path: root/HACKING

================================================================
= This file

* This file attempts to describe the rules to use when hacking
  automake.

================================================================
= Administrivia

* The correct response to most actual bugs is to write a new test case
  which demonstrates the bug.  Then fix the bug, re-run the test suite,
  and check everything in.

* If you incorporate a change from somebody on the net:
  First, if it is a large change, you must make sure they have signed the
  appropriate paperwork.
  Second, be sure to add their name and email address to THANKS

* If a change fixes a test, mention the test in the commit message.
  If a change fixes a bug registered in the Automake debbugs tracker,
  mention the bug number in the commit message.

* If somebody reports a new bug, mention his name in the commit message
  and in the test case you write.  Put him into THANKS.

* When documenting a non-trivial idiom or example in the manual, be
  sure to add a test case for it, and to reference such test case from
  a proper Texinfo comment.

* Some files in the automake package are not owned by automake; these
  files are listed in the $(FETCHFILES) variable in Makefile.am.  They
  should never be edited here.  Almost all of them can be updated from
  respective upstreams with "make fetch" (this should be done especially
  before releases).  The only exception is the 'lib/COPYING' (from FSF),
  which should be updated by hand whenever the GPL gets updated (which
  shouldn't happen that often anyway :-)

* Changes other than bug fixes must be mentioned in NEWS.  Important
  bug fixes should be mentioned in NEWS, too.

================================================================
= Naming

* We've adopted the convention that internal AC_SUBSTs should be
  named with a leading 'am__', and internally generated targets
  should be named with a leading 'am--'.  This convention, although
  in place from at least February 2001, isn't yet universally used.
  But all new code should use it.

  We used to use '_am_' as the prefix for an internal AC_SUBST.
  However, it turns out that NEWS-OS 4.2R complains if a Makefile
  variable begins with the underscore character.  Yay for them.
  I changed the target naming convention just to be safe.

================================================================
= Editing '.am' files

* Always use $(...) and not ${...}

* Use ':', not 'true'.  Use 'exit 1', not 'false'.

* Use '##' comments liberally.  Comment anything even remotely
  unusual.

* Never use basename or dirname.  Instead use sed.

* Do not use 'cd' within back-quotes, use '$(am__cd)' instead.
  Otherwise the directory name may be printed, depending on CDPATH.
  More generally, do not ever use plain 'cd' together with a relative
  directory that does not start with a dot, or you might end up in one
  computed with CDPATH.

* For install and uninstall rules, if a loop is required, it should be
  silent.  Then the body of the loop itself should print each
  "important" command it runs.  The printed commands should be preceded
  by a single space.

* Ensure install rules do not create any installation directory where
  nothing is to be actually installed.  See automake bug#11030.

================================================================
= Editing automake.in and aclocal.in

* Indent using GNU style.  For historical reasons, the perl code
  contains portions indented using Larry Wall's style (perl-mode's
  default), and other portions using the GNU style (cperl-mode's
  default).  Write new code using GNU style.

* Don't use & for function calls, unless required.
  The use of & prevents prototypes from being checked.
  Just as above, don't change massively all the code to strip the
  &, just convert the old code as you work on it, and write new
  code without.

================================================================
= Working with git

* To regenerate dependent files created by aclocal and automake,
  use the 'bootstrap' script.  It uses the code from the source
  tree, so the resulting files (aclocal.m4 and Makefile.in) should
  be the same as you would get if you install this version of
  automake and use it to generate those files.  Be sure to have the
  latest stable version of Autoconf installed and available early
  in your PATH.

* The git tree currently carries a number of branches: master for the
  current development, and release branches named branch-X.Y.  The maint
  branch serves as common ground for both master and the active release
  branches.  Changes intended for both should be applied to maint, which
  should then be merged to release branches and master, of course after
  suitable testing.  It is advisable to merge only after a set of related
  commits have been applied.

* Example work flow for patches to maint:

  # 1. Checkout the "maint" branch:
  git checkout maint

  # 2. Apply the patch(es) with "git am" (or create them with $EDITOR):
  git am -3 0*.patch
  # 2a. Run required tests, if any ...

  # 3. Merge maint into branch-1.11:
  git checkout branch-1.11
  git merge maint
  # 3a. Run required tests, if any ...

  # 4. Redo steps 3 and 3a for master:
  git checkout master
  git merge maint
  # testing ...

  # 5. Push the maint and master branches:
  git push --dry-run origin maint branch-1.11 master
  # if all seems ok, then actually push:
  git push origin maint branch-1.11 master

* When fixing a bug (especially a long-standing one), it may be useful
  to commit the fix to a new temporary branch based off the commit that
  introduced the bug.  Then this "bugfix branch" can be merged into all
  the active branches descending from the buggy commit.  This offers a
  simple way to fix the bug consistently and effectively.

* There may be a number of longer-lived feature branches for new developments.
  They should be based off of a common ancestor of all active branches to
  which the feature should or might be merged later.  The next branch may
  serve as common ground for feature merging and testing, should they not
  be ready for master yet.

* For merges from branches other than maint, prefer 'git merge --log' over
  plain 'git merge', so that a later 'git log' gives an indication of which
  actual patches were merged even when they don't appear early in the list.

