refs: use skip_prefix() in ref_is_hidden()
[git/git.git] / Documentation / SubmittingPatches
1 Here are some guidelines for people who want to contribute their code
2 to this software.
4 (0) Decide what to base your work on.
6 In general, always base your work on the oldest branch that your
7 change is relevant to.
9 - A bugfix should be based on 'maint' in general. If the bug is not
10 present in 'maint', base it on 'master'. For a bug that's not yet
11 in 'master', find the topic that introduces the regression, and
12 base your work on the tip of the topic.
14 - A new feature should be based on 'master' in general. If the new
15 feature depends on a topic that is in 'pu', but not in 'master',
16 base your work on the tip of that topic.
18 - Corrections and enhancements to a topic not yet in 'master' should
19 be based on the tip of that topic. If the topic has not been merged
20 to 'next', it's alright to add a note to squash minor corrections
21 into the series.
23 - In the exceptional case that a new feature depends on several topics
24 not in 'master', start working on 'next' or 'pu' privately and send
25 out patches for discussion. Before the final merge, you may have to
26 wait until some of the dependent topics graduate to 'master', and
27 rebase your work.
29 - Some parts of the system have dedicated maintainers with their own
30 repositories (see the section "Subsystems" below). Changes to
31 these parts should be based on their trees.
33 To find the tip of a topic branch, run "git log --first-parent
34 master..pu" and look for the merge commit. The second parent of this
35 commit is the tip of the topic branch.
37 (1) Make separate commits for logically separate changes.
39 Unless your patch is really trivial, you should not be sending
40 out a patch that was generated between your working tree and
41 your commit head. Instead, always make a commit with complete
42 commit message and generate a series of patches from your
43 repository. It is a good discipline.
45 Give an explanation for the change(s) that is detailed enough so
46 that people can judge if it is good thing to do, without reading
47 the actual patch text to determine how well the code does what
48 the explanation promises to do.
50 If your description starts to get too long, that's a sign that you
51 probably need to split up your commit to finer grained pieces.
52 That being said, patches which plainly describe the things that
53 help reviewers check the patch, and future maintainers understand
54 the code, are the most beautiful patches. Descriptions that summarise
55 the point in the subject well, and describe the motivation for the
56 change, the approach taken by the change, and if relevant how this
57 differs substantially from the prior version, are all good things
58 to have.
60 Make sure that you have tests for the bug you are fixing. See
61 t/README for guidance.
63 When adding a new feature, make sure that you have new tests to show
64 the feature triggers the new behavior when it should, and to show the
65 feature does not trigger when it shouldn't. After any code change, make
66 sure that the entire test suite passes.
68 If you have an account at GitHub (and you can get one for free to work
69 on open source projects), you can use their Travis CI integration to
70 test your changes on Linux, Mac (and hopefully soon Windows). See
71 GitHub-Travis CI hints section for details.
73 Do not forget to update the documentation to describe the updated
74 behavior and make sure that the resulting documentation set formats
75 well. It is currently a liberal mixture of US and UK English norms for
76 spelling and grammar, which is somewhat unfortunate. A huge patch that
77 touches the files all over the place only to correct the inconsistency
78 is not welcome, though. Potential clashes with other changes that can
79 result from such a patch are not worth it. We prefer to gradually
80 reconcile the inconsistencies in favor of US English, with small and
81 easily digestible patches, as a side effect of doing some other real
82 work in the vicinity (e.g. rewriting a paragraph for clarity, while
83 turning en_UK spelling to en_US). Obvious typographical fixes are much
84 more welcomed ("teh -> "the"), preferably submitted as independent
85 patches separate from other documentation changes.
87 Oh, another thing. We are picky about whitespaces. Make sure your
88 changes do not trigger errors with the sample pre-commit hook shipped
89 in templates/hooks--pre-commit. To help ensure this does not happen,
90 run git diff --check on your changes before you commit.
93 (2) Describe your changes well.
95 The first line of the commit message should be a short description (50
96 characters is the soft limit, see DISCUSSION in git-commit(1)), and
97 should skip the full stop. It is also conventional in most cases to
98 prefix the first line with "area: " where the area is a filename or
99 identifier for the general area of the code being modified, e.g.
101 . archive: ustar header checksum is computed unsigned
102 . git-cherry-pick.txt: clarify the use of revision range notation
104 If in doubt which identifier to use, run "git log --no-merges" on the
105 files you are modifying to see the current conventions.
107 The body should provide a meaningful commit message, which:
109 . explains the problem the change tries to solve, iow, what is wrong
110 with the current code without the change.
112 . justifies the way the change solves the problem, iow, why the
113 result with the change is better.
115 . alternate solutions considered but discarded, if any.
117 Describe your changes in imperative mood, e.g. "make xyzzy do frotz"
118 instead of "[This patch] makes xyzzy do frotz" or "[I] changed xyzzy
119 to do frotz", as if you are giving orders to the codebase to change
120 its behaviour. Try to make sure your explanation can be understood
121 without external resources. Instead of giving a URL to a mailing list
122 archive, summarize the relevant points of the discussion.
124 If you want to reference a previous commit in the history of a stable
125 branch, use the format "abbreviated sha1 (subject, date)",
126 with the subject enclosed in a pair of double-quotes, like this:
128 Commit f86a374 ("pack-bitmap.c: fix a memleak", 2015-03-30)
129 noticed that ...
131 The "Copy commit summary" command of gitk can be used to obtain this
132 format.
135 (3) Generate your patch using Git tools out of your commits.
137 Git based diff tools generate unidiff which is the preferred format.
139 You do not have to be afraid to use -M option to "git diff" or
140 "git format-patch", if your patch involves file renames. The
141 receiving end can handle them just fine.
143 Please make sure your patch does not add commented out debugging code,
144 or include any extra files which do not relate to what your patch
145 is trying to achieve. Make sure to review
146 your patch after generating it, to ensure accuracy. Before
147 sending out, please make sure it cleanly applies to the "master"
148 branch head. If you are preparing a work based on "next" branch,
149 that is fine, but please mark it as such.
152 (4) Sending your patches.
154 Learn to use format-patch and send-email if possible. These commands
155 are optimized for the workflow of sending patches, avoiding many ways
156 your existing e-mail client that is optimized for "multipart/*" mime
157 type e-mails to corrupt and render your patches unusable.
159 People on the Git mailing list need to be able to read and
160 comment on the changes you are submitting. It is important for
161 a developer to be able to "quote" your changes, using standard
162 e-mail tools, so that they may comment on specific portions of
163 your code. For this reason, each patch should be submitted
164 "inline" in a separate message.
166 Multiple related patches should be grouped into their own e-mail
167 thread to help readers find all parts of the series. To that end,
168 send them as replies to either an additional "cover letter" message
169 (see below), the first patch, or the respective preceding patch.
171 If your log message (including your name on the
172 Signed-off-by line) is not writable in ASCII, make sure that
173 you send off a message in the correct encoding.
175 WARNING: Be wary of your MUAs word-wrap
176 corrupting your patch. Do not cut-n-paste your patch; you can
177 lose tabs that way if you are not careful.
179 It is a common convention to prefix your subject line with
180 [PATCH]. This lets people easily distinguish patches from other
181 e-mail discussions. Use of additional markers after PATCH and
182 the closing bracket to mark the nature of the patch is also
183 encouraged. E.g. [PATCH/RFC] is often used when the patch is
184 not ready to be applied but it is for discussion, [PATCH v2],
185 [PATCH v3] etc. are often seen when you are sending an update to
186 what you have previously sent.
188 "git format-patch" command follows the best current practice to
189 format the body of an e-mail message. At the beginning of the
190 patch should come your commit message, ending with the
191 Signed-off-by: lines, and a line that consists of three dashes,
192 followed by the diffstat information and the patch itself. If
193 you are forwarding a patch from somebody else, optionally, at
194 the beginning of the e-mail message just before the commit
195 message starts, you can put a "From: " line to name that person.
197 You often want to add additional explanation about the patch,
198 other than the commit message itself. Place such "cover letter"
199 material between the three-dash line and the diffstat. For
200 patches requiring multiple iterations of review and discussion,
201 an explanation of changes between each iteration can be kept in
202 Git-notes and inserted automatically following the three-dash
203 line via `git format-patch --notes`.
205 Do not attach the patch as a MIME attachment, compressed or not.
206 Do not let your e-mail client send quoted-printable. Do not let
207 your e-mail client send format=flowed which would destroy
208 whitespaces in your patches. Many
209 popular e-mail applications will not always transmit a MIME
210 attachment as plain text, making it impossible to comment on
211 your code. A MIME attachment also takes a bit more time to
212 process. This does not decrease the likelihood of your
213 MIME-attached change being accepted, but it makes it more likely
214 that it will be postponed.
216 Exception: If your mailer is mangling patches then someone may ask
217 you to re-send them using MIME, that is OK.
219 Do not PGP sign your patch. Most likely, your maintainer or other people on the
220 list would not have your PGP key and would not bother obtaining it anyway.
221 Your patch is not judged by who you are; a good patch from an unknown origin
222 has a far better chance of being accepted than a patch from a known, respected
223 origin that is done poorly or does incorrect things.
225 If you really really really really want to do a PGP signed
226 patch, format it as "multipart/signed", not a text/plain message
227 that starts with '-----BEGIN PGP SIGNED MESSAGE-----'. That is
228 not a text/plain, it's something else.
230 Send your patch with "To:" set to the mailing list, with "cc:" listing
231 people who are involved in the area you are touching (the output from
232 "git blame $path" and "git shortlog --no-merges $path" would help to
233 identify them), to solicit comments and reviews.
235 After the list reached a consensus that it is a good idea to apply the
236 patch, re-send it with "To:" set to the maintainer [*1*] and "cc:" the
237 list [*2*] for inclusion.
239 Do not forget to add trailers such as "Acked-by:", "Reviewed-by:" and
240 "Tested-by:" lines as necessary to credit people who helped your
241 patch.
243 [Addresses]
244 *1* The current maintainer:
245 *2* The mailing list:
248 (5) Certify your work by adding your "Signed-off-by: " line
250 To improve tracking of who did what, we've borrowed the
251 "sign-off" procedure from the Linux kernel project on patches
252 that are being emailed around. Although core Git is a lot
253 smaller project it is a good discipline to follow it.
255 The sign-off is a simple line at the end of the explanation for
256 the patch, which certifies that you wrote it or otherwise have
257 the right to pass it on as a open-source patch. The rules are
258 pretty simple: if you can certify the below:
260 Developer's Certificate of Origin 1.1
262 By making a contribution to this project, I certify that:
264 (a) The contribution was created in whole or in part by me and I
265 have the right to submit it under the open source license
266 indicated in the file; or
268 (b) The contribution is based upon previous work that, to the best
269 of my knowledge, is covered under an appropriate open source
270 license and I have the right under that license to submit that
271 work with modifications, whether created in whole or in part
272 by me, under the same open source license (unless I am
273 permitted to submit under a different license), as indicated
274 in the file; or
276 (c) The contribution was provided directly to me by some other
277 person who certified (a), (b) or (c) and I have not modified
278 it.
280 (d) I understand and agree that this project and the contribution
281 are public and that a record of the contribution (including all
282 personal information I submit with it, including my sign-off) is
283 maintained indefinitely and may be redistributed consistent with
284 this project or the open source license(s) involved.
286 then you just add a line saying
288 Signed-off-by: Random J Developer <>
290 This line can be automatically added by Git if you run the git-commit
291 command with the -s option.
