo\n";}} results in
@t{@w{"p = \"foo\\n\";"}}.  However, backslashes that are not inside string
or character constants are not duplicated: @samp{\n} by itself
stringifies to @t{"\n"}.

All leading and trailing whitespace in text being stringified is
ignored.  Any sequence of whitespace in the middle of the text is
converted to a single space in the stringified result.  Comments are
replaced by whitespace long before stringification happens, so they
never appear in stringified text.

There is no way to convert a macro argument into a character constant.

If you want to stringify the result of expansion of a macro argument,
you have to use two levels of macros.

@example
#define xstr(s) str(s)
#define str(s) #s
#define foo 4
str (foo)
     @expansion{} "foo"
xstr (foo)
     @expansion{} xstr (4)
     @expansion{} str (4)
     @expansion{} "4"
@end example

@code{s} is stringified when it is used in @code{str}, so it is not
macro-expanded first.  But @code{s} is an ordinary argument to
@code{xstr}, so it is completely macro-expanded before @code{xstr}
itself is expanded (@pxref{Argument Prescan}).  Therefore, by the time
@code{str} gets to its argument, it has already been macro-expanded.

@node Concatenation
@section Concatenation
@cindex concatenation
@cindex token pasting
@cindex token concatenation
@cindex @samp{##} operator

It is often useful to merge two tokens into one while expanding macros.
This is called @dfn{token pasting} or @dfn{token concatenation}.  The
@samp{##} preprocessing operator performs token pasting.  When a macro
is expanded, the two tokens on either side of each @samp{##} operator
are combined into a single token, which then replaces the @samp{##} and
the two original tokens in the macro expansion.  Usually both will be
identifiers, or one will be an identifier and the other a preprocessing
number.  When pasted, they make a longer identifier.  This isn't the
only valid case.  It is also possible to concatenate two numbers (or a
number and a name, such as @code{1.5} and @code{e3}) into a number.
Also, multi-character operators such as @code{+=} can be formed by
token pasting.

However, two tokens that don't together form a valid token cannot be
pasted together.  For example, you cannot concatenate @code{x} with
@code{+} in either order.  If you try, the preprocessor issues a warning
and emits the two tokens.  Whether it puts white space between the
tokens is undefined.  It is common to find unnecessary uses of @samp{##}
in complex macros.  If you get this warning, it is likely that you can
simply remove the @samp{##}.

Both the tokens combined by @samp{##} could come from the macro body,
but you could just as well write them as one token in the first place.
Token pasting is most useful when one or both of the tokens comes from a
macro argument.  If either of the tokens next to an @samp{##} is a
parameter name, it is replaced by its actual argument before @samp{##}
executes.  As with stringification, the actual argument is not
macro-expanded first.  If the argument is empty, that @samp{##} has no
effect.

Keep in mind that the C preprocessor converts comments to whitespace
before macros are even considered.  Therefore, you cannot create a
comment by concatenating @samp{/} and @samp{*}.  You can put as much
whitespace between @samp{##} and its operands as you like, including
comments, and you can put comments in arguments that will be
concatenated.  However, it is an error if @samp{##} appears at either
end of a macro body.

Consider a C program that interprets named commands.  There probably
needs to be a table of commands, perhaps an array of structures declared
as follows:

@example
@group
struct command
@{
  char *name;
  void (*function) (void);
@};
@end group

@group
struct command commands[] =
@{
  @{ "quit", quit_command @},
  @{ "help", help_command @},
  @dots{}
@};
@end group
@end example

It would be cleaner not to have to give each command name twice, once in
the string constant and once in the function name.  A macro which takes the
name of a command as an argument can make this unnecessary.  The string
constant can be created with stringification, and the function name by
concatenating the argument with @samp{_command}.  Here is how it is done:

@example
#define COMMAND(NAME)  @{ #NAME, NAME ## _command @}

struct command commands[] =
@{
  COMMAND (quit),
  COMMAND (help),
  @dots{}
@};
@end example

@node Variadic Macros
@section Variadic Macros
@cindex variable number of arguments
@cindex macros with variable arguments
@cindex variadic macros

A macro can be declared to accept a variable number of arguments much as
a function can.  The syntax for defining the macro is similar to that of
a function.  Here is an example:

@example
#define eprintf(@dots{}) fprintf (stderr, __VA_ARGS__)
@end example

This kind of macro is called @dfn{variadic}.  When the macro is invoked,
all the tokens in its argument list after the last named argument (this
macro has none), including any commas, become the @dfn{variable
argument}.  This sequence of tokens replaces the identifier
@code{@w{__VA_ARGS__}} in the macro body wherever it appears.  Thus, we
have this expansion:

@example
eprintf ("%s:%d: ", input_file, lineno)
     @expansion{}  fprintf (stderr, "%s:%d: ", input_file, lineno)
@end example

The variable argument is completely macro-expanded before it is inserted
into the macro expansion, just like an ordinary argument.  You may use
the @samp{#} and @samp{##} operators to stringify the variable argument
or to paste its leading or trailing token with another token.  (But see
below for an important special case for @samp{##}.)

If your macro is complicated, you may want a more descriptive name for
the variable argument than @code{@w{__VA_ARGS__}}.  CPP permits
this, as an extension.  You may write an argument name immediately
before the @samp{@dots{}}; that name is used for the variable argument.
The @code{eprintf} macro above could be written

@example
#define eprintf(args@dots{}) fprintf (stderr, args)
@end example

@noindent
using this extension.  You cannot use @code{@w{__VA_ARGS__}} and this
extension in the same macro.

You can have named arguments as well as variable arguments in a variadic
macro.  We could define @code{eprintf} like this, instead:

@example
#define eprintf(format, @dots{}) fprintf (stderr, format, __VA_ARGS__)
@end example

@noindent
This formulation looks more descriptive, but unfortunately it is less
flexible: you must now supply at least one argument after the format
string.  In standard C, you cannot omit the comma separating the named
argument from the variable arguments.  Furthermore, if you leave the
variable argument empty, you will get a syntax error, because
there will be an extra comma after the format string.

@example
eprintf("success!\n", );
     @expansion{} fprintf(stderr, "success!\n", );
@end example

GNU CPP has a pair of extensions which deal with this problem.  First,
you are allowed to leave the variable argument out entirely:

@example
eprintf ("success!\n")
     @expansion{} fprintf(stderr, "success!\n", );
@end example

@noindent
Second, the @samp{##} token paste operator has a special meaning when
placed between a comma and a variable argument.  If you write

@example
#define eprintf(format, @dots{}) fprintf (stderr, format, ##__VA_ARGS__)
@end example

@noindent
and the variable argument is left out when the @code{eprintf} macro is
used, then the comma before the @samp{##} will be deleted.  This does
@emph{not} happen if you pass an empty argument, nor does it happen if
the token preceding @samp{##} is anything other than a comma.

@example
eprintf ("success!\n")
     @expansion{} fprintf(stderr, "success!\n");
@end example

@noindent
The above explanation is ambiguous about the case where the only macro
parameter is a variable arguments parameter, as it is meaningless to
try to distinguish whether no argument at all is an empty argument or
a missing argument.  In this case the C99 standard is clear that the
comma must remain, however the existing GCC extension used to swallow
the comma.  So CPP retains the comma when conforming to a specific C
standard, and drops it otherwise.

C99 mandates that the only place the identifier @code{@w{__VA_ARGS__}}
can appear is in the replacement list of a variadic macro.  It may not
be used as a macro name, macro argument name, or within a different type
of macro.  It may also be forbidden in open text; the standard is
ambiguous.  We recommend you avoid using it except for its defined
purpose.

Variadic macros are a new feature in C99.  GNU CPP has supported them
for a long time, but only with a named variable argument
(@samp{args@dots{}}, not @samp{@dots{}} and @code{@w{__VA_ARGS__}}).  If you are
concerned with portability to previous versions of GCC, you should use
only named variable arguments.  On the other hand, if you are concerned
with portability to other conforming implementations of C99, you should
use only @code{@w{__VA_ARGS__}}.

Previous versions of CPP implemented the comma-deletion extension
much more generally.  We have restricted it in this release to minimize
the differences from C99.  To get the same effect with both this and
previous versions of GCC, the token preceding the special @samp{##} must
be a comma, and there must be white space between that comma and
whatever comes immediately before it:

@example
#define eprintf(format, args@dots{}) fprintf (stderr, format , ##args)
@end example

@noindent
@xref{Differences from previous versions}, for the gory details.

@node Predefined Macros
@section Predefined Macros

@cindex predefined macros
Several object-like macros are predefined; you use them without
supplying their definitions.  They fall into three classes: standard,
common, and system-specific.

In C++, there is a fourth category, the named operators.  They act like
predefined macros, but you cannot undefine them.

@menu
* Standard Predefined Macros::
* Common Predefined Macros::
* System-specific Predefined Macros::
* C++ Named Operators::
@end menu

@node Standard Predefined Macros
@subsection Standard Predefined Macros
@cindex standard predefined macros.

The standard predefined macros are specified by the relevant
language standards, so they are available with all compilers that
implement those standards.  Older compilers may not provide all of
them.  Their names all start with double underscores.

@table @code
@item __FILE__
This macro expands to the name of the current input file, in the form of
a C string constant.  This is the path by which the preprocessor opened
the file, not the short name specified in @samp{#include} or as the
input file name argument.  For example,
@code{"/usr/local/include/myheader.h"} is a possible expansion of this
macro.

@item __LINE__
This macro expands to the current input line number, in the form of a
decimal integer constant.  While we call it a predefined macro, it's
a pretty strange macro, since its ``definition'' changes with each
new line of source code.
@end table

@code{__FILE__} and @code{__LINE__} are useful in generating an error
message to report an inconsistency detected by the program; the message
can state the source line at which the inconsistency was detected.  For
example,

@example
fprintf (stderr, "Internal error: "
                 "negative string length "
                 "%d at %s, line %d.",
         length, __FILE__, __LINE__);
@end example

An @samp{#include} directive changes the expansions of @code{__FILE__}
and @code{__LINE__} to correspond to the included file.  At the end of
that file, when processing resumes on the input file that contained
the @samp{#include} directive, the expansions of @code{__FILE__} and
@code{__LINE__} revert to the values they had before the
@samp{#include} (but @code{__LINE__} is then incremented by one as
processing moves to the line after the @samp{#include}).

A @samp{#line} directive changes @code{__LINE__}, and may change
@code{__FILE__} as well.  @xref{Line Control}.

C99 introduces @code{__func__}, and GCC has provided @code{__FUNCTION__}
for a long time.  Both of these are strings containing the name of the
current function (there are slight semantic differences; see the GCC
manual).  Neither of them is a macro; the preprocessor does not know the
name of the current function.  They tend to be useful in conjunction
with @code{__FILE__} and @code{__LINE__}, though.

@table @code

@item __DATE__
This macro expands to a string constant that describes the date on which
the preprocessor is being run.  The string constant contains eleven
characters and looks like @code{@w{"Feb 12 1996"}}.  If the day of the
month is less than 10, it is padded with a space on the left.

If GCC cannot determine the current date, it will emit a warning message
(once per compilation) and @code{__DATE__} will expand to
@code{@w{"??? ?? ????"}}.

