* Makefile.in (removenullpaths): Remove adjacent null paths.

[emacs.git] / doc / misc / ebrowse.texi

\input texinfo   @c -*-texinfo-*-

@comment %**start of header
@setfilename ../../info/ebrowse
@settitle A Class Browser for C++
@setchapternewpage odd
@syncodeindex fn cp
@comment %**end of header

@copying
This file documents Ebrowse, a C++ class browser for GNU Emacs.

Copyright @copyright{} 2000--2013 Free Software Foundation, Inc.

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

(a) The FSF's Back-Cover Text is: ``You have the freedom to copy and
modify this GNU manual.''
@end quotation
@end copying

@dircategory Emacs misc features
@direntry
* Ebrowse: (ebrowse).           A C++ class browser for Emacs.
@end direntry

@titlepage
@title Ebrowse User's Manual
@sp 4
@subtitle Ebrowse/Emacs
@sp 5
@author Gerd Moellmann
@page
@vskip 0pt plus 1filll
@insertcopying
@end titlepage

@contents

@ifnottex
@node Top
@top Ebrowse

You can browse C++ class hierarchies from within Emacs by using
Ebrowse.

@insertcopying
@end ifnottex

@menu
* Overview::                    What is it and how does it work?
* Generating browser files::    How to process C++ source files
* Loading a Tree::              How to start browsing
* Tree Buffers::                Traversing class hierarchies
* Member Buffers::              Looking at member information
* Tags-like Functions::         Finding members from source files
* GNU Free Documentation License:: The license for this documentation.
* Concept Index::               An entry for each concept defined
@end menu




@node Overview
@chapter Introduction

When working in software projects using C++, I frequently missed
software support for two things:

@itemize @bullet
@item
When you get a new class library, or you have to work on source code you
haven't written yourself (or written sufficiently long ago), you need a
tool to let you navigate class hierarchies and investigate
features of the software.  Without such a tool you often end up
@command{grep}ing through dozens or even hundreds of files.

@item
Once you are productive, it would be nice to have a tool that knows your
sources and can help you while you are editing source code.  Imagine to
be able to jump to the definition of an identifier while you are
editing, or something that can complete long identifier names because it
knows what identifiers are defined in your program@dots{}.
@end itemize

The design of Ebrowse reflects these two needs.

How does it work?

@cindex parser for C++ sources
A fast parser written in C is used to process C++ source files.
The parser generates a data base containing information about classes,
members, global functions, defines, types etc.@: found in the sources.

The second part of Ebrowse is a Lisp program.  This program reads
the data base generated by the parser.  It displays its contents in
various forms and allows you to perform operations on it, or do
something with the help of the knowledge contained in the data base.

@cindex major modes, of Ebrowse buffers
@dfn{Navigational} use of Ebrowse is centered around two
types of buffers which define their own major modes:

@cindex tree buffer
@dfn{Tree buffers} are used to view class hierarchies in tree form.
They allow you to quickly find classes, find or view class declarations,
perform operations like query replace on sets of your source files, and
finally tree buffers are used to produce the second buffer form---member
buffers.  @xref{Tree Buffers}.

@cindex member buffer
Members are displayed in @dfn{member buffers}.  Ebrowse
distinguishes between six different types of members; each type is
displayed as a member list of its own:

@itemize @bullet
@item
Instance member variables;

@item
Instance member functions;

@item
Static member variables;

@item
Static member functions;

@item
Friends/Defines.  The list of defines is contained in the friends
list of the pseudo-class @samp{*Globals*};

@item
Types (@code{enum}s, and @code{typedef}s defined with class
scope).@refill
@end itemize

You can switch member buffers from one list to another, or to another
class.  You can include inherited members in the display, you can set
filters that remove categories of members from the display, and most
importantly you can find or view member declarations and definitions
with a keystroke.  @xref{Member Buffers}.

These two buffer types and the commands they provide support the
navigational use of the browser.  The second form resembles Emacs's Tags
package for C and other procedural languages.  Ebrowse's commands of
this type are not confined to special buffers; they are most often used
while you are editing your source code.

To list just a subset of what you can use the Tags part of Ebrowse for:

@itemize @bullet
@item
Jump to the definition or declaration of an identifier in your source
code, with an electric position stack that lets you easily navigate
back and forth.

@item
Complete identifiers in your source with a completion list containing
identifiers from your source code only.

@item
Perform search and query replace operations over some or all of your
source files.

@item
Show all identifiers matching a regular expression---and jump to one of
them, if you like.
@end itemize




@node Generating browser files
@chapter Processing Source Files

@cindex @command{ebrowse}, the program
@cindex class data base creation
Before you can start browsing a class hierarchy, you must run the parser
@command{ebrowse} on your source files in order to generate a Lisp data
base describing your program.

@cindex command line for @command{ebrowse}
The operation of @command{ebrowse} can be tailored with command line
options.  Under normal circumstances it suffices to let the parser use
its default settings.  If you want to do that, call it with a command
line like:

@example
ebrowse *.h *.cc
@end example

@noindent
or, if your shell doesn't allow all the file names to be specified on
the command line,

@example
ebrowse --files=@var{file}
@end example

@noindent
where @var{file} contains the names of the files to be parsed, one
per line.

@findex --help
When invoked with option @samp{--help}, @command{ebrowse} prints a list of
available command line options.@refill

@menu
* Input files::         Specifying which files to parse
* Output file::         Changing the output file name
* Structs and unions::  Omitting @code{struct}s and @code{union}s
* Matching::            Setting regular expression lengths
* Verbosity::           Getting feedback for lengthy operations
@end menu




@comment name,     next,        prev,                     up
@node Input files
@section Specifying Input Files

@table @samp
@cindex input files, for @command{ebrowse}
@item file
Each file name on the command line tells @command{ebrowse} to parse
that file.

@cindex response files
@findex --files
@item --files=@var{file}
This command line switch specifies that @var{file} contains a list of
file names to parse.  Each line in @var{file} must contain one file
name.  More than one option of this kind is allowed.  You might, for
instance, want to use one file for header files, and another for source
files.

@cindex standard input, specifying input files
@item standard input
When @command{ebrowse} finds no file names on the command line, and no
@samp{--file} option is specified, it reads file names from standard
input.  This is sometimes convenient when @command{ebrowse} is used as part
of a command pipe.

@findex --search-path
@item --search-path=@var{paths}
This option lets you specify search paths for your input files.
@var{paths} is a list of directory names, separated from each other by a
either a colon or a semicolon, depending on the operating system.
@end table

@cindex header files
@cindex friend functions
It is generally a good idea to specify input files so that header files
are parsed before source files.  This facilitates the parser's work of
properly identifying friend functions of a class.



@comment name,     next,               prev,        up
@node Output file
@section Changing the Output File Name

@table @samp
@cindex output file name
@findex --output-file
@cindex @file{BROWSE} file
@item --output-file=@var{file}
This option instructs @command{ebrowse} to generate a Lisp data base with
name @var{file}.  By default, the data base is named @file{BROWSE}, and
is written in the directory in which @command{ebrowse} is invoked.

If you regularly use data base names different from the default, you
might want to add this to your init file:

@lisp
(add-to-list 'auto-mode-alist '(@var{NAME} . ebrowse-tree-mode))
@end lisp

@noindent
where @var{NAME} is the Lisp data base name you are using.

@findex --append
@cindex appending output to class data base
@item --append
By default, each run of @command{ebrowse} erases the old contents of the
output file when writing to it.  You can instruct @command{ebrowse} to
append its output to an existing file produced by @command{ebrowse}
with this command line option.
@end table




@comment name,            next,     prev,        up
@node Structs and unions
@section Structs and Unions
@cindex structs
@cindex unions

@table @samp
@findex --no-structs-or-unions
@item --no-structs-or-unions
This switch suppresses all classes in the data base declared as
@code{struct} or @code{union} in the output.

This is mainly useful when you are converting an existing
C program to C++, and do not want to see the old C structs in a class
tree.
@end table




@comment name,  next,      prev,               up
@node Matching
@section Regular Expressions

@cindex regular expressions, recording
The parser @command{ebrowse} normally writes regular expressions to its
output file that help the Lisp part of Ebrowse to find functions,
variables etc.@: in their source files.

