doc/SubmittingPatches: improve text formatting
[git/git.git] / Documentation / SubmittingPatches
1 Submitting Patches
2 ==================
3
4 == Guidelines
5
6 Here are some guidelines for people who want to contribute their code
7 to this software.
8
9 [[base-branch]]
10 === Decide what to base your work on.
11
12 In general, always base your work on the oldest branch that your
13 change is relevant to.
14
15 * A bugfix should be based on `maint` in general. If the bug is not
16 present in `maint`, base it on `master`. For a bug that's not yet
17 in `master`, find the topic that introduces the regression, and
18 base your work on the tip of the topic.
19
20 * A new feature should be based on `master` in general. If the new
21 feature depends on a topic that is in `pu`, but not in `master`,
22 base your work on the tip of that topic.
23
24 * Corrections and enhancements to a topic not yet in `master` should
25 be based on the tip of that topic. If the topic has not been merged
26 to `next`, it's alright to add a note to squash minor corrections
27 into the series.
28
29 * In the exceptional case that a new feature depends on several topics
30 not in `master`, start working on `next` or `pu` privately and send
31 out patches for discussion. Before the final merge, you may have to
32 wait until some of the dependent topics graduate to `master`, and
33 rebase your work.
34
35 * Some parts of the system have dedicated maintainers with their own
36 repositories (see the section "Subsystems" below). Changes to
37 these parts should be based on their trees.
38
39 To find the tip of a topic branch, run `git log --first-parent
40 master..pu` and look for the merge commit. The second parent of this
41 commit is the tip of the topic branch.
42
43 [[separate-commits]]
44 === Make separate commits for logically separate changes.
45
46 Unless your patch is really trivial, you should not be sending
47 out a patch that was generated between your working tree and
48 your commit head. Instead, always make a commit with complete
49 commit message and generate a series of patches from your
50 repository. It is a good discipline.
51
52 Give an explanation for the change(s) that is detailed enough so
53 that people can judge if it is good thing to do, without reading
54 the actual patch text to determine how well the code does what
55 the explanation promises to do.
56
57 If your description starts to get too long, that's a sign that you
58 probably need to split up your commit to finer grained pieces.
59 That being said, patches which plainly describe the things that
60 help reviewers check the patch, and future maintainers understand
61 the code, are the most beautiful patches. Descriptions that summarize
62 the point in the subject well, and describe the motivation for the
63 change, the approach taken by the change, and if relevant how this
64 differs substantially from the prior version, are all good things
65 to have.
66
67 Make sure that you have tests for the bug you are fixing. See
68 `t/README` for guidance.
69
70 [[tests]]
71 When adding a new feature, make sure that you have new tests to show
72 the feature triggers the new behavior when it should, and to show the
73 feature does not trigger when it shouldn't. After any code change, make
74 sure that the entire test suite passes.
75
76 If you have an account at GitHub (and you can get one for free to work
77 on open source projects), you can use their Travis CI integration to
78 test your changes on Linux, Mac (and hopefully soon Windows). See
79 GitHub-Travis CI hints section for details.
80
81 Do not forget to update the documentation to describe the updated
82 behavior and make sure that the resulting documentation set formats
83 well. It is currently a liberal mixture of US and UK English norms for
84 spelling and grammar, which is somewhat unfortunate. A huge patch that
85 touches the files all over the place only to correct the inconsistency
86 is not welcome, though. Potential clashes with other changes that can
87 result from such a patch are not worth it. We prefer to gradually
88 reconcile the inconsistencies in favor of US English, with small and
89 easily digestible patches, as a side effect of doing some other real
90 work in the vicinity (e.g. rewriting a paragraph for clarity, while
91 turning en_UK spelling to en_US). Obvious typographical fixes are much
92 more welcomed ("teh -> "the"), preferably submitted as independent
93 patches separate from other documentation changes.
94
95 [[whitespace-check]]
96 Oh, another thing. We are picky about whitespaces. Make sure your
97 changes do not trigger errors with the sample pre-commit hook shipped
98 in `templates/hooks--pre-commit`. To help ensure this does not happen,
99 run `git diff --check` on your changes before you commit.
100
101 [[describe-changes]]
102 === Describe your changes well.