@item __TIME__
This macro expands to a string constant that describes the time at
which the preprocessor is being run.  The string constant contains
eight characters and looks like @code{"23:59:01"}.

If GCC cannot determine the current time, it will emit a warning message
(once per compilation) and @code{__TIME__} will expand to
@code{"??:??:??"}.

@item __STDC__
In normal operation, this macro expands to the constant 1, to signify
that this compiler conforms to ISO Standard C@.  If GNU CPP is used with
a compiler other than GCC, this is not necessarily true; however, the
preprocessor always conforms to the standard unless the
@option{-traditional-cpp} option is used.

This macro is not defined if the @option{-traditional-cpp} option is used.

On some hosts, the system compiler uses a different convention, where
@code{__STDC__} is normally 0, but is 1 if the user specifies strict
conformance to the C Standard.  CPP follows the host convention when
processing system header files, but when processing user files
@code{__STDC__} is always 1.  This has been reported to cause problems;
for instance, some versions of Solaris provide X Windows headers that
expect @code{__STDC__} to be either undefined or 1.  @xref{Invocation}.

@item __STDC_VERSION__
This macro expands to the C Standard's version number, a long integer
constant of the form @code{@var{yyyy}@var{mm}L} where @var{yyyy} and
@var{mm} are the year and month of the Standard version.  This signifies
which version of the C Standard the compiler conforms to.  Like
@code{__STDC__}, this is not necessarily accurate for the entire
implementation, unless GNU CPP is being used with GCC@.

The value @code{199409L} signifies the 1989 C standard as amended in
1994, which is the current default; the value @code{199901L} signifies
the 1999 revision of the C standard.  Support for the 1999 revision is
not yet complete.

This macro is not defined if the @option{-traditional-cpp} option is
used, nor when compiling C++ or Objective-C@.

@item __STDC_HOSTED__
This macro is defined, with value 1, if the compiler's target is a
@dfn{hosted environment}.  A hosted environment has the complete
facilities of the standard C library available.

@item __cplusplus
This macro is defined when the C++ compiler is in use.  You can use
@code{__cplusplus} to test whether a header is compiled by a C compiler
or a C++ compiler.  This macro is similar to @code{__STDC_VERSION__}, in
that it expands to a version number.  A fully conforming implementation
of the 1998 C++ standard will define this macro to @code{199711L}.  The
GNU C++ compiler is not yet fully conforming, so it uses @code{1}
instead.  We hope to complete our implementation in the near future.

@item __OBJC__
This macro is defined, with value 1, when the Objective-C compiler is in
use.  You can use @code{__OBJC__} to test whether a header is compiled
by a C compiler or a Objective-C compiler.

@item __ASSEMBLER__
This macro is defined with value 1 when preprocessing assembly
language.

@end table

@node Common Predefined Macros
@subsection Common Predefined Macros
@cindex common predefined macros

The common predefined macros are GNU C extensions.  They are available
with the same meanings regardless of the machine or operating system on
which you are using GNU C@.  Their names all start with double
underscores.

@table @code

@item __GNUC__
@itemx __GNUC_MINOR__
@itemx __GNUC_PATCHLEVEL__
These macros are defined by all GNU compilers that use the C
preprocessor: C, C++, and Objective-C@.  Their values are the major
version, minor version, and patch level of the compiler, as integer
constants.  For example, GCC 3.2.1 will define @code{__GNUC__} to 3,
@code{__GNUC_MINOR__} to 2, and @code{__GNUC_PATCHLEVEL__} to 1.  They
are defined only when the entire compiler is in use; if you invoke the
preprocessor directly, they are not defined.

@code{__GNUC_PATCHLEVEL__} is new to GCC 3.0; it is also present in the
widely-used development snapshots leading up to 3.0 (which identify
themselves as GCC 2.96 or 2.97, depending on which snapshot you have).

If all you need to know is whether or not your program is being compiled
by GCC, you can simply test @code{__GNUC__}.  If you need to write code
which depends on a specific version, you must be more careful.  Each
time the minor version is increased, the patch level is reset to zero;
each time the major version is increased (which happens rarely), the
minor version and patch level are reset.  If you wish to use the
predefined macros directly in the conditional, you will need to write it
like this:

@example
/* @r{Test for GCC > 3.2.0} */
#if __GNUC__ > 3 || \
    (__GNUC__ == 3 && (__GNUC_MINOR__ > 2 || \
                       (__GNUC_MINOR__ == 2 && \
                        __GNUC_PATCHLEVEL__ > 0))
@end example

@noindent
Another approach is to use the predefined macros to
calculate a single number, then compare that against a threshold:

@example
#define GCC_VERSION (__GNUC__ * 10000 \
                     + __GNUC_MINOR__ * 100 \
                     + __GNUC_PATCHLEVEL__)
@dots{}
/* @r{Test for GCC > 3.2.0} */
#if GCC_VERSION > 30200
@end example

@noindent
Many people find this form easier to understand.

@item __GNUG__
The GNU C++ compiler defines this.  Testing it is equivalent to
testing @code{@w{(__GNUC__ && __cplusplus)}}.

@item __STRICT_ANSI__
GCC defines this macro if and only if the @option{-ansi} switch, or a
@option{-std} switch specifying strict conformance to some version of ISO C,
was specified when GCC was invoked.  It is defined to @samp{1}.
This macro exists primarily to direct GNU libc's header files to
restrict their definitions to the minimal set found in the 1989 C
standard.

@item __BASE_FILE__
This macro expands to the name of the main input file, in the form
of a C string constant.  This is the source file that was specified
on the command line of the preprocessor or C compiler.

@item __INCLUDE_LEVEL__
This macro expands to a decimal integer constant that represents the
depth of nesting in include files.  The value of this macro is
incremented on every @samp{#include} directive and decremented at the
end of every included file.  It starts out at 0, it's value within the
base file specified on the command line.

@item __ELF__
This macro is defined if the target uses the ELF object format.

@item __VERSION__
This macro expands to a string constant which describes the version of
the compiler in use.  You should not rely on its contents having any
particular form, but it can be counted on to contain at least the
release number.

@item __OPTIMIZE__
@itemx __OPTIMIZE_SIZE__
@itemx __NO_INLINE__
These macros describe the compilation mode.  @code{__OPTIMIZE__} is
defined in all optimizing compilations.  @code{__OPTIMIZE_SIZE__} is
defined if the compiler is optimizing for size, not speed.
@code{__NO_INLINE__} is defined if no functions will be inlined into
their callers (when not optimizing, or when inlining has been
specifically disabled by @option{-fno-inline}).

These macros cause certain GNU header files to provide optimized
definitions, using macros or inline functions, of system library
functions.  You should not use these macros in any way unless you make
sure that programs will execute with the same effect whether or not they
are defined.  If they are defined, their value is 1.

@item __CHAR_UNSIGNED__
GCC defines this macro if and only if the data type @code{char} is
unsigned on the target machine.  It exists to cause the standard header
file @file{limits.h} to work correctly.  You should not use this macro
yourself; instead, refer to the standard macros defined in @file{limits.h}.

@item __WCHAR_UNSIGNED__
Like @code{__CHAR_UNSIGNED__}, this macro is defined if and only if the
data type @code{wchar_t} is unsigned and the front-end is in C++ mode.

@item __REGISTER_PREFIX__
This macro expands to a single token (not a string constant) which is
the prefix applied to CPU register names in assembly language for this
target.  You can use it to write assembly that is usable in multiple
environments.  For example, in the @code{m68k-aout} environment it
expands to nothing, but in the @code{m68k-coff} environment it expands
to a single @samp{%}.

@item __USER_LABEL_PREFIX__
This macro expands to a single token which is the prefix applied to
user labels (symbols visible to C code) in assembly.  For example, in
the @code{m68k-aout} environment it expands to an @samp{_}, but in the
@code{m68k-coff} environment it expands to nothing.

This macro will have the correct definition even if
@option{-f(no-)underscores} is in use, but it will not be correct if
target-specific options that adjust this prefix are used (e.g.@: the
OSF/rose @option{-mno-underscores} option).

@item __SIZE_TYPE__
@itemx __PTRDIFF_TYPE__
@itemx __WCHAR_TYPE__
@itemx __WINT_TYPE__
These macros are defined to the correct underlying types for the
@code{size_t}, @code{ptrdiff_t}, @code{wchar_t}, and @code{wint_t}
typedefs, respectively.  They exist to make the standard header files
@file{stddef.h} and @file{wchar.h} work correctly.  You should not use
these macros directly; instead, include the appropriate headers and use
the typedefs.

@item __CHAR_BIT__
Defined to the number of bits used in the representation of the
@code{char} data type.  It exists to make the standard header given
numerical limits work correctly.  You should not use
this macro directly; instead, include the appropriate headers.

@item __SCHAR_MAX__
@itemx __WCHAR_MAX__
@itemx __SHRT_MAX__
@itemx __INT_MAX__
@itemx __LONG_MAX__
@itemx __LONG_LONG_MAX__
Defined to the maximum value of the @code{signed char}, @code{wchar_t}, 
@code{signed short},
@code{signed int}, @code{signed long}, and @code{signed long long} types
respectively.  They exist to make the standard header given numerical limits
work correctly.  You should not use these macros directly; instead, include
the appropriate headers.

@item __USING_SJLJ_EXCEPTIONS__
This macro is defined, with value 1, if the compiler uses the old
mechanism based on @code{setjmp} and @code{longjmp} for exception
handling.

@item __NEXT_RUNTIME__
This macro is defined, with value 1, if (and only if) the NeXT runtime
(as in @option{-fnext-runtime}) is in use for Objective-C.  If the GNU
runtime is used, this macro is not defined, so that you can use this
macro to determine which runtime (NeXT or GNU) is being used.

@item __LP64__
@item _LP64
These macros are defined, with value 1, if (and only if) the compilation
is for a target where @code{long int} and pointer both use 64-bits and
@code{int} uses 32-bit.
@end table

@node System-specific Predefined Macros
@subsection System-specific Predefined Macros

@cindex system-specific predefined macros
@cindex predefined macros, system-specific
@cindex reserved namespace

The C preprocessor normally predefines several macros that indicate what
type of system and machine is in use.  They are obviously different on
each target supported by GCC@.  This manual, being for all systems and
machines, cannot tell you what their names are, but you can use
@command{cpp -dM} to see them all.  @xref{Invocation}.  All system-specific
predefined macros expand to the constant 1, so you can test them with
either @samp{#ifdef} or @samp{#if}.

The C standard requires that all system-specific macros be part of the
@dfn{reserved namespace}.  All names which begin with two underscores,
or an underscore and a capital letter, are reserved for the compiler and
library to use as they wish.  However, historically system-specific
macros have had names with no special prefix; for instance, it is common
to find @code{unix} defined on Unix systems.  For all such macros, GCC
provides a parallel macro with two underscores added at the beginning
and the end.  If @code{unix} is defined, @code{__unix__} will be defined
too.  There will never be more than two underscores; the parallel of
@code{_mips} is @code{__mips__}.

When the @option{-ansi} option, or any @option{-std} option that
requests strict conformance, is given to the compiler, all the
system-specific predefined macros outside the reserved namespace are
suppressed.  The parallel macros, inside the reserved namespace, remain
defined.

We are slowly phasing out all predefined macros which are outside the
reserved namespace.  You should never use them in new programs, and we
encourage you to correct older code to use the parallel macros whenever
you find it.  We don't recommend you use the system-specific macros that
are in the reserved namespace, either.  It is better in the long run to
check specifically for features you need, using a tool such as
@command{autoconf}.

@node C++ Named Operators
@subsection C++ Named Operators
@cindex named operators
@cindex C++ named operators
@cindex iso646.h

In C++, there are eleven keywords which are simply alternate spellings
of operators normally written with punctuation.  These keywords are
treated as such even in the preprocessor.  They function as operators in
@samp{#if}, and they cannot be defined as macros or poisoned.  In C, you
can request that those keywords take their C++ meaning by including
@file{iso646.h}.  That header defines each one as a normal object-like
macro expanding to the appropriate punctuator.

These are the named operators and their corresponding punctuators:

@multitable {Named Operator} {Punctuator}
@item Named Operator @tab Punctuator
@item @code{and}    @tab @code{&&}
@item @code{and_eq} @tab @code{&=}
@item @code{bitand} @tab @code{&}
@item @code{bitor}  @tab @code{|}
@item @code{compl}  @tab @code{~}
@item @code{not}    @tab @code{!}
@item @code{not_eq} @tab @code{!=}
@item @code{or}     @tab @code{||}
@item @code{or_eq}  @tab @code{|=}
@item @code{xor}    @tab @code{^}
@item @code{xor_eq} @tab @code{^=}
@end multitable

@node Undefining and Redefining Macros
@section Undefining and Redefining Macros
@cindex undefining macros
@cindex redefining macros
@findex #undef

If a macro ceases to be useful, it may be @dfn{undefined} with the
@samp{#undef} directive.  @samp{#undef} takes a single argument, the
name of the macro to undefine.  You use the bare macro name, even if the
macro is function-like.  It is an error if anything appears on the line
after the macro name.  @samp{#undef} has no effect if the name is not a
macro.

@example
#define FOO 4
x = FOO;        @expansion{} x = 4;
#undef FOO
x = FOO;        @expansion{} x = FOO;
@end example

Once a macro has been undefined, that identifier may be @dfn{redefined}
as a macro by a subsequent @samp{#define} directive.  The new definition
need not have any resemblance to the old definition.

However, if an identifier which is currently a macro is redefined, then
the new definition must be @dfn{effectively the same} as the old one.
Two macro definitions are effectively the same if:
@itemize @bullet
@item Both are the same type of macro (object- or function-like).
@item All the tokens of the replacement list are the same.
@item If there are any parameters, they are the same.
@item Whitespace appears in the same places in both.  It need not be
exactly the same amount of whitespace, though.  Remember that comments
count as whitespace.
@end itemize

@noindent
These definitions are effectively the same:
@example
#define FOUR (2 + 2)
#define FOUR         (2    +    2)
#define FOUR (2 /* two */ + 2)
@end example
@noindent
but these are not:
@example
#define FOUR (2 + 2)
#define FOUR ( 2+2 )
#define FOUR (2 * 2)
#define FOUR(score,and,seven,years,ago) (2 + 2)
@end example

If a macro is redefined with a definition that is not effectively the
same as the old one, the preprocessor issues a warning and changes the
macro to use the new definition.  If the new definition is effectively
the same, the redefinition is silently ignored.  This allows, for
instance, two different headers to define a common macro.  The
preprocessor will only complain if the definitions do not match.

@node Directives Within Macro Arguments
@section Directives Within Macro Arguments
@cindex macro arguments and directives

Occasionally it is convenient to use preprocessor directives within
the arguments of a macro.  The C and C++ standards declare that
behavior in these cases is undefined.

Versions of CPP prior to 3.2 would reject such constructs with an
error message.  This was the only syntactic difference between normal
functions and function-like macros, so it seemed attractive to remove
this limitation, and people would often be surprised that they could
not use macros in this way.  Moreover, sometimes people would use
conditional compilation in the argument list to a normal library
function like @samp{printf}, only to find that after a library upgrade
@samp{printf} had changed to be a function-like macro, and their code
would no longer compile.  So from version 3.2 we changed CPP to
successfully process arbitrary directives within macro arguments in
exactly the same way as it would have processed the directive were the
function-like macro invocation not present.

If, within a macro invocation, that macro is redefined, then the new
definition takes effect in time for argument pre-expansion, but the
original definition is still used for argument replacement.  Here is a
pathological example:

@smallexample
#define f(x) x x
f (1
#undef f
#define f 2
f)
@end smallexample

@noindent
which expands to

@smallexample
1 2 1 2
@end smallexample

@noindent
with the semantics described above.

@node Macro Pitfalls
@section Macro Pitfalls
@cindex problems with macros
@cindex pitfalls of macros

In this section we describe some special rules that apply to macros and
macro expansion, and point out certain cases in which the rules have
counter-intuitive consequences that you must watch out for.

@menu
* Misnesting::
* Operator Precedence Problems::
* Swallowing the Semicolon::
* Duplication of Side Effects::
* Self-Referential Macros::
* Argument Prescan::
* Newlines in Arguments::
@end menu

@node Misnesting
@subsection Misnesting

When a macro is called with arguments, the arguments are substituted
into the macro body and the result is checked, together with the rest of
the input file, for more macro calls.  It is possible to piece together
a macro call coming partially from the macro body and partially from the
arguments.  For example,

@example
#define twice(x) (2*(x))
#define call_with_1(x) x(1)
call_with_1 (twice)
     @expansion{} twice(1)
     @expansion{} (2*(1))
@end example

Macro definitions do not have to have balanced parentheses.  By writing
an unbalanced open parenthesis in a macro body, it is possible to create
a macro call that begins inside the macro body but ends outside of it.
For example,

@example
#define strange(file) fprintf (file, "%s %d",
@dots{}
strange(stderr) p, 35)
     @expansion{} fprintf (stderr, "%s %d", p, 35)
@end example

The ability to piece together a macro call can be useful, but the use of
unbalanced open parentheses in a macro body is just confusing, and
should be avoided.

@node Operator Precedence Problems
@subsection Operator Precedence Problems
@cindex parentheses in macro bodies

You may have noticed that in most of the macro definition examples shown
above, each occurrence of a macro argument name had parentheses around
it.  In addition, another pair of parentheses usually surround the
entire macro definition.  Here is why it is best to write macros that
way.

Suppose you define a macro as follows,

@example
#define ceil_div(x, y) (x + y - 1) / y
@end example

@noindent
whose purpose is to divide, rounding up.  (One use for this operation is
to compute how many @code{int} objects are needed to hold a certain
number of @code{char} objects.)  Then suppose it is used as follows:

@example
a = ceil_div (b & c, sizeof (int));
     @expansion{} a = (b & c + sizeof (int) - 1) / sizeof (int);
@end example

@noindent
This does not do what is intended.  The operator-precedence rules of
C make it equivalent to this:

@example
a = (b & (c + sizeof (int) - 1)) / sizeof (int);
@end example

@noindent
What we want is this:

@example
a = ((b & c) + sizeof (int) - 1)) / sizeof (int);
@end example

@noindent
Defining the macro as

@example
#define ceil_div(x, y) ((x) + (y) - 1) / (y)
@end example

@noindent
provides the desired result.

Unintended grouping can result in another way.  Consider @code{sizeof
ceil_div(1, 2)}.  That has the appearance of a C expression that would
compute the size of the type of @code{ceil_div (1, 2)}, but in fact it
means something very different.  Here is what it expands to:

@example
sizeof ((1) + (2) - 1) / (2)
@end example

@noindent
This would take the size of an integer and divide it by two.  The
precedence rules have put the division outside the @code{sizeof} when it
was intended to be inside.

Parentheses around the entire macro definition prevent such problems.
Here, then, is the recommended way to define @code{ceil_div}:

@example
#define ceil_div(x, y) (((x) + (y) - 1) / (y))
@end example

@node Swallowing the Semicolon
@subsection Swallowing the Semicolon
@cindex semicolons (after macro calls)

Often it is desirable to define a macro that expands into a compound
statement.  Consider, for example, the following macro, that advances a
pointer (the argument @code{p} says where to find it) across whitespace
characters:

@example
#define SKIP_SPACES(p, limit)  \
@{ char *lim = (limit);         \
  while (p < lim) @{            \
    if (*p++ != ' ') @{         \
      p--; break; @}@}@}
@end example

@noindent
Here backslash-newline is used to split the macro definition, which must
be a single logical line, so that it resembles the way such code would
be laid out if not part of a macro definition.

A call to this macro might be @code{SKIP_SPACES (p, lim)}.  Strictly
speaking, the call expands to a compound statement, which is a complete
statement with no need for a semicolon to end it.  However, since it
looks like a function call, it minimizes confusion if you can use it
like a function call, writing a semicolon afterward, as in
@code{SKIP_SPACES (p, lim);}

This can cause trouble before @code{else} statements, because the
semicolon is actually a null statement.  Suppose you write

@example
if (*p != 0)
  SKIP_SPACES (p, lim);
else @dots{}
@end example

@noindent
The presence of two statements---the compound statement and a null
statement---in between the @code{if} condition and the @code{else}
makes invalid C code.

The definition of the macro @code{SKIP_SPACES} can be altered to solve
this problem, using a @code{do @dots{} while} statement.  Here is how:

@example
#define SKIP_SPACES(p, limit)     \
do @{ char *lim = (limit);         \
     while (p < lim) @{            \
       if (*p++ != ' ') @{         \
         p--; break; @}@}@}          \
while (0)
@end example

Now @code{SKIP_SPACES (p, lim);} expands into

@example
do @{@dots{}@} while (0);
@end example

@noindent
which is one statement.  The loop executes exactly once; most compilers
generate no extra code for it.

@node Duplication of Side Effects
@subsection Duplication of Side Effects

@cindex side effects (in macro arguments)
@cindex unsafe macros
Many C programs define a macro @code{min}, for ``minimum'', like this:

@example
#define min(X, Y)  ((X) < (Y) ? (X) : (Y))
@end example

When you use this macro with an argument containing a side effect,
as shown here,

@example
next = min (x + y, foo (z));
@end example

@noindent
it expands as follows:

@example
next = ((x + y) < (foo (z)) ? (x + y) : (foo (z)));
@end example

@noindent
where @code{x + y} has been substituted for @code{X} and @code{foo (z)}
for @code{Y}.

The function @code{foo} is used only once in the statement as it appears
in the program, but the expression @code{foo (z)} has been substituted
twice into the macro expansion.  As a result, @code{foo} might be called
two times when the statement is executed.  If it has side effects or if
it takes a long time to compute, the results might not be what you
intended.  We say that @code{min} is an @dfn{unsafe} macro.

The best solution to this problem is to define @code{min} in a way that
computes the value of @code{foo (z)} only once.  The C language offers
no standard way to do this, but it can be done with GNU extensions as
follows:

@example
#define min(X, Y)                \
(@{ typeof (X) x_ = (X);          \
   typeof (Y) y_ = (Y);          \
   (x_ < y_) ? x_ : y_; @})
@end example

The @samp{(@{ @dots{} @})} notation produces a compound statement that
acts as an expression.  Its value is the value of its last statement.
This permits us to define local variables and assign each argument to
one.  The local variables have underscores after their names to reduce
the risk of conflict with an identifier of wider scope (it is impossible
to avoid this entirely).  Now each argument is evaluated exactly once.

