3 perlrepository - Using the Perl source repository
7 All of Perl's source code is kept centrally in a Git repository. The
8 repository contains many Perl revisions from Perl 1 onwards and all the
9 revisions from Perforce, the version control system we were using
10 previously. This repository is accessible in different ways.
12 The full repository takes up about 80MB of disk space. A check out of
13 the blead branch (that is, the master branch, which contains bleadperl,
14 the development version of perl 5) takes up about 160MB of disk space
15 (including the repository). A build of bleadperl takes up about 200MB
16 (including the repository and the check out).
18 =head1 GETTING ACCESS TO THE REPOSITORY
20 =head2 READ ACCESS VIA THE WEB
22 You may access this over the web. This allows you to browse the tree,
23 see recent commits, search for particular commits and more. You may
26 http://perl5.git.perl.org/perl.git
28 =head2 READ ACCESS VIA GIT
30 You will need a copy of Git for your computer. You can fetch a copy of
31 the repository using the Git protocol (which uses port 9418):
33 git clone git://perl5.git.perl.org/perl.git perl-git
35 This clones the repository and makes a local copy in the F<perl-git>
38 If your local network does not allow you to use port 9418, then you can
39 fetch a copy of the repository over HTTP (this is slower):
41 git clone http://perl5.git.perl.org/perl.git perl-http
43 This clones the repository and makes a local copy in the F<perl-http>
46 =head2 WRITE ACCESS TO THE REPOSITORY
48 If you are a committer, then you can fetch a copy of the repository
49 that you can push back on with:
51 git clone ssh://perl5.git.perl.org/gitroot/perl.git perl-ssh
53 This clones the repository and makes a local copy in the 'perl-ssh'
56 If you clone using git, which is faster than ssh, then you will need to
57 modify your config in order to enable pushing. Edit F<.git/config>
58 where you will see something like:
61 url = git://perl5.git.perl.org/perl.git
63 change that to something like this:
66 url = ssh://perl5.git.perl.org/gitroot/perl.git
68 NOTE: there are symlinks set up so that the /gitroot is actually
71 You can also set up your user name and e-mail address. For example
73 % git config user.name "Leon Brocard"
74 % git config user.email acme@astray.com
76 It is also possible to keep C<origin> as a git remote, and add a new
77 remote for ssh access:
79 % git remote add camel user@camel:/gitroot/perl.git
81 This allows you to update your local repository by pulling from
82 C<origin>, which is faster and doesn't require you to authenticate, and
83 to push your changes back with the C<camel> remote:
88 The C<fetch> command just updates the C<camel> refs, as the objects
89 themselves should have been fetched when pulling from C<origin>.
91 =head1 OVERVIEW OF THE REPOSITORY
93 Once you have changed into the repository directory, you can inspect
96 After a clone the repository will contain a single local branch, which
97 will be the current branch as well, as indicated by the asterisk.
102 Using the -a switch to C<branch> will also show the remote tracking
103 branches in the repository:
111 The branches that begin with "origin" correspond to the "git remote"
112 that you cloned from (which is named "origin"). Each branch on the
113 remote will be exactly tracked by theses branches. You should NEVER do
114 work on these remote tracking branches. You only ever do work in a
115 local branch. Local branches can be configured to automerge (on pull)
116 from a designated remote tracking branch. This is the case with the
117 default branch C<blead> which will be configured to merge from the
118 remote tracking branch C<origin/blead>.
120 You can see recent commits:
124 And pull new changes from the repository, and update your local
125 repository (must be clean first)
129 Assuming we are on the branch C<blead> immediately after a pull, this
130 command would be more or less equivalent to:
133 % git merge origin/blead
135 In fact if you want to update your local repository without touching
136 your working directory you do:
140 And if you want to update your remote-tracking branches for all defined
141 remotes simultaneously you can do
145 Neither of these last two commands will update your working directory,
146 however both will update the remote-tracking branches in your
149 To switch to another branch:
151 % git checkout origin/maint-5.8-dor
153 To make a local branch of a remote branch:
155 % git checkout -b maint-5.10 origin/maint-5.10
157 To switch back to blead:
161 =head2 FINDING OUT YOUR STATUS
163 The most common git command you will use will probably be
167 This command will produce as output a description of the current state
168 of the repository, including modified files and unignored untracked
169 files, and in addition it will show things like what files have been
170 staged for the next commit, and usually some useful information about
171 how to change things. For instance the following:
175 # Your branch is ahead of 'origin/blead' by 1 commit.