You can instruct @command{ebrowse} to omit these regular expressions by
calling it with the command line switch @samp{--no-regexps}.

When you do this, the Lisp part of Ebrowse tries to guess, from member
or class names, suitable regular expressions to locate that class or
member in source files.  This works fine in most cases, but the
automatic generation of regular expressions can be too weak if unusual
coding styles are used.

@table @samp
@findex --no-regexps
@item --no-regexps
This option turns off regular expression recording.

@findex --min-regexp-length
@cindex minimum regexp length for recording
@item --min-regexp-length=@var{n}
The number @var{n} following this option specifies the minimum length of
the regular expressions recorded to match class and member declarations
and definitions.  The default value is set at compilation time of
@command{ebrowse}.

The smaller the minimum length, the higher the probability that
Ebrowse will find a wrong match.  The larger the value, the
larger the output file and therefore the memory consumption once the
file is read from Emacs.

@findex --max-regexp-length
@cindex maximum regexp length for recording
@item --max-regexp-length=@var{n}
The number following this option specifies the maximum length of the
regular expressions used to match class and member declarations and
definitions.  The default value is set at compilation time of
@command{ebrowse}.

The larger the maximum length, the higher the probability that the
browser will find a correct match, but the larger the value the larger
the output file and therefore the memory consumption once the data is
read.  As a second effect, the larger the regular expression, the higher
the probability that it will no longer match after editing the file.
@end table




@node Verbosity
@section Verbose Mode
@cindex verbose operation

@table @samp
@findex --verbose
@item --verbose
When this option is specified on the command line, @command{ebrowse} prints
a period for each file parsed, and it displays a @samp{+} for each
class written to the output file.

@findex --very-verbose
@item --very-verbose
This option makes @command{ebrowse} print out the names of the files and
the names of the classes seen.
@end table




@node Loading a Tree
@chapter Starting to Browse
@cindex loading
@cindex browsing

You start browsing a class hierarchy parsed by @command{ebrowse} by just
finding the @file{BROWSE} file with @kbd{C-x C-f}.

An example of a tree buffer display is shown below.

@example
|  Collection
|    IndexedCollection
|      Array
|        FixedArray
|    Set
|    Dictionary
@end example

@cindex mouse highlight in tree buffers
When you run Emacs on a display which supports colors and the mouse, you
will notice that certain areas in the tree buffer are highlighted
when you move the mouse over them.  This highlight marks mouse-sensitive
regions in the buffer.  Please notice the help strings in the echo area
when the mouse moves over a sensitive region.

@cindex context menu
A click with @kbd{Mouse-3} on a mouse-sensitive region opens a context
menu.  In addition to this, each buffer also has a buffer-specific menu
that is opened with a click with @kbd{Mouse-3} somewhere in the buffer
where no highlight is displayed.



@comment ****************************************************************
@comment ***
@comment ***                 TREE BUFFERS
@comment ***
@comment ****************************************************************

@node Tree Buffers
@chapter Tree Buffers
@cindex tree buffer mode
@cindex class trees

Class trees are displayed in @dfn{tree buffers} which install their own
major mode.  Most Emacs keys work in tree buffers in the usual way,
e.g., you can move around in the buffer with the usual @kbd{C-f},
@kbd{C-v} etc., or you can search with @kbd{C-s}.

Tree-specific commands are bound to simple keystrokes, similar to
@code{Gnus}.  You can take a look at the key bindings by entering
@kbd{?} which calls @code{M-x describe-mode} in both tree and member
buffers.

@menu
* Source Display::              Viewing and finding a class declaration
* Member Display::              Showing members, switching to member buffers
* Go to Class::                 Finding a class
* Quitting::                    Discarding and burying the tree buffer
* File Name Display::           Showing file names in the tree
* Expanding and Collapsing::    Expanding and collapsing branches
* Tree Indentation::            Changing the tree indentation
* Killing Classes::             Removing class from the tree
* Saving a Tree::               Saving a modified tree
* Statistics::                  Displaying class tree statistics
* Marking Classes::             Marking and unmarking classes
@end menu



@node Source Display
@section Viewing and Finding Class Declarations
@cindex viewing, class
@cindex finding a class
@cindex class declaration

You can view or find a class declaration when the cursor is on a class
name.