If you do not wish to use GNU C extensions, the only solution is to be
careful when @emph{using} the macro @code{min}.  For example, you can
calculate the value of @code{foo (z)}, save it in a variable, and use
that variable in @code{min}:

@example
@group
#define min(X, Y)  ((X) < (Y) ? (X) : (Y))
@dots{}
@{
  int tem = foo (z);
  next = min (x + y, tem);
@}
@end group
@end example

@noindent
(where we assume that @code{foo} returns type @code{int}).

@node Self-Referential Macros
@subsection Self-Referential Macros
@cindex self-reference

A @dfn{self-referential} macro is one whose name appears in its
definition.  Recall that all macro definitions are rescanned for more
macros to replace.  If the self-reference were considered a use of the
macro, it would produce an infinitely large expansion.  To prevent this,
the self-reference is not considered a macro call.  It is passed into
the preprocessor output unchanged.  Let's consider an example:

@example
#define foo (4 + foo)
@end example

@noindent
where @code{foo} is also a variable in your program.

Following the ordinary rules, each reference to @code{foo} will expand
into @code{(4 + foo)}; then this will be rescanned and will expand into
@code{(4 + (4 + foo))}; and so on until the computer runs out of memory.

The self-reference rule cuts this process short after one step, at
@code{(4 + foo)}.  Therefore, this macro definition has the possibly
useful effect of causing the program to add 4 to the value of @code{foo}
wherever @code{foo} is referred to.

In most cases, it is a bad idea to take advantage of this feature.  A
person reading the program who sees that @code{foo} is a variable will
not expect that it is a macro as well.  The reader will come across the
identifier @code{foo} in the program and think its value should be that
of the variable @code{foo}, whereas in fact the value is four greater.

One common, useful use of self-reference is to create a macro which
expands to itself.  If you write

@example
#define EPERM EPERM
@end example

@noindent
then the macro @code{EPERM} expands to @code{EPERM}.  Effectively, it is
left alone by the preprocessor whenever it's used in running text.  You
can tell that it's a macro with @samp{#ifdef}.  You might do this if you
want to define numeric constants with an @code{enum}, but have
@samp{#ifdef} be true for each constant.

If a macro @code{x} expands to use a macro @code{y}, and the expansion of
@code{y} refers to the macro @code{x}, that is an @dfn{indirect
self-reference} of @code{x}.  @code{x} is not expanded in this case
either.  Thus, if we have

@example
#define x (4 + y)
#define y (2 * x)
@end example

@noindent
then @code{x} and @code{y} expand as follows:

@example
@group
x    @expansion{} (4 + y)
     @expansion{} (4 + (2 * x))

y    @expansion{} (2 * x)
     @expansion{} (2 * (4 + y))
@end group
@end example

@noindent
Each macro is expanded when it appears in the definition of the other
macro, but not when it indirectly appears in its own definition.

@node Argument Prescan
@subsection Argument Prescan
@cindex expansion of arguments
@cindex macro argument expansion
@cindex prescan of macro arguments

Macro arguments are completely macro-expanded before they are
substituted into a macro body, unless they are stringified or pasted
with other tokens.  After substitution, the entire macro body, including
the substituted arguments, is scanned again for macros to be expanded.
The result is that the arguments are scanned @emph{twice} to expand
macro calls in them.

Most of the time, this has no effect.  If the argument contained any
macro calls, they are expanded during the first scan.  The result
therefore contains no macro calls, so the second scan does not change
it.  If the argument were substituted as given, with no prescan, the
single remaining scan would find the same macro calls and produce the
same results.

You might expect the double scan to change the results when a
self-referential macro is used in an argument of another macro
(@pxref{Self-Referential Macros}): the self-referential macro would be
expanded once in the first scan, and a second time in the second scan.
However, this is not what happens.  The self-references that do not
expand in the first scan are marked so that they will not expand in the
second scan either.

You might wonder, ``Why mention the prescan, if it makes no difference?
And why not skip it and make the preprocessor faster?''  The answer is
that the prescan does make a difference in three special cases:

@itemize @bullet
@item
Nested calls to a macro.

We say that @dfn{nested} calls to a macro occur when a macro's argument
contains a call to that very macro.  For example, if @code{f} is a macro
that expects one argument, @code{f (f (1))} is a nested pair of calls to
@code{f}.  The desired expansion is made by expanding @code{f (1)} and
substituting that into the definition of @code{f}.  The prescan causes
the expected result to happen.  Without the prescan, @code{f (1)} itself
would be substituted as an argument, and the inner use of @code{f} would
appear during the main scan as an indirect self-reference and would not
be expanded.

@item
Macros that call other macros that stringify or concatenate.

If an argument is stringified or concatenated, the prescan does not
occur.  If you @emph{want} to expand a macro, then stringify or
concatenate its expansion, you can do that by causing one macro to call
another macro that does the stringification or concatenation.  For
instance, if you have

@example
#define AFTERX(x) X_ ## x
#define XAFTERX(x) AFTERX(x)
#define TABLESIZE 1024
#define BUFSIZE TABLESIZE
@end example

then @code{AFTERX(BUFSIZE)} expands to @code{X_BUFSIZE}, and
@code{XAFTERX(BUFSIZE)} expands to @code{X_1024}.  (Not to
@code{X_TABLESIZE}.  Prescan always does a complete expansion.)

@item
Macros used in arguments, whose expansions contain unshielded commas.

This can cause a macro expanded on the second scan to be called with the
wrong number of arguments.  Here is an example:

@example
#define foo  a,b
#define bar(x) lose(x)
#define lose(x) (1 + (x))
@end example

We would like @code{bar(foo)} to turn into @code{(1 + (foo))}, which
would then turn into @code{(1 + (a,b))}.  Instead, @code{bar(foo)}
expands into @code{lose(a,b)}, and you get an error because @code{lose}
requires a single argument.  In this case, the problem is easily solved
by the same parentheses that ought to be used to prevent misnesting of
arithmetic operations:

@example
#define foo (a,b)
@exdent or
#define bar(x) lose((x))
@end example

The extra pair of parentheses prevents the comma in @code{foo}'s
definition from being interpreted as an argument separator.

@end itemize

@node Newlines in Arguments
@subsection Newlines in Arguments
@cindex newlines in macro arguments

The invocation of a function-like macro can extend over many logical
lines.  However, in the present implementation, the entire expansion
comes out on one line.  Thus line numbers emitted by the compiler or
debugger refer to the line the invocation started on, which might be
different to the line containing the argument causing the problem.

Here is an example illustrating this:

@example
#define ignore_second_arg(a,b,c) a; c

ignore_second_arg (foo (),
                   ignored (),
                   syntax error);
@end example

@noindent
The syntax error triggered by the tokens @code{syntax error} results in
an error message citing line three---the line of ignore_second_arg---
even though the problematic code comes from line five.

We consider this a bug, and intend to fix it in the near future.

@node Conditionals
@chapter Conditionals
@cindex conditionals

A @dfn{conditional} is a directive that instructs the preprocessor to
select whether or not to include a chunk of code in the final token
stream passed to the compiler.  Preprocessor conditionals can test
arithmetic expressions, or whether a name is defined as a macro, or both
simultaneously using the special @code{defined} operator.

A conditional in the C preprocessor resembles in some ways an @code{if}
statement in C, but it is important to understand the difference between
them.  The condition in an @code{if} statement is tested during the
execution of your program.  Its purpose is to allow your program to
behave differently from run to run, depending on the data it is
operating on.  The condition in a preprocessing conditional directive is
tested when your program is compiled.  Its purpose is to allow different
code to be included in the program depending on the situation at the
time of compilation.

However, the distinction is becoming less clear.  Modern compilers often
do test @code{if} statements when a program is compiled, if their
conditions are known not to vary at run time, and eliminate code which
can never be executed.  If you can count on your compiler to do this,
you may find that your program is more readable if you use @code{if}
statements with constant conditions (perhaps determined by macros).  Of
course, you can only use this to exclude code, not type definitions or
other preprocessing directives, and you can only do it if the code
remains syntactically valid when it is not to be used.

GCC version 3 eliminates this kind of never-executed code even when
not optimizing.  Older versions did it only when optimizing.

@menu
* Conditional Uses::
* Conditional Syntax::
* Deleted Code::
@end menu

@node Conditional Uses
@section Conditional Uses

There are three general reasons to use a conditional.

@itemize @bullet
@item
A program may need to use different code depending on the machine or
operating system it is to run on.  In some cases the code for one
operating system may be erroneous on another operating system; for
example, it might refer to data types or constants that do not exist on
the other system.  When this happens, it is not enough to avoid
executing the invalid code.  Its mere presence will cause the compiler
to reject the program.  With a preprocessing conditional, the offending
code can be effectively excised from the program when it is not valid.

@item
You may want to be able to compile the same source file into two
different programs.  One version might make frequent time-consuming
consistency checks on its intermediate data, or print the values of
those data for debugging, and the other not.

@item
A conditional whose condition is always false is one way to exclude code
from the program but keep it as a sort of comment for future reference.
@end itemize

Simple programs that do not need system-specific logic or complex
debugging hooks generally will not need to use preprocessing
conditionals.

@node Conditional Syntax
@section Conditional Syntax

@findex #if
A conditional in the C preprocessor begins with a @dfn{conditional
directive}: @samp{#if}, @samp{#ifdef} or @samp{#ifndef}.

@menu
* Ifdef::
* If::
* Defined::
* Else::
* Elif::
@end menu

@node Ifdef
@subsection Ifdef
@findex #ifdef
@findex #endif

The simplest sort of conditional is

@example
@group
#ifdef @var{MACRO}

@var{controlled text}

#endif /* @var{MACRO} */
@end group
@end example

@cindex conditional group
This block is called a @dfn{conditional group}.  @var{controlled text}
will be included in the output of the preprocessor if and only if
@var{MACRO} is defined.  We say that the conditional @dfn{succeeds} if
@var{MACRO} is defined, @dfn{fails} if it is not.

The @var{controlled text} inside of a conditional can include
preprocessing directives.  They are executed only if the conditional
succeeds.  You can nest conditional groups inside other conditional
groups, but they must be completely nested.  In other words,
@samp{#endif} always matches the nearest @samp{#ifdef} (or
@samp{#ifndef}, or @samp{#if}).  Also, you cannot start a conditional
group in one file and end it in another.

Even if a conditional fails, the @var{controlled text} inside it is
still run through initial transformations and tokenization.  Therefore,
it must all be lexically valid C@.  Normally the only way this matters is
that all comments and string literals inside a failing conditional group
must still be properly ended.

The comment following the @samp{#endif} is not required, but it is a
good practice if there is a lot of @var{controlled text}, because it
helps people match the @samp{#endif} to the corresponding @samp{#ifdef}.
Older programs sometimes put @var{MACRO} directly after the
@samp{#endif} without enclosing it in a comment.  This is invalid code
according to the C standard.  CPP accepts it with a warning.  It
never affects which @samp{#ifndef} the @samp{#endif} matches.