103
104 The first line of the commit message should be a short description (50
105 characters is the soft limit, see DISCUSSION in linkgit:git-commit[1]),
106 and should skip the full stop. It is also conventional in most cases to
107 prefix the first line with "area: " where the area is a filename or
108 identifier for the general area of the code being modified, e.g.
109
110 * doc: clarify distinction between sign-off and pgp-signing
111 * githooks.txt: improve the intro section
112
113 If in doubt which identifier to use, run `git log --no-merges` on the
114 files you are modifying to see the current conventions.
115
116 [[summary-section]]
117 It's customary to start the remainder of the first line after "area: "
118 with a lower-case letter. E.g. "doc: clarify...", not "doc:
119 Clarify...", or "githooks.txt: improve...", not "githooks.txt:
120 Improve...".
121
122 [[meaningful-message]]
123 The body should provide a meaningful commit message, which:
124
125 . explains the problem the change tries to solve, i.e. what is wrong
126 with the current code without the change.
127
128 . justifies the way the change solves the problem, i.e. why the
129 result with the change is better.
130
131 . alternate solutions considered but discarded, if any.
132
133 [[imperative-mood]]
134 Describe your changes in imperative mood, e.g. "make xyzzy do frotz"
135 instead of "[This patch] makes xyzzy do frotz" or "[I] changed xyzzy
136 to do frotz", as if you are giving orders to the codebase to change
137 its behavior. Try to make sure your explanation can be understood
138 without external resources. Instead of giving a URL to a mailing list
139 archive, summarize the relevant points of the discussion.
140
141 [[commit-reference]]
142 If you want to reference a previous commit in the history of a stable
143 branch, use the format "abbreviated sha1 (subject, date)",
144 with the subject enclosed in a pair of double-quotes, like this:
145
146 ....
147 Commit f86a374 ("pack-bitmap.c: fix a memleak", 2015-03-30)
148 noticed that ...
149 ....
150
151 The "Copy commit summary" command of gitk can be used to obtain this
152 format, or this invocation of `git show`:
153
154 ....
155 git show -s --date=short --pretty='format:%h ("%s", %ad)' <commit>
156 ....
157
158 [[git-tools]]
159 === Generate your patch using Git tools out of your commits.
160
161 Git based diff tools generate unidiff which is the preferred format.
162
163 You do not have to be afraid to use `-M` option to `git diff` or
164 `git format-patch`, if your patch involves file renames. The
165 receiving end can handle them just fine.
166
167 [[review-patch]]
168 Please make sure your patch does not add commented out debugging code,
169 or include any extra files which do not relate to what your patch
170 is trying to achieve. Make sure to review
171 your patch after generating it, to ensure accuracy. Before
172 sending out, please make sure it cleanly applies to the `master`
173 branch head. If you are preparing a work based on "next" branch,
174 that is fine, but please mark it as such.
175
176 [[send-patches]]
177 === Sending your patches.
178
179 Learn to use format-patch and send-email if possible. These commands
180 are optimized for the workflow of sending patches, avoiding many ways
181 your existing e-mail client that is optimized for "multipart/*" mime
182 type e-mails to corrupt and render your patches unusable.
183
184 People on the Git mailing list need to be able to read and
185 comment on the changes you are submitting. It is important for
186 a developer to be able to "quote" your changes, using standard
187 e-mail tools, so that they may comment on specific portions of
188 your code. For this reason, each patch should be submitted
189 "inline" in a separate message.
190
191 Multiple related patches should be grouped into their own e-mail
192 thread to help readers find all parts of the series. To that end,
193 send them as replies to either an additional "cover letter" message
194 (see below), the first patch, or the respective preceding patch.
195
196 If your log message (including your name on the
197 Signed-off-by line) is not writable in ASCII, make sure that
198 you send off a message in the correct encoding.
199
200 WARNING: Be wary of your MUAs word-wrap
201 corrupting your patch. Do not cut-n-paste your patch; you can
202 lose tabs that way if you are not careful.
203
204 It is a common convention to prefix your subject line with
205 [PATCH]. This lets people easily distinguish patches from other
206 e-mail discussions. Use of additional markers after PATCH and
207 the closing bracket to mark the nature of the patch is also
208 encouraged. E.g. [PATCH/RFC] is often used when the patch is
209 not ready to be applied but it is for discussion, [PATCH v2],
210 [PATCH v3] etc. are often seen when you are sending an update to
211 what you have previously sent.