@table @kbd
@item SPC
This command views the class declaration if the database
contains information about it.  If you don't parse the entire source
you are working on, some classes will only be known to exist but the
location of their declarations and definitions will not be known.@refill

@item RET
Works like @kbd{SPC}, except that it finds the class
declaration rather than viewing it, so that it is ready for
editing.@refill
@end table

The same functionality is available from the menu opened with
@kbd{Mouse-3} on the class name.




@node Member Display
@section Displaying Members
@cindex @samp{*Members*} buffer
@cindex @samp{*Globals*}
@cindex freezing a member buffer
@cindex member lists, in tree buffers

Ebrowse distinguishes six different kinds of members, each of
which is displayed as a separate @dfn{member list}: instance variables,
instance functions, static variables, static functions, friend
functions, and types.

Each of these lists can be displayed in a member buffer with a command
starting with @kbd{L} when the cursor is on a class name.  By default,
there is only one member buffer named @dfn{*Members*} that is reused
each time you display a member list---this has proven to be more
practical than to clutter up the buffer list with dozens of member
buffers.

If you want to display more than one member list at a time you can
@dfn{freeze} its member buffer. Freezing a member buffer prevents it
from being overwritten the next time you display a member list. You can
toggle this buffer status at any time.

Every member list display command in the tree buffer can be used with a
prefix argument (@kbd{C-u}).  Without a prefix argument, the command will
pop to a member buffer displaying the member list.  With prefix argument,
the member buffer will additionally be @dfn{frozen}.

@table @kbd
@cindex instance member variables, list
@item L v
This command displays the list of instance member variables.

@cindex static variables, list
@item L V
Display the list of static variables.

@cindex friend functions, list
@item L d
Display the list of friend functions.  This list is used for defines if
you are viewing the class @samp{*Globals*} which is a place holder for
global symbols.

@cindex member functions, list
@item L f
Display the list of member functions.

@cindex static member functions, list
@item L F
Display the list of static member functions.

@cindex types, list
@item L t
Display a list of types.
@end table

These lists are also available from the class' context menu invoked with
@kbd{Mouse-3} on the class name.




@node Go to Class
@section Finding a Class
@cindex locate class
@cindex expanding branches
@cindex class location

@table @kbd
@cindex search for class
@item /
This command reads a class name from the minibuffer with completion and
positions the cursor on the class in the class tree.

If the branch of the class tree containing the class searched for is
currently collapsed, the class itself and all its base classes are
recursively made visible.  (See also @ref{Expanding and
Collapsing}.)@refill

This function is also available from the tree buffer's context menu.

@item n
Repeat the last search done with @kbd{/}.  Each tree buffer has its own
local copy of the regular expression last searched in it.
@end table




@node Quitting
@section Burying a Tree Buffer
@cindex burying tree buffer

@table @kbd
@item q
Is a synonym for @kbd{M-x bury-buffer}.
@end table




@node File Name Display
@section Displaying File Names

@table @kbd
@cindex file names in tree buffers
@item T f
This command toggles the display of file names in a tree buffer.  If
file name display is switched on, the names of the files containing the
class declaration are shown to the right of the class names.  If the
file is not known, the string @samp{unknown} is displayed.

This command is also provided in the tree buffer's context menu.

@item s
Display file names for the current line, or for the number of lines
given by a prefix argument.
@end table

Here is an example of a tree buffer with file names displayed.

@example
|  Collection           (unknown)
|    IndexedCollection  (indexedcltn.h)
|      Array            (array.h)
|        FixedArray     (fixedarray.h)
|    Set                (set.h)
|    Dictionary         (dict.h)
@end example


@node Expanding and Collapsing
@section Expanding and Collapsing a Tree
@cindex expand tree branch
@cindex collapse tree branch
@cindex branches of class tree
@cindex class tree, collapse or expand

You can expand and collapse parts of a tree to reduce the complexity of
large class hierarchies.  Expanding or collapsing branches of a tree has
no impact on the functionality of other commands, like @kbd{/}.  (See
also @ref{Go to Class}.)@refill

Collapsed branches are indicated with an ellipsis following the class
name like in the example below.