293 Notice that you can place your own Signed-off-by: line when
294 forwarding somebody else's patch with the above rules for
295 D-C-O. Indeed you are encouraged to do so. Do not forget to
296 place an in-body "From: " line at the beginning to properly attribute
297 the change to its true author (see (2) above).
299 Also notice that a real name is used in the Signed-off-by: line. Please
300 don't hide your real name.
302 If you like, you can put extra tags at the end:
304 1. "Reported-by:" is used to credit someone who found the bug that
305 the patch attempts to fix.
306 2. "Acked-by:" says that the person who is more familiar with the area
307 the patch attempts to modify liked the patch.
308 3. "Reviewed-by:", unlike the other tags, can only be offered by the
309 reviewer and means that she is completely satisfied that the patch
310 is ready for application. It is usually offered only after a
311 detailed review.
312 4. "Tested-by:" is used to indicate that the person applied the patch
313 and found it to have the desired effect.
315 You can also create your own tag or use one that's in common usage
316 such as "Thanks-to:", "Based-on-patch-by:", or "Mentored-by:".
318 ------------------------------------------------
319 Subsystems with dedicated maintainers
321 Some parts of the system have dedicated maintainers with their own
322 repositories.
324 - git-gui/ comes from git-gui project, maintained by Pat Thoyts:
326 git://
328 - gitk-git/ comes from Paul Mackerras's gitk project:
330 git://
332 - po/ comes from the localization coordinator, Jiang Xin:
336 Patches to these parts should be based on their trees.
338 ------------------------------------------------
339 An ideal patch flow
341 Here is an ideal patch flow for this project the current maintainer
342 suggests to the contributors:
344 (0) You come up with an itch. You code it up.
346 (1) Send it to the list and cc people who may need to know about
347 the change.
349 The people who may need to know are the ones whose code you
350 are butchering. These people happen to be the ones who are
351 most likely to be knowledgeable enough to help you, but
352 they have no obligation to help you (i.e. you ask for help,
353 don't demand). "git log -p -- $area_you_are_modifying" would
354 help you find out who they are.
356 (2) You get comments and suggestions for improvements. You may
357 even get them in a "on top of your change" patch form.
359 (3) Polish, refine, and re-send to the list and the people who
360 spend their time to improve your patch. Go back to step (2).
362 (4) The list forms consensus that the last round of your patch is
363 good. Send it to the maintainer and cc the list.
365 (5) A topic branch is created with the patch and is merged to 'next',
366 and cooked further and eventually graduates to 'master'.
368 In any time between the (2)-(3) cycle, the maintainer may pick it up
369 from the list and queue it to 'pu', in order to make it easier for
370 people play with it without having to pick up and apply the patch to
371 their trees themselves.
373 ------------------------------------------------
374 Know the status of your patch after submission
376 * You can use Git itself to find out when your patch is merged in
377 master. 'git pull --rebase' will automatically skip already-applied
378 patches, and will let you know. This works only if you rebase on top
379 of the branch in which your patch has been merged (i.e. it will not
380 tell you if your patch is merged in pu if you rebase on top of
381 master).
383 * Read the Git mailing list, the maintainer regularly posts messages
384 entitled "What's cooking in git.git" and "What's in git.git" giving
385 the status of various proposed changes.
387 --------------------------------------------------
388 GitHub-Travis CI hints
390 With an account at GitHub (you can get one for free to work on open
391 source projects), you can use Travis CI to test your changes on Linux,
392 Mac (and hopefully soon Windows). You can find a successful example
393 test build here:
395 Follow these steps for the initial setup:
397 (1) Fork to your GitHub account.
398 You can find detailed instructions how to fork here:
401 (2) Open the Travis CI website:
403 (3) Press the "Sign in with GitHub" button.
405 (4) Grant Travis CI permissions to access your GitHub account.
406 You can find more information about the required permissions here:
409 (5) Open your Travis CI profile page:
411 (6) Enable Travis CI builds for your Git fork.
413 After the initial setup, Travis CI will run whenever you push new changes
414 to your fork of Git on GitHub. You can monitor the test state of all your
415 branches here:<Your GitHub handle>/git/branches
417 If a branch did not pass all test cases then it is marked with a red
418 cross. In that case you can click on the failing Travis CI job and
419 scroll all the way down in the log. Find the line "<-- Click here to see
420 detailed test output!" and click on the triangle next to the log line
421 number to expand the detailed test output. Here is such a failing
422 example:
424 Fix the problem and push your fix to your Git fork. This will trigger
425 a new Travis CI build to ensure all tests pass.
428 ------------------------------------------------
429 MUA specific hints
431 Some of patches I receive or pick up from the list share common
432 patterns of breakage. Please make sure your MUA is set up
433 properly not to corrupt whitespaces.
435 See the DISCUSSION section of git-format-patch(1) for hints on
436 checking your patch by mailing it to yourself and applying with
437 git-am(1).
439 While you are at it, check the resulting commit log message from
440 a trial run of applying the patch. If what is in the resulting
441 commit is not exactly what you would want to see, it is very
442 likely that your maintainer would end up hand editing the log
443 message when he applies your patch. Things like "Hi, this is my
444 first patch.\n", if you really want to put in the patch e-mail,
445 should come after the three-dash line that signals the end of the
446 commit message.
449 Pine
450 ----
452 (Johannes Schindelin)
454 I don't know how many people still use pine, but for those poor
455 souls it may be good to mention that the quell-flowed-text is
456 needed for recent versions.
458 ... the "no-strip-whitespace-before-send" option, too. AFAIK it
459 was introduced in 4.60.
461 (Linus Torvalds)
463 And 4.58 needs at least this.
465 ---
466 diff-tree 8326dd8350be64ac7fc805f6563a1d61ad10d32c (from e886a61f76edf5410573e92e38ce22974f9c40f1)
467 Author: Linus Torvalds <>
468 Date: Mon Aug 15 17:23:51 2005 -0700
470 Fix pine whitespace-corruption bug
472 There's no excuse for unconditionally removing whitespace from
473 the pico buffers on close.
475 diff --git a/pico/pico.c b/pico/pico.c
476 --- a/pico/pico.c
477 +++ b/pico/pico.c
478 @@ -219,7 +219,9 @@ PICO *pm;
479 switch(pico_all_done){ /* prepare for/handle final events */
480 case COMP_EXIT : /* already confirmed */
481 packheader();
482 +#if 0
483 stripwhitespace();
484 +#endif
485 c |= COMP_EXIT;
486 break;
489 (Daniel Barkalow)
491 > A patch to SubmittingPatches, MUA specific help section for
492 > users of Pine 4.63 would be very much appreciated.
494 Ah, it looks like a recent version changed the default behavior to do the
495 right thing, and inverted the sense of the configuration option. (Either
496 that or Gentoo did it.) So you need to set the
497 "no-strip-whitespace-before-send" option, unless the option you have is
498 "strip-whitespace-before-send", in which case you should avoid checking
499 it.
502 Thunderbird, KMail, GMail
503 -------------------------
505 See the MUA-SPECIFIC HINTS section of git-format-patch(1).
507 Gnus
508 ----
510 '|' in the *Summary* buffer can be used to pipe the current
511 message to an external program, and this is a handy way to drive
512 "git am". However, if the message is MIME encoded, what is
513 piped into the program is the representation you see in your
514 *Article* buffer after unwrapping MIME. This is often not what
515 you would want for two reasons. It tends to screw up non ASCII
516 characters (most notably in people's names), and also
517 whitespaces (fatal in patches). Running 'C-u g' to display the
518 message in raw form before using '|' to run the pipe can work
519 this problem around.