212
213 `git format-patch` command follows the best current practice to
214 format the body of an e-mail message. At the beginning of the
215 patch should come your commit message, ending with the
216 Signed-off-by: lines, and a line that consists of three dashes,
217 followed by the diffstat information and the patch itself. If
218 you are forwarding a patch from somebody else, optionally, at
219 the beginning of the e-mail message just before the commit
220 message starts, you can put a "From: " line to name that person.
221
222 You often want to add additional explanation about the patch,
223 other than the commit message itself. Place such "cover letter"
224 material between the three-dash line and the diffstat. For
225 patches requiring multiple iterations of review and discussion,
226 an explanation of changes between each iteration can be kept in
227 Git-notes and inserted automatically following the three-dash
228 line via `git format-patch --notes`.
229
230 [[attachment]]
231 Do not attach the patch as a MIME attachment, compressed or not.
232 Do not let your e-mail client send quoted-printable. Do not let
233 your e-mail client send format=flowed which would destroy
234 whitespaces in your patches. Many
235 popular e-mail applications will not always transmit a MIME
236 attachment as plain text, making it impossible to comment on
237 your code. A MIME attachment also takes a bit more time to
238 process. This does not decrease the likelihood of your
239 MIME-attached change being accepted, but it makes it more likely
240 that it will be postponed.
241
242 Exception: If your mailer is mangling patches then someone may ask
243 you to re-send them using MIME, that is OK.
244
245 [[pgp-signature]]
246 Do not PGP sign your patch. Most likely, your maintainer or other people on the
247 list would not have your PGP key and would not bother obtaining it anyway.
248 Your patch is not judged by who you are; a good patch from an unknown origin
249 has a far better chance of being accepted than a patch from a known, respected
250 origin that is done poorly or does incorrect things.
251
252 If you really really really really want to do a PGP signed
253 patch, format it as "multipart/signed", not a text/plain message
254 that starts with `-----BEGIN PGP SIGNED MESSAGE-----`. That is
255 not a text/plain, it's something else.
256
257 Send your patch with "To:" set to the mailing list, with "cc:" listing
258 people who are involved in the area you are touching (the output from
259 `git blame $path` and `git shortlog --no-merges $path` would help to
260 identify them), to solicit comments and reviews.
261
262 :1: footnote:[The current maintainer: gitster@pobox.com]
263 :2: footnote:[The mailing list: git@vger.kernel.org]
264
265 After the list reached a consensus that it is a good idea to apply the
266 patch, re-send it with "To:" set to the maintainer{1} and "cc:" the
267 list{2} for inclusion.
268
269 Do not forget to add trailers such as `Acked-by:`, `Reviewed-by:` and
270 `Tested-by:` lines as necessary to credit people who helped your
271 patch.
272
273 [[sign-off]]
274 === Certify your work by adding your "Signed-off-by: " line
275
276 To improve tracking of who did what, we've borrowed the
277 "sign-off" procedure from the Linux kernel project on patches
278 that are being emailed around. Although core Git is a lot
279 smaller project it is a good discipline to follow it.
280
281 The sign-off is a simple line at the end of the explanation for
282 the patch, which certifies that you wrote it or otherwise have
283 the right to pass it on as a open-source patch. The rules are
284 pretty simple: if you can certify the below D-C-O:
285
286 [[dco]]
287 .Developer's Certificate of Origin 1.1
288 ____
289 By making a contribution to this project, I certify that:
290
291 a. The contribution was created in whole or in part by me and I
292 have the right to submit it under the open source license
293 indicated in the file; or
294
295 b. The contribution is based upon previous work that, to the best
296 of my knowledge, is covered under an appropriate open source
297 license and I have the right under that license to submit that
298 work with modifications, whether created in whole or in part
299 by me, under the same open source license (unless I am
300 permitted to submit under a different license), as indicated
301 in the file; or
302
303 c. The contribution was provided directly to me by some other
304 person who certified (a), (b) or (c) and I have not modified
305 it.
306
307 d. I understand and agree that this project and the contribution
308 are public and that a record of the contribution (including all
309 personal information I submit with it, including my sign-off) is
310 maintained indefinitely and may be redistributed consistent with
311 this project or the open source license(s) involved.
312 ____
313
314 then you just add a line saying
315
316 ....
317 Signed-off-by: Random J Developer <random@developer.example.org>
318 ....
319
320 This line can be automatically added by Git if you run the git-commit
321 command with the -s option.
322
323 Notice that you can place your own Signed-off-by: line when
324 forwarding somebody else's patch with the above rules for
325 D-C-O. Indeed you are encouraged to do so. Do not forget to
326 place an in-body "From: " line at the beginning to properly attribute
327 the change to its true author (see (2) above).
328
329 [[real-name]]
330 Also notice that a real name is used in the Signed-off-by: line. Please
331 don't hide your real name.
332
333 [[commit-trailers]]
334 If you like, you can put extra tags at the end:
335
336 . `Reported-by:` is used to credit someone who found the bug that
337 the patch attempts to fix.
338 . `Acked-by:` says that the person who is more familiar with the area
339 the patch attempts to modify liked the patch.
340 . `Reviewed-by:`, unlike the other tags, can only be offered by the
341 reviewer and means that she is completely satisfied that the patch
342 is ready for application. It is usually offered only after a
343 detailed review.
344 . `Tested-by:` is used to indicate that the person applied the patch
345 and found it to have the desired effect.
346
347 You can also create your own tag or use one that's in common usage
348 such as "Thanks-to:", "Based-on-patch-by:", or "Mentored-by:".
349
350 == Subsystems with dedicated maintainers
351
352 Some parts of the system have dedicated maintainers with their own
353 repositories.
354
355 - 'git-gui/' comes from git-gui project, maintained by Pat Thoyts:
356
357 git://repo.or.cz/git-gui.git
358
359 - 'gitk-git/' comes from Paul Mackerras's gitk project:
360
361 git://ozlabs.org/~paulus/gitk
362
363 - 'po/' comes from the localization coordinator, Jiang Xin:
364
365 https://github.com/git-l10n/git-po/
366
367 Patches to these parts should be based on their trees.
368
369 [[patch-flow]]
370 == An ideal patch flow
371
372 Here is an ideal patch flow for this project the current maintainer
373 suggests to the contributors:
374
375 . You come up with an itch. You code it up.
376
377 . Send it to the list and cc people who may need to know about
378 the change.
379 +
380 The people who may need to know are the ones whose code you
381 are butchering. These people happen to be the ones who are
382 most likely to be knowledgeable enough to help you, but
383 they have no obligation to help you (i.e. you ask for help,
384 don't demand). +git log -p {litdd} _$area_you_are_modifying_+ would
385 help you find out who they are.
386
387 . You get comments and suggestions for improvements. You may
388 even get them in a "on top of your change" patch form.
389
390 . Polish, refine, and re-send to the list and the people who
391 spend their time to improve your patch. Go back to step (2).
392
393 . The list forms consensus that the last round of your patch is
394 good. Send it to the maintainer and cc the list.
395
396 . A topic branch is created with the patch and is merged to `next`,
397 and cooked further and eventually graduates to `master`.
398
399 In any time between the (2)-(3) cycle, the maintainer may pick it up
400 from the list and queue it to `pu`, in order to make it easier for
401 people play with it without having to pick up and apply the patch to
402 their trees themselves.
403
404 [[patch-status]]
405 == Know the status of your patch after submission
406
407 * You can use Git itself to find out when your patch is merged in
408 master. `git pull --rebase` will automatically skip already-applied
409 patches, and will let you know. This works only if you rebase on top
410 of the branch in which your patch has been merged (i.e. it will not
411 tell you if your patch is merged in pu if you rebase on top of
412 master).
413
414 * Read the Git mailing list, the maintainer regularly posts messages
415 entitled "What's cooking in git.git" and "What's in git.git" giving
416 the status of various proposed changes.
417
418 [[travis]]
419 == GitHub-Travis CI hints
420
421 With an account at GitHub (you can get one for free to work on open
422 source projects), you can use Travis CI to test your changes on Linux,
423 Mac (and hopefully soon Windows). You can find a successful example
424 test build here: https://travis-ci.org/git/git/builds/120473209
425
426 Follow these steps for the initial setup:
427
428 . Fork https://github.com/git/git to your GitHub account.
429 You can find detailed instructions how to fork here:
430 https://help.github.com/articles/fork-a-repo/
431
432 . Open the Travis CI website: https://travis-ci.org
433
434 . Press the "Sign in with GitHub" button.
435
436 . Grant Travis CI permissions to access your GitHub account.