@example
|  Collection
|    IndexedCollection...
|    Set
|    Dictionary
@end example

@table @kbd
@item -
This command collapses the branch of the tree starting at the class the
cursor is on.

@item +
This command expands the branch of the tree starting at the class the
cursor is on.  Both commands for collapsing and expanding branches are
also available from the class' object menu.

@item *
This command expands all collapsed branches in the tree.
@end table




@node Tree Indentation
@section Changing the Tree Indentation
@cindex tree indentation
@cindex indentation of the tree

@table @kbd
@item T w
This command reads a new indentation width from the minibuffer and
redisplays the tree buffer with the new indentation  It is also
available from the tree buffer's context menu.
@end table




@node Killing Classes
@section Removing Classes from the Tree
@cindex killing classes
@cindex class, remove from tree

@table @kbd
@item C-k
This command removes the class the cursor is on and all its derived
classes from the tree.  The user is asked for confirmation before the
deletion is actually performed.
@end table




@node Saving a Tree
@section Saving a Tree
@cindex save tree to a file
@cindex tree, save to a file
@cindex class tree, save to a file

@table @kbd
@item C-x C-s
This command writes a class tree to the file from which it was read.
This is useful after classes have been deleted from a tree.

@item  C-x C-w
Writes the tree to a file whose name is read from the minibuffer.
@end table




@node     Statistics
@section Statistics
@cindex statistics for a tree
@cindex tree statistics
@cindex class statistics

@table @kbd
@item x
Display statistics for the tree, like number of classes in it, number of
member functions, etc.  This command can also be found in the buffer's
context menu.
@end table




@node     Marking Classes
@section Marking Classes
@cindex marking classes
@cindex operations on marked classes

Classes can be marked for operations similar to the standard Emacs
commands @kbd{M-x tags-search} and @kbd{M-x tags-query-replace} (see
also @xref{Tags-like Functions}.)@refill

@table @kbd
@cindex toggle mark
@item M t
Toggle the mark of the line point is in or for as many lines as given by
a prefix command.  This command can also be found in the class' context
menu.

@cindex unmark all
@item M a
Unmark all classes.  With prefix argument @kbd{C-u}, mark all classes in
the tree. Since this command operates on the whole buffer, it can also be
found in the buffer's object menu.
@end table

Marked classes are displayed with an @code{>} in column one of the tree
display, like in the following example

@example
|> Collection
|    IndexedCollection...
|>   Set
|    Dictionary
@end example




@c ****************************************************************
@c ***
@c ***                 MEMBER BUFFERS
@c ***
@c ****************************************************************

@node Member Buffers
@chapter Member Buffers
@cindex members
@cindex member buffer mode

@cindex class members, types
@cindex types of class members
@dfn{Member buffers} are used to operate on lists of members of a class.
Ebrowse distinguishes six kinds of lists:

@itemize @bullet
@item
Instance variables (normal member variables);
@item
Instance functions (normal member functions);
@item
Static variables;
@item
Static member functions;
@item
Friend functions;
@item
Types (@code{enum}s and @code{typedef}s defined with class scope.
Nested classes will be shown in the class tree like normal classes.
@end itemize

Like tree buffers, member buffers install their own major mode.  Also
like in tree buffers, menus are provided for certain areas in the
buffer: members, classes, and the buffer itself.

@menu
* Switching Member Lists::      Choosing which members to display
* Finding/Viewing::             Modifying source code
* Inherited Members::           Display of Inherited Members
* Searching Members::           Finding members in member buffer
* Switching to Tree::           Going back to the tree buffer
* Filters::                     Selective member display
* Attributes::                  Display of @code{virtual} etc.
* Long and Short Display::      Comprehensive and verbose display
* Regexp Display::              Showing matching regular expressions
* Switching Classes::           Displaying another class
* Killing/Burying::             Getting rid of the member buffer
* Column Width::                Display style
* Redisplay::                   Redrawing the member list
* Getting Help::                How to get help for key bindings
@end menu




@node Switching Member Lists
@section Switching Member Lists
@cindex member lists, in member buffers
@cindex static members
@cindex friends
@cindex types
@cindex defines

@table @kbd
@cindex next member list
@item L n
This command switches the member buffer display to the next member list.