177 # Changes to be committed:
178 # (use "git reset HEAD <file>..." to unstage)
180 # modified: pod/perlrepository.pod
182 # Changed but not updated:
183 # (use "git add <file>..." to update what will be committed)
185 # modified: pod/perlrepository.pod
188 # (use "git add <file>..." to include in what will be committed)
190 # deliberate.untracked
192 This shows that there were changes to this document staged for commit,
193 and that there were further changes in the working directory not yet
194 staged. It also shows that there was an untracked file in the working
195 directory, and as you can see shows how to change all of this. It also
196 shows that there is one commit on the working branch C<blead> which
197 has not been pushed to the C<origin> remote yet. B<NOTE>: that this
198 output is also what you see as a template if you do not provide a
199 message to C<git commit>.
201 Assuming we commit all the mentioned changes above:
203 % git commit -a -m'explain git status and stuff about remotes'
204 Created commit daf8e63: explain git status and stuff about remotes
205 1 files changed, 83 insertions(+), 3 deletions(-)
207 We can re-run git status and see something like this:
211 # Your branch is ahead of 'origin/blead' by 2 commits.
214 # (use "git add <file>..." to include in what will be committed)
216 # deliberate.untracked
217 nothing added to commit but untracked files present (use "git add" to track)
220 When in doubt, before you do anything else, check your status and read
221 it carefully, many questions are answered directly by the git status
224 =head1 SUBMITTING A PATCH
226 If you have a patch in mind for Perl, you should first get a copy of
229 % git clone git://perl5.git.perl.org/perl.git perl-git
231 Then change into the directory:
235 Alternatively, if you already have a Perl repository, you should ensure
236 that you're on the I<blead> branch, and your repository is up to date:
241 (It's preferable to patch against the latest blead version, since patches
242 are usually integrated from blead to the maintenance branches. This
243 does not apply, obviously, in the rare case where your patch is specific
244 to a maintaince release.)
246 Now that we have everything up to date, we need to create a temporary
247 new branch for these changes and switch into it:
249 % git checkout -b orange
251 which is the short form of
254 % git checkout orange
256 Then make your changes. For example, if Leon Brocard changes his name
257 to Orange Brocard, we should change his name in the AUTHORS file:
259 % perl -pi -e 's{Leon Brocard}{Orange Brocard}' AUTHORS
261 You can see what files are changed:
265 # Changes to be committed:
266 # (use "git reset HEAD <file>..." to unstage)
271 And you can see the changes:
274 diff --git a/AUTHORS b/AUTHORS
275 index 293dd70..722c93e 100644
278 @@ -541,7 +541,7 @@ Lars Hecking <lhecking@nmrc.ucc.ie>
279 Laszlo Molnar <laszlo.molnar@eth.ericsson.se>
280 Leif Huhn <leif@hale.dkstat.com>
281 Len Johnson <lenjay@ibm.net>
282 -Leon Brocard <acme@astray.com>
283 +Orange Brocard <acme@astray.com>
284 Les Peters <lpeters@aol.net>
285 Lesley Binks <lesley.binks@gmail.com>
286 Lincoln D. Stein <lstein@cshl.org>
288 Now commit your change locally:
291 % git commit -m 'Rename Leon Brocard to Orange Brocard'
292 Created commit 6196c1d: Rename Leon Brocard to Orange Brocard
293 1 files changed, 1 insertions(+), 1 deletions(-)
295 Now you should create a patch file for all your local changes:
297 % git format-patch origin
298 0001-Rename-Leon-Brocard-to-Orange-Brocard.patch
300 You should now send an email to perl5-porters@perl.org with a
301 description of your changes, and attach this patch file as an
304 If you want to delete your temporary branch, you may do so with:
307 % git branch -d orange
308 error: The branch 'orange' is not an ancestor of your current HEAD.
309 If you are sure you want to delete it, run 'git branch -D orange'.
310 % git branch -D orange
311 Deleted branch orange.
313 =head2 A note on derived files
315 Be aware that many files in the distribution are derivative--avoid
316 patching them, because git won't see the changes to them, and the
317 build process will overwrite them.
318 Patch the originals instead. Most utilities (like perldoc) are in
319 this category, i.e. patch utils/perldoc.PL rather than utils/perldoc.
320 Similarly, don't create patches for files under $src_root/ext from
321 their copies found in $install_root/lib. If you are unsure about the
322 proper location of a file that may have gotten copied while building
323 the source distribution, consult the C<MANIFEST>.