@findex #ifndef
Sometimes you wish to use some code if a macro is @emph{not} defined.
You can do this by writing @samp{#ifndef} instead of @samp{#ifdef}.
One common use of @samp{#ifndef} is to include code only the first
time a header file is included.  @xref{Once-Only Headers}.

Macro definitions can vary between compilations for several reasons.
Here are some samples.

@itemize @bullet
@item
Some macros are predefined on each kind of machine
(@pxref{System-specific Predefined Macros}).  This allows you to provide
code specially tuned for a particular machine.

@item
System header files define more macros, associated with the features
they implement.  You can test these macros with conditionals to avoid
using a system feature on a machine where it is not implemented.

@item
Macros can be defined or undefined with the @option{-D} and @option{-U}
command line options when you compile the program.  You can arrange to
compile the same source file into two different programs by choosing a
macro name to specify which program you want, writing conditionals to
test whether or how this macro is defined, and then controlling the
state of the macro with command line options, perhaps set in the
Makefile.  @xref{Invocation}.

@item
Your program might have a special header file (often called
@file{config.h}) that is adjusted when the program is compiled.  It can
define or not define macros depending on the features of the system and
the desired capabilities of the program.  The adjustment can be
automated by a tool such as @command{autoconf}, or done by hand.
@end itemize

@node If
@subsection If

The @samp{#if} directive allows you to test the value of an arithmetic
expression, rather than the mere existence of one macro.  Its syntax is

@example
@group
#if @var{expression}

@var{controlled text}

#endif /* @var{expression} */
@end group
@end example

@var{expression} is a C expression of integer type, subject to stringent
restrictions.  It may contain

@itemize @bullet
@item
Integer constants.

@item
Character constants, which are interpreted as they would be in normal
code.

@item
Arithmetic operators for addition, subtraction, multiplication,
division, bitwise operations, shifts, comparisons, and logical
operations (@code{&&} and @code{||}).  The latter two obey the usual
short-circuiting rules of standard C@.

@item
Macros.  All macros in the expression are expanded before actual
computation of the expression's value begins.

@item
Uses of the @code{defined} operator, which lets you check whether macros
are defined in the middle of an @samp{#if}.

@item
Identifiers that are not macros, which are all considered to be the
number zero.  This allows you to write @code{@w{#if MACRO}} instead of
@code{@w{#ifdef MACRO}}, if you know that MACRO, when defined, will
always have a nonzero value.  Function-like macros used without their
function call parentheses are also treated as zero.

In some contexts this shortcut is undesirable.  The @option{-Wundef}
option causes GCC to warn whenever it encounters an identifier which is
not a macro in an @samp{#if}.
@end itemize

The preprocessor does not know anything about types in the language.
Therefore, @code{sizeof} operators are not recognized in @samp{#if}, and
neither are @code{enum} constants.  They will be taken as identifiers
which are not macros, and replaced by zero.  In the case of
@code{sizeof}, this is likely to cause the expression to be invalid.

The preprocessor calculates the value of @var{expression}.  It carries
out all calculations in the widest integer type known to the compiler;
on most machines supported by GCC this is 64 bits.  This is not the same
rule as the compiler uses to calculate the value of a constant
expression, and may give different results in some cases.  If the value
comes out to be nonzero, the @samp{#if} succeeds and the @var{controlled
text} is included; otherwise it is skipped.

If @var{expression} is not correctly formed, GCC issues an error and
treats the conditional as having failed.

@node Defined
@subsection Defined

@cindex @code{defined}
The special operator @code{defined} is used in @samp{#if} and
@samp{#elif} expressions to test whether a certain name is defined as a
macro.  @code{defined @var{name}} and @code{defined (@var{name})} are
both expressions whose value is 1 if @var{name} is defined as a macro at
the current point in the program, and 0 otherwise.  Thus,  @code{@w{#if
defined MACRO}} is precisely equivalent to @code{@w{#ifdef MACRO}}.

@code{defined} is useful when you wish to test more than one macro for
existence at once.  For example,

@example
#if defined (__vax__) || defined (__ns16000__)
@end example

@noindent
would succeed if either of the names @code{__vax__} or
@code{__ns16000__} is defined as a macro.

Conditionals written like this:

@example
#if defined BUFSIZE && BUFSIZE >= 1024
@end example

@noindent
can generally be simplified to just @code{@w{#if BUFSIZE >= 1024}},
since if @code{BUFSIZE} is not defined, it will be interpreted as having
the value zero.

If the @code{defined} operator appears as a result of a macro expansion,
the C standard says the behavior is undefined.  GNU cpp treats it as a
genuine @code{defined} operator and evaluates it normally.  It will warn
wherever your code uses this feature if you use the command-line option
@option{-pedantic}, since other compilers may handle it differently.

@node Else
@subsection Else

@findex #else
The @samp{#else} directive can be added to a conditional to provide
alternative text to be used if the condition fails.  This is what it
looks like:

@example
@group
#if @var{expression}
@var{text-if-true}
#else /* Not @var{expression} */
@var{text-if-false}
#endif /* Not @var{expression} */
@end group
@end example

@noindent
If @var{expression} is nonzero, the @var{text-if-true} is included and
the @var{text-if-false} is skipped.  If @var{expression} is zero, the
opposite happens.

You can use @samp{#else} with @samp{#ifdef} and @samp{#ifndef}, too.

@node Elif
@subsection Elif

@findex #elif
One common case of nested conditionals is used to check for more than two
possible alternatives.  For example, you might have

@example
#if X == 1
@dots{}
#else /* X != 1 */
#if X == 2
@dots{}
#else /* X != 2 */
@dots{}
#endif /* X != 2 */
#endif /* X != 1 */
@end example

Another conditional directive, @samp{#elif}, allows this to be
abbreviated as follows:

@example
#if X == 1
@dots{}
#elif X == 2
@dots{}
#else /* X != 2 and X != 1*/
@dots{}
#endif /* X != 2 and X != 1*/
@end example

@samp{#elif} stands for ``else if''.  Like @samp{#else}, it goes in the
middle of a conditional group and subdivides it; it does not require a
matching @samp{#endif} of its own.  Like @samp{#if}, the @samp{#elif}
directive includes an expression to be tested.  The text following the
@samp{#elif} is processed only if the original @samp{#if}-condition
failed and the @samp{#elif} condition succeeds.

More than one @samp{#elif} can go in the same conditional group.  Then
the text after each @samp{#elif} is processed only if the @samp{#elif}
condition succeeds after the original @samp{#if} and all previous
@samp{#elif} directives within it have failed.

@samp{#else} is allowed after any number of @samp{#elif} directives, but
@samp{#elif} may not follow @samp{#else}.

@node Deleted Code
@section Deleted Code
@cindex commenting out code

If you replace or delete a part of the program but want to keep the old
code around for future reference, you often cannot simply comment it
out.  Block comments do not nest, so the first comment inside the old
code will end the commenting-out.  The probable result is a flood of
syntax errors.

One way to avoid this problem is to use an always-false conditional
instead.  For instance, put @code{#if 0} before the deleted code and
@code{#endif} after it.  This works even if the code being turned
off contains conditionals, but they must be entire conditionals
(balanced @samp{#if} and @samp{#endif}).

Some people use @code{#ifdef notdef} instead.  This is risky, because
@code{notdef} might be accidentally defined as a macro, and then the
conditional would succeed.  @code{#if 0} can be counted on to fail.

Do not use @code{#if 0} for comments which are not C code.  Use a real
comment, instead.  The interior of @code{#if 0} must consist of complete
tokens; in particular, single-quote characters must balance.  Comments
often contain unbalanced single-quote characters (known in English as
apostrophes).  These confuse @code{#if 0}.  They don't confuse
@samp{/*}.

@node Diagnostics
@chapter Diagnostics
@cindex diagnostic
@cindex reporting errors
@cindex reporting warnings

@findex #error
The directive @samp{#error} causes the preprocessor to report a fatal
error.  The tokens forming the rest of the line following @samp{#error}
are used as the error message.

You would use @samp{#error} inside of a conditional that detects a
combination of parameters which you know the program does not properly
support.  For example, if you know that the program will not run
properly on a VAX, you might write

@example
@group
#ifdef __vax__
#error "Won't work on VAXen.  See comments at get_last_object."
#endif
@end group
@end example

If you have several configuration parameters that must be set up by
the installation in a consistent way, you can use conditionals to detect
an inconsistency and report it with @samp{#error}.  For example,

@example
#if !defined(UNALIGNED_INT_ASM_OP) && defined(DWARF2_DEBUGGING_INFO)
#error "DWARF2_DEBUGGING_INFO requires UNALIGNED_INT_ASM_OP."
#endif
@end example

@findex #warning
The directive @samp{#warning} is like @samp{#error}, but causes the
preprocessor to issue a warning and continue preprocessing.  The tokens
following @samp{#warning} are used as the warning message.

You might use @samp{#warning} in obsolete header files, with a message
directing the user to the header file which should be used instead.

Neither @samp{#error} nor @samp{#warning} macro-expands its argument.
Internal whitespace sequences are each replaced with a single space.
The line must consist of complete tokens.  It is wisest to make the
argument of these directives be a single string constant; this avoids
problems with apostrophes and the like.

@node Line Control
@chapter Line Control
@cindex line control

The C preprocessor informs the C compiler of the location in your source
code where each token came from.  Presently, this is just the file name
and line number.  All the tokens resulting from macro expansion are
reported as having appeared on the line of the source file where the
outermost macro was used.  We intend to be more accurate in the future.

If you write a program which generates source code, such as the
@command{bison} parser generator, you may want to adjust the preprocessor's
notion of the current file name and line number by hand.  Parts of the
output from @command{bison} are generated from scratch, other parts come
from a standard parser file.  The rest are copied verbatim from
@command{bison}'s input.  You would like compiler error messages and
symbolic debuggers to be able to refer to @code{bison}'s input file.

@findex #line
@command{bison} or any such program can arrange this by writing
@samp{#line} directives into the output file.  @samp{#line} is a
directive that specifies the original line number and source file name
for subsequent input in the current preprocessor input file.
@samp{#line} has three variants:

@table @code
@item #line @var{linenum}
@var{linenum} is a non-negative decimal integer constant.  It specifies
the line number which should be reported for the following line of
input.  Subsequent lines are counted from @var{linenum}.

@item #line @var{linenum} @var{filename}
@var{linenum} is the same as for the first form, and has the same
effect.  In addition, @var{filename} is a string constant.  The
following line and all subsequent lines are reported to come from the
file it specifies, until something else happens to change that.
@var{filename} is interpreted according to the normal rules for a string
constant: backslash escapes are interpreted.  This is different from
@samp{#include}.

Previous versions of CPP did not interpret escapes in @samp{#line};
we have changed it because the standard requires they be interpreted,
and most other compilers do.

@item #line @var{anything else}
@var{anything else} is checked for macro calls, which are expanded.
The result should match one of the above two forms.
@end table

@samp{#line} directives alter the results of the @code{__FILE__} and
@code{__LINE__} predefined macros from that point on.  @xref{Standard
Predefined Macros}.  They do not have any effect on @samp{#include}'s
idea of the directory containing the current file.  This is a change
from GCC 2.95.  Previously, a file reading

@smallexample
#line 1 "../src/gram.y"
#include "gram.h"
@end smallexample

would search for @file{gram.h} in @file{../src}, then the @option{-I}
chain; the directory containing the physical source file would not be
searched.  In GCC 3.0 and later, the @samp{#include} is not affected by
the presence of a @samp{#line} referring to a different directory.

We made this change because the old behavior caused problems when
generated source files were transported between machines.  For instance,
it is common practice to ship generated parsers with a source release,
so that people building the distribution do not need to have yacc or
Bison installed.  These files frequently have @samp{#line} directives
referring to the directory tree of the system where the distribution was
created.  If GCC tries to search for headers in those directories, the
build is likely to fail.

The new behavior can cause failures too, if the generated file is not
in the same directory as its source and it attempts to include a header
which would be visible searching from the directory containing the
source file.  However, this problem is easily solved with an additional
@option{-I} switch on the command line.  The failures caused by the old
semantics could sometimes be corrected only by editing the generated
files, which is difficult and error-prone.

@node Pragmas
@chapter Pragmas

The @samp{#pragma} directive is the method specified by the C standard
for providing additional information to the compiler, beyond what is
conveyed in the language itself.  Three forms of this directive
(commonly known as @dfn{pragmas}) are specified by the 1999 C standard.
A C compiler is free to attach any meaning it likes to other pragmas.

GCC has historically preferred to use extensions to the syntax of the
language, such as @code{__attribute__}, for this purpose.  However, GCC
does define a few pragmas of its own.  These mostly have effects on the
entire translation unit or source file.

In GCC version 3, all GNU-defined, supported pragmas have been given a
@code{GCC} prefix.  This is in line with the @code{STDC} prefix on all
pragmas defined by C99.  For backward compatibility, pragmas which were
recognized by previous versions are still recognized without the
@code{GCC} prefix, but that usage is deprecated.  Some older pragmas are
deprecated in their entirety.  They are not recognized with the
@code{GCC} prefix.  @xref{Obsolete Features}.

@cindex @code{_Pragma}
C99 introduces the @code{@w{_Pragma}} operator.  This feature addresses a
major problem with @samp{#pragma}: being a directive, it cannot be
produced as the result of macro expansion.  @code{@w{_Pragma}} is an
operator, much like @code{sizeof} or @code{defined}, and can be embedded
in a macro.

Its syntax is @code{@w{_Pragma (@var{string-literal})}}, where
@var{string-literal} can be either a normal or wide-character string
literal.  It is destringized, by replacing all @samp{\\} with a single
@samp{\} and all @samp{\"} with a @samp{"}.  The result is then
processed as if it had appeared as the right hand side of a
@samp{#pragma} directive.  For example,

@example
_Pragma ("GCC dependency \"parse.y\"")
@end example

@noindent
has the same effect as @code{#pragma GCC dependency "parse.y"}.  The
same effect could be achieved using macros, for example

@example
#define DO_PRAGMA(x) _Pragma (#x)
DO_PRAGMA (GCC dependency "parse.y")
@end example

The standard is unclear on where a @code{_Pragma} operator can appear.
The preprocessor does not accept it within a preprocessing conditional
directive like @samp{#if}.  To be safe, you are probably best keeping it
out of directives other than @samp{#define}, and putting it on a line of
its own.

This manual documents the pragmas which are meaningful to the
preprocessor itself.  Other pragmas are meaningful to the C or C++
compilers.  They are documented in the GCC manual.

@ftable @code
@item #pragma GCC dependency
@code{#pragma GCC dependency} allows you to check the relative dates of
the current file and another file.  If the other file is more recent than
the current file, a warning is issued.  This is useful if the current
file is derived from the other file, and should be regenerated.  The
other file is searched for using the normal include search path.
Optional trailing text can be used to give more information in the
warning message.

@example
#pragma GCC dependency "parse.y"
#pragma GCC dependency "/usr/include/time.h" rerun fixincludes
@end example

@item #pragma GCC poison
Sometimes, there is an identifier that you want to remove completely
from your program, and make sure that it never creeps back in.  To
enforce this, you can @dfn{poison} the identifier with this pragma.
@code{#pragma GCC poison} is followed by a list of identifiers to
poison.  If any of those identifiers appears anywhere in the source
after the directive, it is a hard error.  For example,

@example
#pragma GCC poison printf sprintf fprintf
sprintf(some_string, "hello");
@end example

@noindent
will produce an error.

If a poisoned identifier appears as part of the expansion of a macro
which was defined before the identifier was poisoned, it will @emph{not}
cause an error.  This lets you poison an identifier without worrying
about system headers defining macros that use it.

For example,

@example
#define strrchr rindex
#pragma GCC poison rindex
strrchr(some_string, 'h');
@end example

@noindent
will not produce an error.

@item #pragma GCC system_header
This pragma takes no arguments.  It causes the rest of the code in the
current file to be treated as if it came from a system header.
@xref{System Headers}.

@end ftable

@node Other Directives
@chapter Other Directives

@findex #ident
The @samp{#ident} directive takes one argument, a string constant.  On
some systems, that string constant is copied into a special segment of
the object file.  On other systems, the directive is ignored.

This directive is not part of the C standard, but it is not an official
GNU extension either.  We believe it came from System V@.

@findex #sccs
The @samp{#sccs} directive is recognized, because it appears in the
header files of some systems.  It is a very old, obscure, extension
which we did not invent, and we have been unable to find any
documentation of what it should do, so GCC simply ignores it.

@cindex null directive
The @dfn{null directive} consists of a @samp{#} followed by a newline,
with only whitespace (including comments) in between.  A null directive
is understood as a preprocessing directive but has no effect on the
preprocessor output.  The primary significance of the existence of the
null directive is that an input line consisting of just a @samp{#} will
produce no output, rather than a line of output containing just a
@samp{#}.  Supposedly some old C programs contain such lines.

@node Preprocessor Output
@chapter Preprocessor Output

When the C preprocessor is used with the C, C++, or Objective-C
compilers, it is integrated into the compiler and communicates a stream
of binary tokens directly to the compiler's parser.  However, it can
also be used in the more conventional standalone mode, where it produces
textual output.
@c FIXME: Document the library interface.

@cindex output format
The output from the C preprocessor looks much like the input, except
that all preprocessing directive lines have been replaced with blank
lines and all comments with spaces.  Long runs of blank lines are
discarded.

The ISO standard specifies that it is implementation defined whether a
preprocessor preserves whitespace between tokens, or replaces it with
e.g.@: a single space.  In GNU CPP, whitespace between tokens is collapsed
to become a single space, with the exception that the first token on a
non-directive line is preceded with sufficient spaces that it appears in
the same column in the preprocessed output that it appeared in the
original source file.  This is so the output is easy to read.
@xref{Differences from previous versions}.  CPP does not insert any
whitespace where there was none in the original source, except where
necessary to prevent an accidental token paste.

@cindex linemarkers
Source file name and line number information is conveyed by lines
of the form

@example
# @var{linenum} @var{filename} @var{flags}
@end example

@noindent
These are called @dfn{linemarkers}.  They are inserted as needed into
the output (but never within a string or character constant).  They mean
that the following line originated in file @var{filename} at line
@var{linenum}.  @var{filename} will never contain any non-printing
characters; they are replaced with octal escape sequences.

After the file name comes zero or more flags, which are @samp{1},
@samp{2}, @samp{3}, or @samp{4}.  If there are multiple flags, spaces
separate them.  Here is what the flags mean:

@table @samp
@item 1
This indicates the start of a new file.
@item 2
This indicates returning to a file (after having included another file).
@item 3
This indicates that the following text comes from a system header file,
so certain warnings should be suppressed.
@item 4
This indicates that the following text should be treated as being
wrapped in an implicit @code{extern "C"} block.
@c maybe cross reference NO_IMPLICIT_EXTERN_C
@end table

As an extension, the preprocessor accepts linemarkers in non-assembler
input files.  They are treated like the corresponding @samp{#line}
directive, (@pxref{Line Control}), except that trailing flags are
permitted, and are interpreted with the meanings described above.  If
multiple flags are given, they must be in ascending order.

Some directives may be duplicated in the output of the preprocessor.
These are @samp{#ident} (always), @samp{#pragma} (only if the
preprocessor does not handle the pragma itself), and @samp{#define} and
@samp{#undef} (with certain debugging options).  If this happens, the
@samp{#} of the directive will always be in the first column, and there
will be no space between the @samp{#} and the directive name.  If macro
expansion happens to generate tokens which might be mistaken for a
duplicated directive, a space will be inserted between the @samp{#} and
the directive name.

@node Traditional Mode
@chapter Traditional Mode

Traditional (pre-standard) C preprocessing is rather different from
the preprocessing specified by the standard.  When GCC is given the
@option{-traditional-cpp} option, it attempts to emulate a traditional
preprocessor.

GCC versions 3.2 and later only support traditional mode semantics in
the preprocessor, and not in the compiler front ends.  This chapter
outlines the traditional preprocessor semantics we implemented.

The implementation does not correspond precisely to the behavior of
earlier versions of GCC, nor to any true traditional preprocessor.
After all, inconsistencies among traditional implementations were a
major motivation for C standardization.  However, we intend that it
should be compatible with true traditional preprocessors in all ways
that actually matter.

@menu
* Traditional lexical analysis::
* Traditional macros::
* Traditional miscellany::
* Traditional warnings::
@end menu

@node Traditional lexical analysis
@section Traditional lexical analysis

The traditional preprocessor does not decompose its input into tokens
the same way a standards-conforming preprocessor does.  The input is
simply treated as a stream of text with minimal internal form.

This implementation does not treat trigraphs (@pxref{trigraphs})
specially since they were an invention of the standards committee.  It
handles arbitrarily-positioned escaped newlines properly and splices
the lines as you would expect; many traditional preprocessors did not
do this.

The form of horizontal whitespace in the input file is preserved in
the output.  In particular, hard tabs remain hard tabs.  This can be
useful if, for example, you are preprocessing a Makefile.

Traditional CPP only recognizes C-style block comments, and treats the
@samp{/*} sequence as introducing a comment only if it lies outside
quoted text.  Quoted text is introduced by the usual single and double
quotes, and also by an initial @samp{<} in a @code{#include}
directive.

Traditionally, comments are completely removed and are not replaced
with a space.  Since a traditional compiler does its own tokenization
of the output of the preprocessor, this means that comments can
effectively be used as token paste operators.  However, comments
behave like separators for text handled by the preprocessor itself,
since it doesn't re-lex its input.  For example, in

@smallexample
#if foo/**/bar
@end smallexample

@noindent
@samp{foo} and @samp{bar} are distinct identifiers and expanded
separately if they happen to be macros.  In other words, this
directive is equivalent to

@smallexample
#if foo bar
@end smallexample

@noindent
rather than

@smallexample
#if foobar
@end smallexample

Generally speaking, in traditional mode an opening quote need not have
a matching closing quote.  In particular, a macro may be defined with
replacement text that contains an unmatched quote.  Of course, if you
attempt to compile preprocessed output containing an unmatched quote
you will get a syntax error.

