[git/git.git] / Documentation / SubmittingPatches
1 Checklist (and a short version for the impatient):
3 Commits:
5 - make commits of logical units
6 - check for unnecessary whitespace with "git diff --check"
7 before committing
8 - do not check in commented out code or unneeded files
9 - provide a meaningful commit message
10 - the first line of the commit message should be a short
11 description and should skip the full stop
12 - if you want your work included in git.git, add a
13 "Signed-off-by: Your Name <you@example.com>" line to the
14 commit message (or just use the option "-s" when
15 committing) to confirm that you agree to the Developer's
16 Certificate of Origin
17 - make sure that you have tests for the bug you are fixing
18 - make sure that the test suite passes after your commit
20 Patch:
22 - use "git format-patch -M" to create the patch
23 - do not PGP sign your patch
24 - do not attach your patch, but read in the mail
25 body, unless you cannot teach your mailer to
26 leave the formatting of the patch alone.
27 - be careful doing cut & paste into your mailer, not to
28 corrupt whitespaces.
29 - provide additional information (which is unsuitable for
30 the commit message) between the "---" and the diffstat
31 - if you change, add, or remove a command line option or
32 make some other user interface change, the associated
33 documentation should be updated as well.
34 - if your name is not writable in ASCII, make sure that
35 you send off a message in the correct encoding.
36 - send the patch to the list (git@vger.kernel.org) and the
37 maintainer (gitster@pobox.com) if (and only if) the patch
38 is ready for inclusion. If you use git-send-email(1),
39 please test it first by sending email to yourself.
41 Long version:
43 I started reading over the SubmittingPatches document for Linux
44 kernel, primarily because I wanted to have a document similar to
45 it for the core GIT to make sure people understand what they are
46 doing when they write "Signed-off-by" line.
48 But the patch submission requirements are a lot more relaxed
49 here on the technical/contents front, because the core GIT is
50 thousand times smaller ;-). So here is only the relevant bits.
53 (1) Make separate commits for logically separate changes.
55 Unless your patch is really trivial, you should not be sending
56 out a patch that was generated between your working tree and
57 your commit head. Instead, always make a commit with complete
58 commit message and generate a series of patches from your
59 repository. It is a good discipline.
61 Describe the technical detail of the change(s).
63 If your description starts to get too long, that's a sign that you
64 probably need to split up your commit to finer grained pieces.
66 Oh, another thing. I am picky about whitespaces. Make sure your
67 changes do not trigger errors with the sample pre-commit hook shipped
68 in templates/hooks--pre-commit. To help ensure this does not happen,
69 run git diff --check on your changes before you commit.
72 (1a) Try to be nice to older C compilers
74 We try to support wide range of C compilers to compile
75 git with. That means that you should not use C99 initializers, even
76 if a lot of compilers grok it.
78 Also, variables have to be declared at the beginning of the block
79 (you can check this with gcc, using the -Wdeclaration-after-statement
80 option).
82 Another thing: NULL pointers shall be written as NULL, not as 0.
85 (2) Generate your patch using git tools out of your commits.
87 git based diff tools (git, Cogito, and StGIT included) generate
88 unidiff which is the preferred format.
90 You do not have to be afraid to use -M option to "git diff" or
91 "git format-patch", if your patch involves file renames. The
92 receiving end can handle them just fine.
94 Please make sure your patch does not include any extra files
95 which do not belong in a patch submission. Make sure to review
96 your patch after generating it, to ensure accuracy. Before
97 sending out, please make sure it cleanly applies to the "master"
98 branch head. If you are preparing a work based on "next" branch,
99 that is fine, but please mark it as such.
102 (3) Sending your patches.
104 People on the git mailing list need to be able to read and
105 comment on the changes you are submitting. It is important for
106 a developer to be able to "quote" your changes, using standard
107 e-mail tools, so that they may comment on specific portions of
108 your code. For this reason, all patches should be submitted
109 "inline". WARNING: Be wary of your MUAs word-wrap
110 corrupting your patch. Do not cut-n-paste your patch; you can
111 lose tabs that way if you are not careful.