325 =head2 A note on binary files
327 Since the patch(1) utility cannot deal with binary files, it's important
328 that you either avoid the use of binary files in your patch, generate the
329 files dynamically, or that you encode any binary files using the
330 F<uupacktool.pl> utility.
332 Assuming you needed to include a gzip-encoded file for a module's test
333 suite, you might do this as follows using the F<uupacktool.pl> utility:
335 $ perl uupacktool.pl -v -p -D lib/Some/Module/t/src/t.gz
336 Writing lib/Some/Module/t/src/t.gz into lib/Some/Module/t/src/t.gz.packed
338 This will replace the C<t.gz> file with an encoded counterpart. During
339 C<make test>, before any tests are run, perl's Makefile will restore all
340 the C<.packed> files mentioned in the MANIFEST to their original name.
341 This means that the test suite does not need to be aware of this packing
342 scheme and will not need to be altered.
344 =head2 Getting your patch accepted
346 The first thing you should include with your patch is a description of the
347 problem that the patch corrects. If it is a code patch (rather than a
348 documentation patch) you should also include a small test case that
349 illustrates the bug (a patch to an existing test file is preferred).
351 If you are submitting a code patch there are several other things that
356 =item Comments, Comments, Comments
358 Be sure to adequately comment your code. While commenting every
359 line is unnecessary, anything that takes advantage of side effects of
360 operators, that creates changes that will be felt outside of the
361 function being patched, or that others may find confusing should
362 be documented. If you are going to err, it is better to err on the
363 side of adding too many comments than too few.
367 In general, please follow the particular style of the code you are patching.
369 In particular, follow these general guidelines for patching Perl sources:
371 8-wide tabs (no exceptions!)
372 4-wide indents for code, 2-wide indents for nested CPP #defines
373 try hard not to exceed 79-columns
375 uncuddled elses and "K&R" style for indenting control constructs
376 no C++ style (//) comments
377 mark places that need to be revisited with XXX (and revisit often!)
378 opening brace lines up with "if" when conditional spans multiple
379 lines; should be at end-of-line otherwise
380 in function definitions, name starts in column 0 (return value is on
382 single space after keywords that are followed by parens, no space
383 between function name and following paren
384 avoid assignments in conditionals, but if they're unavoidable, use
385 extra paren, e.g. "if (a && (b = c)) ..."
386 "return foo;" rather than "return(foo);"
387 "if (!foo) ..." rather than "if (foo == FALSE) ..." etc.
391 When submitting a patch you should make every effort to also include
392 an addition to perl's regression tests to properly exercise your
393 patch. Your testsuite additions should generally follow these
394 guidelines (courtesy of Gurusamy Sarathy <gsar@activestate.com>):
396 Know what you're testing. Read the docs, and the source.
397 Tend to fail, not succeed.
398 Interpret results strictly.
399 Use unrelated features (this will flush out bizarre interactions).
400 Use non-standard idioms (otherwise you are not testing TIMTOWTDI).
401 Avoid using hardcoded test numbers whenever possible (the
402 EXPECTED/GOT found in t/op/tie.t is much more maintainable,
403 and gives better failure reports).
404 Give meaningful error messages when a test fails.
405 Avoid using qx// and system() unless you are testing for them. If you
406 do use them, make sure that you cover _all_ perl platforms.
407 Unlink any temporary files you create.
408 Promote unforeseen warnings to errors with $SIG{__WARN__}.
409 Be sure to use the libraries and modules shipped with the version
410 being tested, not those that were already installed.
411 Add comments to the code explaining what you are testing for.
412 Make updating the '1..42' string unnecessary. Or make sure that
414 Test _all_ behaviors of a given operator, library, or function:
415 - All optional arguments
416 - Return values in various contexts (boolean, scalar, list, lvalue)
417 - Use both global and lexical variables
418 - Don't forget the exceptional, pathological cases.
422 =head1 ACCEPTING A PATCH
424 If you have received a patch file generated using the above section,
425 you should try out the patch.