* master and release branches should not be rewound, i.e., should always
  fast-forward, except maybe for privacy issues.  The maint branch should not
  be rewound except maybe after retiring a release branch or a new stable
  release.  For next, and for feature branches, the announcement for the
  branch should document rewinding policy.

================================================================
= Writing a good commit message

* Here is the general format that Automake's commit messages are expected
  to follow.  See the further points below for clarifications and minor
  corrections.

      topic: brief description (this is the "summary line")

      <reference to relevant bugs, if any>

      Here goes a more detailed explanation of why the commit is needed,
      and a general overview of what it does, and how.  This section is
      optional, but you are expected to provide it more often than not.

      And if the detailed explanation is quite long or detailed, you can
      want to break it in more paragraphs.

      Then you can add references to relevant mailing list discussions
      (if any), with proper links.  But don't take this as an excuse for
      writing incomplete commit messages!  The "distilled" conclusions
      reached in such discussions should have been placed in the
      paragraphs above.

      Finally, here you can thank people that motivated or helped the
      change.  So, thanks to John Doe for bringing up the issue, and to
      J. Random Hacker for providing suggestions and testing the patch.

      <detailed list of touched files>

* The <detailed list of touched files> is mandatory but for the most
  trivial changes, and should follows the GNU guidelines for ChangeLog
  entries (described explicitly in the GNU Coding Standards); it might
  be something of this sort:

    * some/file (func1): Improved frobnication.
    (func2): Adjusted accordingly.
    * another/file (foo, bar): Likewise.
    * tests/foo.tap: New test.
    * tests/Makefile.am (TESTS): Add it.

* If your commit fixes an automake bug registered in the tracker (say
  numbered 1234), you should put the following line after the summary
  line:

      This change fixes automake bug#1234.

* If your commit is just related to the given bug report, but does not
  fix it, you might want to add a line like this instead:

      This change is related to automake bug#1234.

* When referring to older commits, use 'git describe' output as pointer.
  But also try to identify the given commit by date and/or summary line
  if possible.  Examples:

      Since yesterday's commit, v1.11-2019-g4d2bf42, ...

      ... removed in commit 'v1.11-1674-g02e9072' of 01-01-2012,
      "dist: ditch support for lzma"...

================================================================
= Test suite

* Use "make check" and "make maintainer-check" liberally.

* Make sure each test file is executable.

* Export the 'keep_testdirs' environment variable to "yes" to keep
  test directories for successful tests also.

* Use perl coverage information to ensure your new code is thoroughly
  tested by your new tests.

* See file 'tests/README' for more information.

================================================================
= Release procedure

* The steps outlined here are meant to be followed for alpha and stable
  releases as well.  Where differences are expected, they will be
  explicitly described.

* Fetch new versions of the files that are maintained by the FSF.
  Commit.  Unfortunately you need an FSF account to do this.
  (You can also use "make fetch", but that is still woefully incomplete.)

* Update NEWS.

* Update the version number in configure.ac.
  (The idea is that every other alpha number will be a net release.
  The repository will always have its own "odd" number so we can easily
  distinguish net and repo versions.)

* Run this:
  ./bootstrap && ./configure && make && make check && make distcheck

* Run "make git-tag-release".
  This will run the maintainer checks, check that the NEWS file is
  up-to-date, check that the local git repository and working tree
  are clean and up-to-date, and create a proper signed git tag for
  the release (based on the contents of $(VERSION)).

* Run "make git-upload-release".
  This will first verify that you are releasing from a tagged version
  and that the local git repository and working tree are clean and
  up-to-date, and will then run "make dist" to create the tarballs,
  and invoke the 'gnupload' script sign and upload them to the correct
  locations.  In case you need to sign with a non-default key, you can
  use "make GNUPLOADFLAGS='--user KEY' git-upload-release".

* Update version number in configure.ac to next alpha number.
  Re-run ./bootstrap and commit.

* Don't forget to "git push" your changes so they appear in the public
  git tree.

* For stable releases, update the manuals at www.gnu.org:
  - Generate manuals:
    cd doc
    wget "http://savannah.gnu.org/cgi-bin/viewcvs/~checkout~/texinfo/texinfo/util/gendocs.sh"
    wget "http://savannah.gnu.org/cgi-bin/viewcvs/~checkout~/texinfo/texinfo/util/gendocs_template"
    sh ./gendocs.sh --email bug-automake@gnu.org automake "GNU Automake"
  - copy manuals recursively to web cvs,
  - commit.
  - Check for link errors, fix them, recheck until convergence:
    <http://validator.w3.org/checklink>

* Send the announcement at least to <autotools-announce@gnu.org> and
  <automake@gnu.org>.  If the release is a stable one, the announcement
  must also go to <info-gnu@gnu.org>; if it is an alpha or beta release,
  announcement should be sent also to <platform-testers@gnu.org>, to
  maximize the possibility of early testing on exotic or proprietary
  systems.  Finally, copy the announcement into the NEWS feed at
  <https://savannah.gnu.org/projects/automake>.

-----

Copyright (C) 2003-2012 Free Software Foundation, Inc.

This program 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 2, or (at your option)
any later version.

This program 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 this program.  If not, see <http://www.gnu.org/licenses/>.

Local Variables:
mode: text
End: