@c Copyright (C) 2012 John W. Eaton
@c Copyright (C) 2008-2013 Jaroslav Hajek
@c
@c This file is part of Octave.
@c
@c Octave is free software; you can redistribute it and/or modify it
@c under the terms of the GNU General Public License as published by the
@c Free Software Foundation; either version 3 of the License, or (at
@c your option) any later version.
@c
@c Octave is distributed in the hope that it will be useful, but WITHOUT
@c ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
@c FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
@c for more details.
@c
@c You should have received a copy of the GNU General Public License
@c along with Octave; see the file COPYING.  If not, see
@c <http://www.gnu.org/licenses/>.

@node Contributing Guidelines
@appendix Contributing Guidelines
@cindex coding standards
@cindex Octave development

This chapter is dedicated to those who wish to contribute code to Octave.

@menu
* How to Contribute::
* Building the Development Sources::
* Basics of Generating a Changeset::
* General Guidelines::
* Octave Sources (m-files)::
* C++ Sources::
* Other Sources::
@end menu

@node How to Contribute
@section How to Contribute
The mailing list for Octave development discussions is
@email{maintainers@@octave.org}.  Patches should be submitted to
@url{https://savannah.gnu.org/patch/?func=additem&group=octave, Octave's patch tracker}.
This concerns the development of Octave core, i.e., code that goes in to Octave
directly.  You may consider developing and publishing a package instead;
a great place for this is the allied Octave-Forge project
(@url{http://octave.sourceforge.net}).  Note that the Octave core
project is inherently more conservative and follows narrower rules.

@node Building the Development Sources
@section Building the Development Sources

The directions for building from the development sources change from
time to time, so you should read the resources for developers on the web
or in the development sources archive.  Start here:
@url{http://www.octave.org/get-involved.html}.

@node Basics of Generating a Changeset
@section Basics of Generating a Changeset

The best way to contribute is to create a Mercurial changeset and submit it to
the @url{http://savannah.gnu.org/bugs/?group=octave, bug} or
@url{http://savannah.gnu.org/patch/?func=additem&group=octave, patch}
trackers@footnote{Please use the patch tracker only for patches which add new
features.  If you have a patch to submit that fixes a bug, you should use the
bug tracker instead.}.
Mercurial is the source code management system currently used to develop
Octave.  Other forms of contributions (e.g., simple diff patches) are
also acceptable, but they slow down the review process.  If you want to
make more contributions, you should really get familiar with Mercurial.
A good place to start is
@url{http://www.selenic.com/mercurial/wiki/index.cgi/Tutorial}.  There
you will also find help about how to install Mercurial.

A simple contribution sequence could look like this:

@example
@group
hg clone http://www.octave.org/hg/octave
                             # make a local copy of the octave
                             # source repository
cd octave
# change some sources@dots{}
hg commit -m "make Octave the coolest software ever"
                             # commit the changeset into your
                             # local repository
hg export -o ../cool.diff tip
                             # export the changeset to a diff
                             # file
# attach ../cool.diff to your bug report
@end group
@end example

You may want to get familiar with Mercurial queues to manage your
changesets.  To work with queues you must activate the extension
@nospell{mq} with the following entry in Mercurial's configuration file
@file{.hgrc} (or @file{Mercurial.ini} on Windows):

@example
@group
[extensions]
mq=
@end group
@end example

Here is a slightly more complex example using Mercurial
queues, where work on two unrelated changesets is done in parallel and
one of the changesets is updated after discussion on the bug tracker:

@example
hg qnew nasty_bug            # create a new patch
# change sources@dots{}
hg qref                      # save the changes into the patch
# change even more@dots{}
hg qref -m "solution to nasty bug!"
                             # save again with commit message
hg export -o ../nasty.diff tip
                             # export the patch
# attach ../nasty.diff to your bug report
hg qpop                      # undo the application of the patch
                             # and remove the changes from the
                             # source tree
hg qnew doc_improvements     # create an unrelated patch
# change doc sources@dots{}
hg qref -m "could not find myfav.m in the doc"
                             # save the changes into the patch
hg export -o ../doc.diff tip
                             # export the second patch
# attach ../doc.diff to your bug report
hg qpop
# discussion in the bug tracker @dots{}
hg qpush nasty_bug           # apply the patch again
# change sources yet again @dots{}
hg qref
hg export -o ../nasty2.diff tip
# attach ../nasty2.diff to your bug report
@end example

Mercurial has a few more useful extensions that really should be enabled.
They are not enabled by default due to a number of factors
(mostly because they don't work in all terminal types).

The following entries in the @file{.hgrc} are recommended

@example
@group
[extensions]
graphlog=
color=
progress=
pager=
@end group
@end example

For the color extension, default color and formatting
of @code{hg status} can be modified by

@example
@group
[color]
status.modified = magenta bold
status.added = green bold
status.removed = red bold
status.deleted = cyan bold
status.unknown = black  bold
status.ignored = black bold
@end group
@end example

Sometimes a few further improvements for the pager extension are
necessary.  The following options should not be enabled unless paging
is not working correctly.

@example
@group
[pager]
# Some options for the less pager, see less(1) for their meaning.
pager = LESS='FSRX' less

# Some commands that aren't paged by default; also enable paging
# for them
attend = tags, help, annotate, cat, diff, export, status, \
         outgoing, incoming
@end group
@end example

Enabling the described extensions should immediately lead to a difference
when using the command line version of @nospell{hg}.  Of these options, the
only one that enables a new command is @nospell{graphlog}.  It is recommanded
that to use the command @code{hg glog}, instead of @code{hg log}, for a better
feel about what commits are being based on.

@node General Guidelines
@section General Guidelines

All Octave's sources are distributed under the GNU General Public License
(GPL).  Currently, Octave uses GPL version 3.  For details about this
license, see @url{http://www.gnu.org/licenses/gpl.html}.  Therefore,
whenever you create a new source file, it should have the following
comment header (use appropriate year, name and comment marks):

@example
## Copyright (C) 1996-2013 John W. Eaton <jwe@@octave.org>
##
## This file is part of Octave.
##
## Octave is free software; you can redistribute it and/or modify it
## under the terms of the GNU General Public License as published by
## the Free Software Foundation; either version 3 of the License, or
## (at your option) any later version.
##
## Octave is distributed in the hope that it will be useful, but
## WITHOUT ANY WARRANTY; without even the implied warranty of
## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
## GNU General Public License for more details.
##
## You should have received a copy of the GNU General Public License
## along with Octave; see the file COPYING.  If not,
## see <http://www.gnu.org/licenses/>.
@end example

Always include commit messages in changesets.  After making your source
changes, record and briefly describe the changes in your commit message.
You should have previously configured your @file{.hgrc} (or
@file{Mercurial.ini} on Windows) with your name and email, which will
be automatically added to your commit message.  Your commit message
should have a brief one-line explanation of what the commit does.  If you
are patching a bug, this one-line explanation should mention the bug
number at the end.  If your change is small and only touches one file then
this is typically sufficient.  If you are modifying several files, or
several parts of one file, you should enumerate your changes roughly
following the GNU coding standards for changelogs, as in the following
example:

@example
@group
look for methods before constructors

* symtab.cc (symbol_table::fcn_info::fcn_info_rep::find):
Look for class methods before constructors, contrary to @sc{matlab}
documentation.

* test/ctor-vs-method: New directory of test classes.
* test/test_ctor_vs_method.m: New file.
* test/Makefile.am: Include ctor-vs-method/module.mk.
(FCN_FILES): Include test_ctor_vs_method.m in the list.
@end group
@end example

@noindent
In this example, the names of the file changed is listed first, and in
parentheses the name of the function in that file that was modified.  There
is no need to mention the function for m-files that only contain one function.
The commit message should describe what was changed, not why it was changed.
Any explanation for why a change is needed should appear as comments in the
code, particularly if there is something that might not be obvious to someone
reading it later.

When submitting code which addresses a known bug on the Octave bug
tracker (@url{http://bugs.octave.org}), please add '(bug #XXXXX)' to the
first line of the commit messages.  For example:

@example
@group
Fix bug for complex input for gradient (bug #34292).
@end group
@end example

The preferred comment mark for places that may need further attention is
@code{FIXME:}.

@node Octave Sources (m-files)
@section Octave Sources (m-files)

Don't use tabs.  Tabs cause trouble.  If you are used to them, set up
your editor so that it converts tabs to spaces.  Indent the bodies of
statement blocks.  The recommended indent is 2 spaces.  When calling
functions, put spaces after commas and before the calling parentheses,
like this:

@example
  x = max (sin (y+3), 2);
@end example

@noindent
An exception are matrix or cell constructors:

@example
  [sin(x), cos(x)]
  @{sin(x), cos(x)@}
@end example

@noindent
Here, putting spaces after @code{sin}, @code{cos} would result in a
parse error.  For an indexing expression, do not put a space after the
identifier (this differentiates indexing and function calls nicely).
The space after a comma is not necessary if index expressions are simple,
i.e., you may write

@example
  A(:,i,j)
@end example

@noindent
but

@example
  A([1:i-1;i+1:n], XI(:,2:n-1))
@end example

Use lowercase names if possible.  Uppercase is acceptable for variable
names consisting of 1-2 letters.  Do not use mixed case names.  Function
names must be lowercase.  Function names are global, so choose them
wisely.