113 It is a common convention to prefix your subject line with
114 [PATCH]. This lets people easily distinguish patches from other
115 e-mail discussions. Use of additional markers after PATCH and
116 the closing bracket to mark the nature of the patch is also
117 encouraged. E.g. [PATCH/RFC] is often used when the patch is
118 not ready to be applied but it is for discussion, [PATCH v2],
119 [PATCH v3] etc. are often seen when you are sending an update to
120 what you have previously sent.
122 "git format-patch" command follows the best current practice to
123 format the body of an e-mail message. At the beginning of the
124 patch should come your commit message, ending with the
125 Signed-off-by: lines, and a line that consists of three dashes,
126 followed by the diffstat information and the patch itself. If
127 you are forwarding a patch from somebody else, optionally, at
128 the beginning of the e-mail message just before the commit
129 message starts, you can put a "From: " line to name that person.
131 You often want to add additional explanation about the patch,
132 other than the commit message itself. Place such "cover letter"
133 material between the three dash lines and the diffstat.
135 Do not attach the patch as a MIME attachment, compressed or not.
136 Do not let your e-mail client send quoted-printable. Do not let
137 your e-mail client send format=flowed which would destroy
138 whitespaces in your patches. Many
139 popular e-mail applications will not always transmit a MIME
140 attachment as plain text, making it impossible to comment on
141 your code. A MIME attachment also takes a bit more time to
142 process. This does not decrease the likelihood of your
143 MIME-attached change being accepted, but it makes it more likely
144 that it will be postponed.
146 Exception: If your mailer is mangling patches then someone may ask
147 you to re-send them using MIME, that is OK.
149 Do not PGP sign your patch, at least for now. Most likely, your
150 maintainer or other people on the list would not have your PGP
151 key and would not bother obtaining it anyway. Your patch is not
152 judged by who you are; a good patch from an unknown origin has a
153 far better chance of being accepted than a patch from a known,
154 respected origin that is done poorly or does incorrect things.
156 If you really really really really want to do a PGP signed
157 patch, format it as "multipart/signed", not a text/plain message
158 that starts with '-----BEGIN PGP SIGNED MESSAGE-----'. That is
159 not a text/plain, it's something else.
161 Note that your maintainer does not necessarily read everything
162 on the git mailing list. If your patch is for discussion first,
163 send it "To:" the mailing list, and optionally "cc:" him. If it
164 is trivially correct or after the list reached a consensus, send
165 it "To:" the maintainer and optionally "cc:" the list for
166 inclusion.
168 Also note that your maintainer does not actively involve himself in
169 maintaining what are in contrib/ hierarchy. When you send fixes and
170 enhancements to them, do not forget to "cc: " the person who primarily
171 worked on that hierarchy in contrib/.
174 (4) Sign your work
176 To improve tracking of who did what, we've borrowed the
177 "sign-off" procedure from the Linux kernel project on patches
178 that are being emailed around. Although core GIT is a lot
179 smaller project it is a good discipline to follow it.
181 The sign-off is a simple line at the end of the explanation for
182 the patch, which certifies that you wrote it or otherwise have
183 the right to pass it on as a open-source patch. The rules are
184 pretty simple: if you can certify the below:
186 Developer's Certificate of Origin 1.1
188 By making a contribution to this project, I certify that:
190 (a) The contribution was created in whole or in part by me and I
191 have the right to submit it under the open source license
192 indicated in the file; or
194 (b) The contribution is based upon previous work that, to the best
195 of my knowledge, is covered under an appropriate open source
196 license and I have the right under that license to submit that
197 work with modifications, whether created in whole or in part
198 by me, under the same open source license (unless I am
199 permitted to submit under a different license), as indicated
200 in the file; or
202 (c) The contribution was provided directly to me by some other
203 person who certified (a), (b) or (c) and I have not modified
204 it.