437 You can find more information about the required permissions here:
438 https://docs.travis-ci.com/user/github-oauth-scopes
439
440 . Open your Travis CI profile page: https://travis-ci.org/profile
441
442 . Enable Travis CI builds for your Git fork.
443
444 After the initial setup, Travis CI will run whenever you push new changes
445 to your fork of Git on GitHub. You can monitor the test state of all your
446 branches here: https://travis-ci.org/__<Your GitHub handle>__/git/branches
447
448 If a branch did not pass all test cases then it is marked with a red
449 cross. In that case you can click on the failing Travis CI job and
450 scroll all the way down in the log. Find the line "<-- Click here to see
451 detailed test output!" and click on the triangle next to the log line
452 number to expand the detailed test output. Here is such a failing
453 example: https://travis-ci.org/git/git/jobs/122676187
454
455 Fix the problem and push your fix to your Git fork. This will trigger
456 a new Travis CI build to ensure all tests pass.
457
458 [[mua]]
459 == MUA specific hints
460
461 Some of patches I receive or pick up from the list share common
462 patterns of breakage. Please make sure your MUA is set up
463 properly not to corrupt whitespaces.
464
465 See the DISCUSSION section of linkgit:git-format-patch[1] for hints on
466 checking your patch by mailing it to yourself and applying with
467 linkgit:git-am[1].
468
469 While you are at it, check the resulting commit log message from
470 a trial run of applying the patch. If what is in the resulting
471 commit is not exactly what you would want to see, it is very
472 likely that your maintainer would end up hand editing the log
473 message when he applies your patch. Things like "Hi, this is my
474 first patch.\n", if you really want to put in the patch e-mail,
475 should come after the three-dash line that signals the end of the
476 commit message.
477
478
479 === Pine
480
481 (Johannes Schindelin)
482
483 ....
484 I don't know how many people still use pine, but for those poor
485 souls it may be good to mention that the quell-flowed-text is
486 needed for recent versions.
487
488 ... the "no-strip-whitespace-before-send" option, too. AFAIK it
489 was introduced in 4.60.
490 ....
491
492 (Linus Torvalds)
493
494 ....
495 And 4.58 needs at least this.
496
497 diff-tree 8326dd8350be64ac7fc805f6563a1d61ad10d32c (from e886a61f76edf5410573e92e38ce22974f9c40f1)
498 Author: Linus Torvalds <torvalds@g5.osdl.org>
499 Date: Mon Aug 15 17:23:51 2005 -0700
500
501 Fix pine whitespace-corruption bug
502
503 There's no excuse for unconditionally removing whitespace from
504 the pico buffers on close.
505
506 diff --git a/pico/pico.c b/pico/pico.c
507 --- a/pico/pico.c
508 +++ b/pico/pico.c
509 @@ -219,7 +219,9 @@ PICO *pm;
510 switch(pico_all_done){ /* prepare for/handle final events */
511 case COMP_EXIT : /* already confirmed */
512 packheader();
513 +#if 0
514 stripwhitespace();
515 +#endif
516 c |= COMP_EXIT;
517 break;
518 ....
519
520 (Daniel Barkalow)
521
522 ....
523 > A patch to SubmittingPatches, MUA specific help section for
524 > users of Pine 4.63 would be very much appreciated.
525
526 Ah, it looks like a recent version changed the default behavior to do the
527 right thing, and inverted the sense of the configuration option. (Either
528 that or Gentoo did it.) So you need to set the
529 "no-strip-whitespace-before-send" option, unless the option you have is
530 "strip-whitespace-before-send", in which case you should avoid checking
531 it.
532 ....
533
534 === Thunderbird, KMail, GMail
535
536 See the MUA-SPECIFIC HINTS section of linkgit:git-format-patch[1].
537
538 === Gnus
539
540 "|" in the `*Summary*` buffer can be used to pipe the current
541 message to an external program, and this is a handy way to drive
542 `git am`. However, if the message is MIME encoded, what is
543 piped into the program is the representation you see in your
544 `*Article*` buffer after unwrapping MIME. This is often not what
545 you would want for two reasons. It tends to screw up non ASCII
546 characters (most notably in people's names), and also
547 whitespaces (fatal in patches). Running "C-u g" to display the
548 message in raw form before using "|" to run the pipe can work
549 this problem around.