Always use a specific end-of-block statement (like @code{endif},
@code{endswitch}) rather than the generic @code{end}.

Enclose the @code{if}, @code{while}, @code{until}, and @code{switch}
conditions in parentheses, as in C:

@example
@group
if (isvector (a))
  s = sum (a);
endif
@end group
@end example

@noindent
Do not do this, however, with the iteration counter portion of a
@code{for} statement.  Write:

@example
@group
for i = 1:n
  b(i) = sum (a(:,i));
endfor
@end group
@end example

@node C++ Sources
@section C++ Sources

Don't use tabs.  Tabs cause trouble.  If you are used to them, set up
your editor so that it converts tabs to spaces.  Format function headers
like this:

@example
@group
static bool
matches_patterns (const string_vector& patterns, int pat_idx,
                  int num_pat, const std::string& name)
@end group
@end example

@noindent
The function name should start in column 1, and multi-line argument
lists should be aligned on the first char after the open parenthesis.
You should put a space before the left open parenthesis and after commas,
for both function definitions and function calls.

The recommended indent is 2 spaces.  When indenting, indent the statement
after control structures (like @code{if}, @code{while}, etc.).  If there
is a compound statement, indent @emph{both} the curly braces and the
body of the statement (so that the body gets indented by @emph{two}
indents).  Example:

@example
@group
if (have_args)
  @{
    idx.push_back (first_args);
    have_args = false;
  @}
else
  idx.push_back (make_value_list (*p_args, *p_arg_nm, &tmp));
@end group
@end example

@noindent
If you have nested @code{if} statements, use extra braces for extra
clarification.

Split long expressions in such a way that a continuation line starts
with an operator rather than identifier.  If the split occurs inside
braces, continuation should be aligned with the first char after the
innermost braces enclosing the split.  Example:

@example
@group
SVD::type type = ((nargout == 0 || nargout == 1)
                  ? SVD::sigma_only
                  : (nargin == 2) ? SVD::economy : SVD::std);
@end group
@end example

@noindent
Consider putting extra braces around a multi-line expression to make it
more readable, even if they are not necessary.  Also, do not hesitate to
put extra braces anywhere if it improves clarity.

Declare variables just before they are needed.  Use local variables of
blocks---it helps optimization.  Don't write a multi-line variable
declaration with a single type specification and multiple variables.  If
the variables don't fit on single line, repeat the type specification.
Example:

@example
@group
octave_value retval;

octave_idx_type nr = b.rows ();
octave_idx_type nc = b.cols ();

double d1, d2;
@end group
@end example

Use lowercase names if possible.  Uppercase is acceptable for variable
names consisting of 1-2 letters.  Do not use mixed case names.

Use Octave's types and classes if possible.  Otherwise, use the C++
standard library.  Use of STL containers and algorithms is encouraged.
Use templates wisely to reduce code duplication.  Avoid comma
expressions, labels and gotos, and explicit typecasts.  If you need to
typecast, use the modern C++ casting operators.  In functions, minimize
the number of @code{return} statements---use nested @code{if} statements
if possible.

@node Other Sources
@section Other Sources
Apart from C++ and Octave language (m-files), Octave's sources include
files written in C, Fortran, M4, Perl, Unix shell, AWK, Texinfo, and
@TeX{}.  There are not many rules to follow when using these other
languages; some of them are summarized below.  In any case, the golden
rule is: if you modify a source file, try to follow any conventions you
can detect in the file or other similar files.

For C you should obviously follow all C++ rules that can apply.

If you modify a Fortran file, you should stay within Fortran 77 with
common extensions like @code{END DO}.  Currently, we want all sources to
be compilable with the f2c and g77 compilers, without special flags if
possible.  This usually means that non-legacy compilers also accept the
sources.

The M4 macro language is mainly used for Autoconf configuration files.
You should follow normal M4 rules when contributing to these files.
Some M4 files come from external source, namely the Autoconf archive
@url{http://autoconf-archive.cryp.to}.

If you give a code example in the documentation written in Texinfo with
the @code{@@example} environment, you should be aware that the text
within such an environment will not be wrapped.  It is recommended that
you keep the lines short enough to fit on pages in the generated pdf or
ps documents.  Here is a ruler (in an @code{@@example} environment) for
finding the appropriate line width:

@example
@group
         1         2         3         4         5         6
123456789012345678901234567890123456789012345678901234567890
@end group
@end example