However, all preprocessing directives other than @code{#define}
require matching quotes.  For example:

@smallexample
#define m This macro's fine and has an unmatched quote
"/* This is not a comment.  */
/* This is a comment.  The following #include directive
   is ill-formed.  */
#include <stdio.h
@end smallexample

Just as for the ISO preprocessor, what would be a closing quote can be
escaped with a backslash to prevent the quoted text from closing.

@node Traditional macros
@section Traditional macros

The major difference between traditional and ISO macros is that the
former expand to text rather than to a token sequence.  CPP removes
all leading and trailing horizontal whitespace from a macro's
replacement text before storing it, but preserves the form of internal
whitespace.

One consequence is that it is legitimate for the replacement text to
contain an unmatched quote (@pxref{Traditional lexical analysis}). An
unclosed string or character constant continues into the text
following the macro call.  Similarly, the text at the end of a macro's
expansion can run together with the text after the macro invocation to
produce a single token.

Normally comments are removed from the replacement text after the
macro is expanded, but if the @option{-CC} option is passed on the
command line comments are preserved.  (In fact, the current
implementation removes comments even before saving the macro
replacement text, but it careful to do it in such a way that the
observed effect is identical even in the function-like macro case.)

The ISO stringification operator @samp{#} and token paste operator
@samp{##} have no special meaning.  As explained later, an effect
similar to these operators can be obtained in a different way.  Macro
names that are embedded in quotes, either from the main file or after
macro replacement, do not expand.

CPP replaces an unquoted object-like macro name with its replacement
text, and then rescans it for further macros to replace.  Unlike
standard macro expansion, traditional macro expansion has no provision
to prevent recursion.  If an object-like macro appears unquoted in its
replacement text, it will be replaced again during the rescan pass,
and so on @emph{ad infinitum}.  GCC detects when it is expanding
recursive macros, emits an error message, and continues after the
offending macro invocation.

@smallexample
#define PLUS +
#define INC(x) PLUS+x
INC(foo);
     @expansion{} ++foo;
@end smallexample

Function-like macros are similar in form but quite different in
behavior to their ISO counterparts.  Their arguments are contained
within parentheses, are comma-separated, and can cross physical lines.
Commas within nested parentheses are not treated as argument
separators.  Similarly, a quote in an argument cannot be left
unclosed; a following comma or parenthesis that comes before the
closing quote is treated like any other character.  There is no
facility for handling variadic macros.

This implementation removes all comments from macro arguments, unless
the @option{-C} option is given.  The form of all other horizontal
whitespace in arguments is preserved, including leading and trailing
whitespace.  In particular

@smallexample
f( )
@end smallexample

@noindent
is treated as an invocation of the macro @samp{f} with a single
argument consisting of a single space.  If you want to invoke a
function-like macro that takes no arguments, you must not leave any
whitespace between the parentheses.

If a macro argument crosses a new line, the new line is replaced with
a space when forming the argument.  If the previous line contained an
unterminated quote, the following line inherits the quoted state.

Traditional preprocessors replace parameters in the replacement text
with their arguments regardless of whether the parameters are within
quotes or not.  This provides a way to stringize arguments.  For
example

@smallexample
#define str(x) "x"
str(/* A comment */some text )
     @expansion{} "some text "
@end smallexample

@noindent
Note that the comment is removed, but that the trailing space is
preserved.  Here is an example of using a comment to effect token
pasting.

@smallexample
#define suffix(x) foo_/**/x
suffix(bar)
     @expansion{} foo_bar
@end smallexample

@node Traditional miscellany
@section Traditional miscellany

Here are some things to be aware of when using the traditional
preprocessor.

@itemize @bullet
@item
Preprocessing directives are recognized only when their leading
@samp{#} appears in the first column.  There can be no whitespace
between the beginning of the line and the @samp{#}, but whitespace can
follow the @samp{#}.

@item
A true traditional C preprocessor does not recognize @samp{#error} or
@samp{#pragma}, and may not recognize @samp{#elif}.  CPP supports all
the directives in traditional mode that it supports in ISO mode,
including extensions, with the exception that the effects of
@samp{#pragma GCC poison} are undefined.

@item
__STDC__ is not defined.

@item
If you use digraphs the behavior is undefined.

@item
If a line that looks like a directive appears within macro arguments,
the behavior is undefined.

@end itemize

@node Traditional warnings
@section Traditional warnings
You can request warnings about features that did not exist, or worked
differently, in traditional C with the @option{-Wtraditional} option.
GCC does not warn about features of ISO C which you must use when you
are using a conforming compiler, such as the @samp{#} and @samp{##}
operators.

Presently @option{-Wtraditional} warns about:

@itemize @bullet
@item
Macro parameters that appear within string literals in the macro body.
In traditional C macro replacement takes place within string literals,
but does not in ISO C@.

@item
In traditional C, some preprocessor directives did not exist.
Traditional preprocessors would only consider a line to be a directive
if the @samp{#} appeared in column 1 on the line.  Therefore
@option{-Wtraditional} warns about directives that traditional C
understands but would ignore because the @samp{#} does not appear as the
first character on the line.  It also suggests you hide directives like
@samp{#pragma} not understood by traditional C by indenting them.  Some
traditional implementations would not recognize @samp{#elif}, so it
suggests avoiding it altogether.

@item
A function-like macro that appears without an argument list.  In some
traditional preprocessors this was an error.  In ISO C it merely means
that the macro is not expanded.

@item
The unary plus operator.  This did not exist in traditional C@.

@item
The @samp{U} and @samp{LL} integer constant suffixes, which were not
available in traditional C@.  (Traditional C does support the @samp{L}
suffix for simple long integer constants.)  You are not warned about
uses of these suffixes in macros defined in system headers.  For
instance, @code{UINT_MAX} may well be defined as @code{4294967295U}, but
you will not be warned if you use @code{UINT_MAX}.

You can usually avoid the warning, and the related warning about
constants which are so large that they are unsigned, by writing the
integer constant in question in hexadecimal, with no U suffix.  Take
care, though, because this gives the wrong result in exotic cases.
@end itemize

@node Implementation Details
@chapter Implementation Details

Here we document details of how the preprocessor's implementation
affects its user-visible behavior.  You should try to avoid undue
reliance on behavior described here, as it is possible that it will
change subtly in future implementations.

Also documented here are obsolete features and changes from previous
versions of CPP@.

@menu
* Implementation-defined behavior::
* Implementation limits::
* Obsolete Features::
* Differences from previous versions::
@end menu

@node Implementation-defined behavior
@section Implementation-defined behavior
@cindex implementation-defined behavior

This is how CPP behaves in all the cases which the C standard
describes as @dfn{implementation-defined}.  This term means that the
implementation is free to do what it likes, but must document its choice
and stick to it.
@c FIXME: Check the C++ standard for more implementation-defined stuff.

@itemize @bullet
@need 1000
@item The mapping of physical source file multi-byte characters to the
execution character set.

Currently, CPP requires its input to be ASCII or UTF-8.  The execution
character set may be controlled by the user, with the
@code{-ftarget-charset} and @code{-ftarget-wide-charset} options.

@item Identifier characters.
@anchor{Identifier characters}

The C and C++ standards allow identifiers to be composed of @samp{_}
and the alphanumeric characters.  C++ and C99 also allow universal
character names (not implemented in GCC), and C99 further permits
implementation-defined characters.

GCC allows the @samp{$} character in identifiers as an extension for
most targets.  This is true regardless of the @option{std=} switch,
since this extension cannot conflict with standards-conforming
programs.  When preprocessing assembler, however, dollars are not
identifier characters by default.

Currently the targets that by default do not permit @samp{$} are AVR,
IP2K, MMIX, MIPS Irix 3, ARM aout, and PowerPC targets for the AIX and
BeOS operating systems.

You can override the default with @option{-fdollars-in-identifiers} or
@option{fno-dollars-in-identifiers}.  @xref{fdollars-in-identifiers}.

@item Non-empty sequences of whitespace characters.

In textual output, each whitespace sequence is collapsed to a single
space.  For aesthetic reasons, the first token on each non-directive
line of output is preceded with sufficient spaces that it appears in the
same column as it did in the original source file.

@item The numeric value of character constants in preprocessor expressions.

The preprocessor and compiler interpret character constants in the
same way; i.e.@: escape sequences such as @samp{\a} are given the
values they would have on the target machine.

The compiler values a multi-character character constant a character
at a time, shifting the previous value left by the number of bits per
target character, and then or-ing in the bit-pattern of the new
character truncated to the width of a target character.  The final
bit-pattern is given type @code{int}, and is therefore signed,
regardless of whether single characters are signed or not (a slight
change from versions 3.1 and earlier of GCC).  If there are more
characters in the constant than would fit in the target @code{int} the
compiler issues a warning, and the excess leading characters are
ignored.

For example, 'ab' for a target with an 8-bit @code{char} would be
interpreted as @w{(int) ((unsigned char) 'a' * 256 + (unsigned char)
'b')}, and '\234a' as @w{(int) ((unsigned char) '\234' * 256 + (unsigned
char) 'a')}.

@item Source file inclusion.

For a discussion on how the preprocessor locates header files,
@ref{Include Operation}.

@item Interpretation of the filename resulting from a macro-expanded
@samp{#include} directive.

@xref{Computed Includes}.

@item Treatment of a @samp{#pragma} directive that after macro-expansion
results in a standard pragma.

No macro expansion occurs on any @samp{#pragma} directive line, so the
question does not arise.

Note that GCC does not yet implement any of the standard
pragmas.

@end itemize

@node Implementation limits
@section Implementation limits
@cindex implementation limits

CPP has a small number of internal limits.  This section lists the
limits which the C standard requires to be no lower than some minimum,
and all the others we are aware of.  We intend there to be as few limits
as possible.  If you encounter an undocumented or inconvenient limit,
please report that to us as a bug.  (See the section on reporting bugs in
the GCC manual.)

Where we say something is limited @dfn{only by available memory}, that
means that internal data structures impose no intrinsic limit, and space
is allocated with @code{malloc} or equivalent.  The actual limit will
therefore depend on many things, such as the size of other things
allocated by the compiler at the same time, the amount of memory
consumed by other processes on the same computer, etc.

@itemize @bullet

@item Nesting levels of @samp{#include} files.

We impose an arbitrary limit of 200 levels, to avoid runaway recursion.
The standard requires at least 15 levels.

@item Nesting levels of conditional inclusion.

The C standard mandates this be at least 63.  CPP is limited only by
available memory.

@item Levels of parenthesized expressions within a full expression.

The C standard requires this to be at least 63.  In preprocessor
conditional expressions, it is limited only by available memory.

@item Significant initial characters in an identifier or macro name.

The preprocessor treats all characters as significant.  The C standard
requires only that the first 63 be significant.

@item Number of macros simultaneously defined in a single translation unit.

The standard requires at least 4095 be possible.  CPP is limited only
by available memory.

@item Number of parameters in a macro definition and arguments in a macro call.

We allow @code{USHRT_MAX}, which is no smaller than 65,535.  The minimum
required by the standard is 127.

@item Number of characters on a logical source line.