427 First we need to create a temporary new branch for these changes and
430 % git checkout -b experimental
432 Patches that were formatted by C<git format-patch> are applied with
435 % git am 0001-Rename-Leon-Brocard-to-Orange-Brocard.patch
436 Applying Rename Leon Brocard to Orange Brocard
438 If just a raw diff is provided, it is also possible use this two-step
441 % git apply bugfix.diff
442 % git commit -am "Some fixing" --author="That Guy <that.guy@internets.com>"
444 Now we can inspect the change:
447 commit b1b3dab48344cff6de4087efca3dbd63548ab5e2
448 Author: Leon Brocard <acme@astray.com>
449 Date: Fri Dec 19 17:02:59 2008 +0000
451 Rename Leon Brocard to Orange Brocard
455 diff --git a/AUTHORS b/AUTHORS
456 index 293dd70..722c93e 100644
459 @@ -541,7 +541,7 @@ Lars Hecking <lhecking@nmrc.ucc.ie>
460 Laszlo Molnar <laszlo.molnar@eth.ericsson.se>
461 Leif Huhn <leif@hale.dkstat.com>
462 Len Johnson <lenjay@ibm.net>
463 -Leon Brocard <acme@astray.com>
464 +Orange Brocard <acme@astray.com>
465 Les Peters <lpeters@aol.net>
466 Lesley Binks <lesley.binks@gmail.com>
467 Lincoln D. Stein <lstein@cshl.org>
469 If you are a committer to Perl and you think the patch is good, you can
470 then merge it into blead then push it out to the main repository:
473 % git merge experimental
476 If you want to delete your temporary branch, you may do so with:
479 % git branch -d experimental
480 error: The branch 'experimental' is not an ancestor of your current HEAD.
481 If you are sure you want to delete it, run 'git branch -D experimental'.
482 % git branch -D experimental
483 Deleted branch experimental.
485 =head1 CLEANING A WORKING DIRECTORY
487 The command C<git clean> can with varying arguments be used as a
488 replacement for make-clean.
490 To reset your working directory to a pristine condition you can do:
494 However, be aware this will delete ALL untracked content. You can use
498 to remove all ignored untracked files, such as build and test
499 byproduct, but leave any manually created files alone.
501 If you only want to cancel some uncommitted edits, you can use
502 C<git checkout> and give it a list of files to be reverted.
504 If you want to cancel one or several commits, you can use C<git reset>.
508 C<git> provides a built-in way to determine, with a binary search in
509 the history, which commit should be blamed for introducing a given bug.
511 Suppose that we have a script F<~/testcase.pl> that exits with C<0>
512 when some behaviour is correct, and with C<1> when it's faulty. We need
513 an helper script that automates building C<perl> and running the
519 # If you can use ccache, add -Dcc=ccache\ gcc -Dld=gcc to the Configure line
520 sh Configure -des -Dusedevel -Doptimize="-g" || exit 125
522 ./perl -Ilib ~/testcase.pl
524 This script may return C<125> to indicate that the corresponding commit
525 should be skipped. Otherwise, it returns the status of
528 We first enter in bisect mode with:
532 For example, if the bug is present on C<HEAD> but wasn't in 5.10.0,
533 C<git> will learn about this when you enter:
536 % git bisect good perl-5.10.0
537 Bisecting: 853 revisions left to test after this
539 This results in checking out the median commit between C<HEAD> and
540 C<perl-5.10.0>. We can then run the bisecting process with:
542 % git bisect run ~/run
544 When the first bad commit is isolated, C<git bisect> will tell you so:
546 ca4cfd28534303b82a216cfe83a1c80cbc3b9dc5 is first bad commit
547 commit ca4cfd28534303b82a216cfe83a1c80cbc3b9dc5
548 Author: Dave Mitchell <davem@fdisolutions.com>
549 Date: Sat Feb 9 14:56:23 2008 +0000
551 [perl #49472] Attributes + Unknown Error
556 You can peek into the bisecting process with C<git bisect log> and
557 C<git bisect visualize>. C<git bisect reset> will get you out of bisect
560 Please note that the first C<good> state must be an ancestor of the
561 first C<bad> state. If you want to search for the commit that I<solved>
562 some bug, you have to negate your test case (i.e. exit with C<1> if OK
563 and C<0> if not) and still mark the lower bound as C<good> and the
564 upper as C<bad>. The "first bad commit" has then to be understood as
565 the "first commit where the bug is solved".
567 C<git help bisect> has much more information on how you can tweak your
570 =head1 COMMITTING TO MAINTENANCE VERSIONS
572 To commit to a maintenance version of perl, you need to create a local
575 % git checkout --track -b maint-5.005 origin/maint-5.005
577 This creates a local branch named C<maint-5.005>, which tracks the remote
578 branch C<origin/maint-5.005>. Then you can pull, commit, merge and push as
581 You can also cherry-pick commits from blead and another branch, by
582 using the C<git cherry-pick> command. It is recommended to use the B<-x>
583 option to C<git cherry-pick> in order to record the SHA1 of the original
584 commit in the new commit message.
588 The git documentation, accessible via C<git help command>.