@cindex previous member list
@item L p
This command switches the member buffer display to the previous member
list.

@item L f
Switch to the list of member functions.

@cindex static
@item L F
Switch to the list of static member functions.

@item L v
Switch to the list of member variables.

@item L V
Switch to the list of static member variables.

@item L d
Switch to the list of friends or defines.

@item L t
Switch to the list of types.
@end table

Both commands cycle through the member list.

Most of the commands are also available from the member buffer's
context menu.




@node Finding/Viewing
@section Finding and Viewing Member Source
@cindex finding members, in member buffers
@cindex viewing members, in member buffers
@cindex member definitions, in member buffers
@cindex member declarations, in member buffers
@cindex definition of a member, in member buffers
@cindex declaration of a member, in member buffers

@table @kbd
@item RET
This command finds the definition of the member the cursor is on.
Finding involves roughly the same as the standard Emacs tags facility
does---loading the file and searching for a regular expression matching
the member.

@item f
This command finds the declaration of the member the cursor is on.

@item SPC
This is the same command as @kbd{RET}, but views the member definition
instead of finding the member's source file.

@item v
This is the same command as @kbd{f}, but views the member's declaration
instead of finding the file the declaration is in.
@end table

You can install a hook function to perform actions after a member or
class declaration or definition has been found, or when it is not found.

All the commands described above can also be found in the context menu
displayed when clicking @kbd{Mouse-2} on a member name.




@node Inherited Members
@section Display of Inherited Members
@cindex superclasses, members
@cindex base classes, members
@cindex inherited members

@table @kbd
@item D b
This command toggles the display of inherited members in the member
buffer.  This is also in the buffer's context menu.
@end table




@node Searching Members
@section Searching Members
@cindex searching members

@table @kbd
@item G v
Position the cursor on a member whose name is read from the minibuffer;
only members shown in the current member buffer appear in the completion
list.

@item G m
Like the above command, but all members for the current class appear in
the completion list.  If necessary, the current member list is switched
to the one containing the member.

With a prefix argument (@kbd{C-u}), all members in the class tree,
i.e., all members the browser knows about appear in the completion
list.  The member display will be switched to the class and member list
containing the member.

@item G n
Repeat the last member search.
@end table

Look into the buffer's context menu for a convenient way to do this with
a mouse.



@node Switching to Tree
@section Switching to Tree Buffer
@cindex tree buffer, switch to
@cindex buffer switching
@cindex switching buffers

@table @kbd
@item @key{TAB}
Pop up the tree buffer to which the member buffer belongs.

@item t
Do the same as @key{TAB} but also position the cursor on the class
displayed in the member buffer.
@end table




@node Filters
@section Filters
@cindex filters

@table @kbd
@cindex @code{public} members
@item F a u
This command toggles the display of @code{public} members.  The
@samp{a} stands for `access'.

@cindex @code{protected} members
@item F a o
This command toggles the display of @code{protected} members.

@cindex @code{private} members
@item F a i
This command toggles the display of @code{private} members.

@cindex @code{virtual} members
@item F v
This command toggles the display of @code{virtual} members.

@cindex @code{inline} members
@item F i
This command toggles the display of @code{inline} members.

@cindex @code{const} members
@item F c
This command toggles the display of @code{const} members.

@cindex pure virtual members
@item F p
This command toggles the display of pure virtual members.

@cindex remove filters
@item F r
This command removes all filters.
@end table

These commands are also found in the buffer's context menu.




@node Attributes
@section Displaying Member Attributes
@cindex attributes
@cindex member attribute display

@table @kbd
@item D a
Toggle the display of member attributes (default is on).

The nine member attributes Ebrowse knows about are displayed
as a list a single-characters flags enclosed in angle brackets in front
the of the member's name.  A @samp{-} at a given position means that
the attribute is false.  The list of attributes from left to right is

@table @samp
@cindex @code{template} attribute
@item T
The member is a template.

@cindex @code{extern "C"} attribute
@item C
The member is declared @code{extern "C"}.

@cindex @code{virtual} attribute
@item v
Means the member is declared @code{virtual}.

@cindex @code{inline}
@item i
The member is declared @code{inline}.

@cindex @code{const} attribute
@item c
The member is @code{const}.

@cindex pure virtual function attribute
@item 0
The member is a pure virtual function.

@cindex @code{mutable} attribute
@item m
The member is declared @code{mutable}.

@cindex @code{explicit} attribute
@item e
The member is declared @code{explicit}.

@item t
The member is a function with a throw list.
@end table
@end table

This command is also in the buffer's context menu.



@node Long and Short Display
@section Long and Short Member Display
@cindex display form
@cindex long display
@cindex short display

@table @kbd
@item D l
This command toggles the member buffer between short and long display
form.  The short display form displays member names, only:

@example
| isEmpty        contains       hasMember      create
| storeSize      hash           isEqual        restoreGuts
| saveGuts
@end example

The long display shows one member per line with member name and regular
expressions matching the member (if known):

@example
| isEmpty               Bool isEmpty () const...
| hash                  unsigned hash () const...
| isEqual               int isEqual (...
@end example

Regular expressions will only be displayed when the Lisp database has
not been produced with the @command{ebrowse} option @samp{--no-regexps}.
@xref{Matching, --no-regexps, Regular Expressions}.
@end table




@node Regexp Display
@section Display of Regular Expressions
@cindex regular expression display

@table @kbd
@item D r
This command toggles the long display form from displaying the regular
expressions matching the member declarations to those expressions
matching member definitions.
@end table

Regular expressions will only be displayed when the Lisp database has
not been produced with the @command{ebrowse} option @samp{--no-regexps},
see @ref{Matching, --no-regexps, Regular Expressions}.




@node Switching Classes
@section Displaying Another Class
@cindex base class, display
@cindex derived class, display
@cindex superclass, display
@cindex subclass, display
@cindex class display

@table @kbd
@item C c
This command lets you switch the member buffer to another class.  It
reads the name of the new class from the minibuffer with completion.

@item C b
This is the same command as @kbd{C c} but restricts the classes shown in
the completion list to immediate base classes, only.  If only one base
class exists, this one is immediately shown in the minibuffer.

@item C d
Same as @kbd{C b}, but for derived classes.

@item C p
Switch to the previous class in the class hierarchy on the same level as
the class currently displayed.

@item C n
Switch to the next sibling of the class in the class tree.
@end table




@node Killing/Burying
@section Burying a Member Buffer
@cindex burying member buffers

@table @kbd
@item q
This command is a synonym for @kbd{M-x bury-buffer}.
@end table




@node Column Width
@section Setting the Column Width
@cindex column width
@cindex member indentation
@cindex indentation, member

@table @kbd
@item D w
This command sets the column width depending on the display form used
(long or short display).
@end table




@node Redisplay
@section Forced Redisplay
@cindex redisplay of member buffers

@table @kbd
@item C-l
This command forces a redisplay of the member buffer.  If the width
of the window displaying the member buffer is changed this command
redraws the member list with the appropriate column widths and number of
columns.
@end table




@node Getting Help
@section Getting Help
@cindex help

@table @kbd
@item ?
This key is bound to @code{describe-mode}.
@end table




@comment **************************************************************
@comment ***                TAGS LIKE FUNCTIONS
@comment **************************************************************

@node Tags-like Functions
@chapter Tags-like Functions

Ebrowse provides tags functions similar to those of the standard
Emacs Tags facility, but better suited to the needs of C++ programmers.

@menu
* Finding and Viewing::   Going to a member declaration/definition
* Position Stack::        Moving to previous locations
* Search & Replace::      Searching and replacing over class tree files
* Members in Files::      Listing all members in a given file
* Apropos::               Listing members matching a regular expression
* Symbol Completion::     Completing names while editing
* Member Buffer Display:: Quickly display a member buffer for some
                            identifier
@end menu



@node Finding and Viewing
@section Finding and Viewing Members
@cindex finding class member, in C++ source
@cindex viewing class member, in C++ source
@cindex tags
@cindex member definition, finding, in C++ source
@cindex member declaration, finding, in C++ source

The functions in this section are similar to those described in
@ref{Source Display}, and also in @ref{Finding/Viewing}, except that
they work in a C++ source buffer, not in member and tree buffers created
by Ebrowse.

@table @kbd
@item C-c C-m f
Find the definition of the member around point.  If you invoke this
function with a prefix argument, the declaration is searched.

If more than one class contains a member with the given name you can
select the class with completion.  If there is a scope declaration in
front of the member name, this class name is used as initial input for
the completion.

@item C-c C-m F
Find the declaration of the member around point.

@item C-c C-m v
View the definition of the member around point.

@item C-c C-m V
View the declaration of the member around point.

@item C-c C-m 4 f
Find a member's definition in another window.

@item C-c C-m 4 F
Find a member's declaration in another window.

@item C-c C-m 4 v
View a member's definition in another window.

@item C-c C-m 4 V
View a member's declaration in another window.

@item C-c C-m 5 f
Find a member's definition in another frame.

@item C-c C-m 5 F
Find a member's declaration in another frame.

@item C-c C-m 5 v
View a member's definition in another frame.

@item C-c C-m 5 V
View a member's declaration in another frame.
@end table



@node Position Stack
@section The Position Stack
@cindex position stack

When jumping to a member declaration or definition with one of
Ebrowse's commands, the position from where you performed the
jump and the position where you jumped to are recorded in a
@dfn{position stack}.  There are several ways in which you can quickly
move to positions in the stack:@refill

@table @kbd
@cindex return to original position
@item C-c C-m -
This command sets point to the previous position in the position stack.
Directly after you performed a jump, this will put you back to the
position where you came from.

The stack is not popped, i.e., you can always switch back and forth
between positions in the stack.  To avoid letting the stack grow to
infinite size there is a maximum number of positions defined.  When this
number is reached, older positions are discarded when new positions are
pushed on the stack.

@item C-c C-m +
This command moves forward in the position stack, setting point to
the next position stored in the position stack.

@item C-c C-m p
Displays an electric buffer showing all positions saved in the stack.
You can select a position by pressing @kbd{SPC} in a line.  You can
view a position with @kbd{v}.
@end table




@node Search & Replace
@section Searching and Replacing
@cindex searching multiple C++ files
@cindex replacing in multiple C++ files
@cindex restart tags-operation

Ebrowse allows you to perform operations on all or a subset of the files
mentioned in a class tree.  When you invoke one of the following
functions and more than one class tree is loaded, you must choose a
class tree to use from an electric tree menu.  If the selected tree
contains marked classes, the following commands operate on the files
mentioned in the marked classes only.  Otherwise all files in the class
tree are used.

@table @kbd
@item C-c C-m s
This function performs a regular expression search in the chosen set of
files.

@item C-c C-m u
This command performs a search for calls of a given member which is
selected in the usual way with completion.

@item C-c C-m %
Perform a query replace over the set of files.

@item C-c C-m ,
All three operations above stop when finding a match.  You can restart
the operation with this command.

@item C-c C-m n
This restarts the last tags operation with the next file in the list.
@end table




@node Members in Files
@section Members in Files
@cindex files
@cindex members in file, listing
@cindex list class members in a file
@cindex file, members

The command @kbd{C-c C-m l}, lists all members in a given file.  The file
name is read from the minibuffer with completion.




@node Apropos
@section Member Apropos
@cindex apropos on class members
@cindex members, matching regexp

The command @kbd{C-c C-m a} can be used to display all members matching a
given regular expression.  This command can be very useful if you
remember only part of a member name, and not its beginning.

A special buffer is popped up containing all identifiers matching the
regular expression, and what kind of symbol it is (e.g., a member
function, or a type).  You can then switch to this buffer, and use the
command @kbd{C-c C-m f}, for example, to jump to a specific member.




@node Symbol Completion
@section Symbol Completion
@cindex completion
@cindex symbol completion

The command @kbd{C-c C-m @key{TAB}} completes the symbol in front of point.




@node Member Buffer Display
@section Quick Member Display
@cindex member buffer, for member at point

You can quickly display a member buffer containing the member the cursor
in on with the command @kbd{C-c C-m m}.


@node GNU Free Documentation License
@appendix GNU Free Documentation License
@include doclicense.texi


@node Concept Index
@unnumbered Concept Index
@printindex cp

@bye