The C standard requires a minimum of 4096 be permitted.  CPP places
no limits on this, but you may get incorrect column numbers reported in
diagnostics for lines longer than 65,535 characters.

@item Maximum size of a source file.

The standard does not specify any lower limit on the maximum size of a
source file.  GNU cpp maps files into memory, so it is limited by the
available address space.  This is generally at least two gigabytes.
Depending on the operating system, the size of physical memory may or
may not be a limitation.

@end itemize

@node Obsolete Features
@section Obsolete Features

CPP has a number of features which are present mainly for
compatibility with older programs.  We discourage their use in new code.
In some cases, we plan to remove the feature in a future version of GCC@.

@menu
* Assertions::
* Obsolete once-only headers::
@end menu

@node Assertions
@subsection Assertions
@cindex assertions

@dfn{Assertions} are a deprecated alternative to macros in writing
conditionals to test what sort of computer or system the compiled
program will run on.  Assertions are usually predefined, but you can
define them with preprocessing directives or command-line options.

Assertions were intended to provide a more systematic way to describe
the compiler's target system.  However, in practice they are just as
unpredictable as the system-specific predefined macros.  In addition, they
are not part of any standard, and only a few compilers support them.
Therefore, the use of assertions is @strong{less} portable than the use
of system-specific predefined macros.  We recommend you do not use them at
all.

@cindex predicates
An assertion looks like this:

@example
#@var{predicate} (@var{answer})
@end example

@noindent
@var{predicate} must be a single identifier.  @var{answer} can be any
sequence of tokens; all characters are significant except for leading
and trailing whitespace, and differences in internal whitespace
sequences are ignored.  (This is similar to the rules governing macro
redefinition.)  Thus, @code{(x + y)} is different from @code{(x+y)} but
equivalent to @code{@w{( x + y )}}.  Parentheses do not nest inside an
answer.

@cindex testing predicates
To test an assertion, you write it in an @samp{#if}.  For example, this
conditional succeeds if either @code{vax} or @code{ns16000} has been
asserted as an answer for @code{machine}.

@example
#if #machine (vax) || #machine (ns16000)
@end example

@noindent
You can test whether @emph{any} answer is asserted for a predicate by
omitting the answer in the conditional:

@example
#if #machine
@end example

@findex #assert
Assertions are made with the @samp{#assert} directive.  Its sole
argument is the assertion to make, without the leading @samp{#} that
identifies assertions in conditionals.

@example
#assert @var{predicate} (@var{answer})
@end example

@noindent
You may make several assertions with the same predicate and different
answers.  Subsequent assertions do not override previous ones for the
same predicate.  All the answers for any given predicate are
simultaneously true.

@cindex assertions, canceling
@findex #unassert
Assertions can be canceled with the @samp{#unassert} directive.  It
has the same syntax as @samp{#assert}.  In that form it cancels only the
answer which was specified on the @samp{#unassert} line; other answers
for that predicate remain true.  You can cancel an entire predicate by
leaving out the answer:

@example
#unassert @var{predicate}
@end example

@noindent
In either form, if no such assertion has been made, @samp{#unassert} has
no effect.

You can also make or cancel assertions using command line options.
@xref{Invocation}.

@node Obsolete once-only headers
@subsection Obsolete once-only headers

CPP supports two more ways of indicating that a header file should be
read only once.  Neither one is as portable as a wrapper @samp{#ifndef},
and we recommend you do not use them in new programs.

@findex #import
In the Objective-C language, there is a variant of @samp{#include}
called @samp{#import} which includes a file, but does so at most once.
If you use @samp{#import} instead of @samp{#include}, then you don't
need the conditionals inside the header file to prevent multiple
inclusion of the contents.  GCC permits the use of @samp{#import} in C
and C++ as well as Objective-C@.  However, it is not in standard C or C++
and should therefore not be used by portable programs.

@samp{#import} is not a well designed feature.  It requires the users of
a header file to know that it should only be included once.  It is much
better for the header file's implementor to write the file so that users
don't need to know this.  Using a wrapper @samp{#ifndef} accomplishes
this goal.

In the present implementation, a single use of @samp{#import} will
prevent the file from ever being read again, by either @samp{#import} or
@samp{#include}.  You should not rely on this; do not use both
@samp{#import} and @samp{#include} to refer to the same header file.

Another way to prevent a header file from being included more than once
is with the @samp{#pragma once} directive.  If @samp{#pragma once} is
seen when scanning a header file, that file will never be read again, no
matter what.

@samp{#pragma once} does not have the problems that @samp{#import} does,
but it is not recognized by all preprocessors, so you cannot rely on it
in a portable program.

@node Differences from previous versions
@section Differences from previous versions
@cindex differences from previous versions

This section details behavior which has changed from previous versions
of CPP@.  We do not plan to change it again in the near future, but
we do not promise not to, either.

The ``previous versions'' discussed here are 2.95 and before.  The
behavior of GCC 3.0 is mostly the same as the behavior of the widely
used 2.96 and 2.97 development snapshots.  Where there are differences,
they generally represent bugs in the snapshots.

@itemize @bullet

@item Order of evaluation of @samp{#} and @samp{##} operators

The standard does not specify the order of evaluation of a chain of
@samp{##} operators, nor whether @samp{#} is evaluated before, after, or
at the same time as @samp{##}.  You should therefore not write any code
which depends on any specific ordering.  It is possible to guarantee an
ordering, if you need one, by suitable use of nested macros.

An example of where this might matter is pasting the arguments @samp{1},
@samp{e} and @samp{-2}.  This would be fine for left-to-right pasting,
but right-to-left pasting would produce an invalid token @samp{e-2}.

GCC 3.0 evaluates @samp{#} and @samp{##} at the same time and strictly
left to right.  Older versions evaluated all @samp{#} operators first,
then all @samp{##} operators, in an unreliable order.

@item The form of whitespace between tokens in preprocessor output

@xref{Preprocessor Output}, for the current textual format.  This is
also the format used by stringification.  Normally, the preprocessor
communicates tokens directly to the compiler's parser, and whitespace
does not come up at all.

Older versions of GCC preserved all whitespace provided by the user and
inserted lots more whitespace of their own, because they could not
accurately predict when extra spaces were needed to prevent accidental
token pasting.

@item Optional argument when invoking rest argument macros

As an extension, GCC permits you to omit the variable arguments entirely
when you use a variable argument macro.  This is forbidden by the 1999 C
standard, and will provoke a pedantic warning with GCC 3.0.  Previous
versions accepted it silently.

@item @samp{##} swallowing preceding text in rest argument macros

Formerly, in a macro expansion, if @samp{##} appeared before a variable
arguments parameter, and the set of tokens specified for that argument
in the macro invocation was empty, previous versions of CPP would
back up and remove the preceding sequence of non-whitespace characters
(@strong{not} the preceding token).  This extension is in direct
conflict with the 1999 C standard and has been drastically pared back.

In the current version of the preprocessor, if @samp{##} appears between
a comma and a variable arguments parameter, and the variable argument is
omitted entirely, the comma will be removed from the expansion.  If the
variable argument is empty, or the token before @samp{##} is not a
comma, then @samp{##} behaves as a normal token paste.

@item @samp{#line} and @samp{#include}

The @samp{#line} directive used to change GCC's notion of the
``directory containing the current file,'' used by @samp{#include} with
a double-quoted header file name.  In 3.0 and later, it does not.
@xref{Line Control}, for further explanation.

@item Syntax of @samp{#line}

In GCC 2.95 and previous, the string constant argument to @samp{#line}
was treated the same way as the argument to @samp{#include}: backslash
escapes were not honored, and the string ended at the second @samp{"}.
This is not compliant with the C standard.  In GCC 3.0, an attempt was
made to correct the behavior, so that the string was treated as a real
string constant, but it turned out to be buggy.  In 3.1, the bugs have
been fixed.  (We are not fixing the bugs in 3.0 because they affect
relatively few people and the fix is quite invasive.)

@end itemize

@node Invocation
@chapter Invocation
@cindex invocation
@cindex command line

Most often when you use the C preprocessor you will not have to invoke it
explicitly: the C compiler will do so automatically.  However, the
preprocessor is sometimes useful on its own.  All the options listed
here are also acceptable to the C compiler and have the same meaning,
except that the C compiler has different rules for specifying the output
file.

@strong{Note:} Whether you use the preprocessor by way of @command{gcc}
or @command{cpp}, the @dfn{compiler driver} is run first.  This
program's purpose is to translate your command into invocations of the
programs that do the actual work.  Their command line interfaces are
similar but not identical to the documented interface, and may change
without notice.

@ignore
@c man begin SYNOPSIS
cpp [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
    [@option{-I}@var{dir}@dots{}] [@option{-W}@var{warn}@dots{}]
    [@option{-M}|@option{-MM}] [@option{-MG}] [@option{-MF} @var{filename}]
    [@option{-MP}] [@option{-MQ} @var{target}@dots{}]
    [@option{-MT} @var{target}@dots{}]
    [@option{-P}] [@option{-fno-working-directory}] 
    [@option{-x} @var{language}] [@option{-std=}@var{standard}]
    @var{infile} @var{outfile}

Only the most useful options are listed here; see below for the remainder.
@c man end
@c man begin SEEALSO
gpl(7), gfdl(7), fsf-funding(7),
gcc(1), as(1), ld(1), and the Info entries for @file{cpp}, @file{gcc}, and
@file{binutils}.
@c man end
@end ignore

@c man begin OPTIONS
The C preprocessor expects two file names as arguments, @var{infile} and
@var{outfile}.  The preprocessor reads @var{infile} together with any
other files it specifies with @samp{#include}.  All the output generated
by the combined input files is written in @var{outfile}.

Either @var{infile} or @var{outfile} may be @option{-}, which as
@var{infile} means to read from standard input and as @var{outfile}
means to write to standard output.  Also, if either file is omitted, it
means the same as if @option{-} had been specified for that file.

Unless otherwise noted, or the option ends in @samp{=}, all options
which take an argument may have that argument appear either immediately
after the option, or with a space between option and argument:
@option{-Ifoo} and @option{-I foo} have the same effect.

@cindex grouping options
@cindex options, grouping
Many options have multi-letter names; therefore multiple single-letter
options may @emph{not} be grouped: @option{-dM} is very different from
@w{@samp{-d -M}}.

@cindex options
@include cppopts.texi
@c man end

@node Environment Variables
@chapter Environment Variables
@cindex environment variables
@c man begin ENVIRONMENT

This section describes the environment variables that affect how CPP
operates.  You can use them to specify directories or prefixes to use
when searching for include files, or to control dependency output.

Note that you can also specify places to search using options such as
@option{-I}, and control dependency output with options like
@option{-M} (@pxref{Invocation}).  These take precedence over
environment variables, which in turn take precedence over the
configuration of GCC@.
 
@include cppenv.texi
@c man end

@page
@include fdl.texi

@page
@node Index of Directives
@unnumbered Index of Directives
@printindex fn

@node Option Index
@unnumbered Option Index
@noindent
CPP's command line options and environment variables are indexed here
without any initial @samp{-} or @samp{--}.
@printindex op

@page
@node Concept Index
@unnumbered Concept Index
@printindex cp

@bye