206 (d) I understand and agree that this project and the contribution
207 are public and that a record of the contribution (including all
208 personal information I submit with it, including my sign-off) is
209 maintained indefinitely and may be redistributed consistent with
210 this project or the open source license(s) involved.
212 then you just add a line saying
214 Signed-off-by: Random J Developer <random@developer.example.org>
216 This line can be automatically added by git if you run the git-commit
217 command with the -s option.
219 Notice that you can place your own Signed-off-by: line when
220 forwarding somebody else's patch with the above rules for
221 D-C-O. Indeed you are encouraged to do so. Do not forget to
222 place an in-body "From: " line at the beginning to properly attribute
223 the change to its true author (see (2) above).
225 Some people also put extra tags at the end.
227 "Acked-by:" says that the patch was reviewed by the person who
228 is more familiar with the issues and the area the patch attempts
229 to modify. "Tested-by:" says the patch was tested by the person
230 and found to have the desired effect.
232 ------------------------------------------------
233 An ideal patch flow
235 Here is an ideal patch flow for this project the current maintainer
236 suggests to the contributors:
238 (0) You come up with an itch. You code it up.
240 (1) Send it to the list and cc people who may need to know about
241 the change.
243 The people who may need to know are the ones whose code you
244 are butchering. These people happen to be the ones who are
245 most likely to be knowledgeable enough to help you, but
246 they have no obligation to help you (i.e. you ask for help,
247 don't demand). "git log -p -- $area_you_are_modifying" would
248 help you find out who they are.
250 (2) You get comments and suggestions for improvements. You may
251 even get them in a "on top of your change" patch form.
253 (3) Polish, refine, and re-send to the list and the people who
254 spend their time to improve your patch. Go back to step (2).
256 (4) The list forms consensus that the last round of your patch is
257 good. Send it to the list and cc the maintainer.
259 (5) A topic branch is created with the patch and is merged to 'next',
260 and cooked further and eventually graduates to 'master'.
262 In any time between the (2)-(3) cycle, the maintainer may pick it up
263 from the list and queue it to 'pu', in order to make it easier for
264 people play with it without having to pick up and apply the patch to
265 their trees themselves.
267 ------------------------------------------------
268 MUA specific hints
270 Some of patches I receive or pick up from the list share common
271 patterns of breakage. Please make sure your MUA is set up
272 properly not to corrupt whitespaces. Here are two common ones
273 I have seen:
275 * Empty context lines that do not have _any_ whitespace.
277 * Non empty context lines that have one extra whitespace at the
278 beginning.
280 One test you could do yourself if your MUA is set up correctly is:
282 * Send the patch to yourself, exactly the way you would, except
283 To: and Cc: lines, which would not contain the list and
284 maintainer address.
286 * Save that patch to a file in UNIX mailbox format. Call it say
287 a.patch.
289 * Try to apply to the tip of the "master" branch from the
290 git.git public repository:
292 $ git fetch http://kernel.org/pub/scm/git/git.git master:test-apply
293 $ git checkout test-apply
294 $ git reset --hard
295 $ git am a.patch
297 If it does not apply correctly, there can be various reasons.
299 * Your patch itself does not apply cleanly. That is _bad_ but
300 does not have much to do with your MUA. Please rebase the
301 patch appropriately.
303 * Your MUA corrupted your patch; "am" would complain that
304 the patch does not apply. Look at .git/rebase-apply/ subdirectory and
305 see what 'patch' file contains and check for the common
306 corruption patterns mentioned above.
308 * While you are at it, check what are in 'info' and
309 'final-commit' files as well. If what is in 'final-commit' is
310 not exactly what you would want to see in the commit log
311 message, it is very likely that your maintainer would end up
312 hand editing the log message when he applies your patch.
313 Things like "Hi, this is my first patch.\n", if you really
314 want to put in the patch e-mail, should come after the
315 three-dash line that signals the end of the commit message.
318 Pine
319 ----
321 (Johannes Schindelin)
323 I don't know how many people still use pine, but for those poor
324 souls it may be good to mention that the quell-flowed-text is
325 needed for recent versions.
327 ... the "no-strip-whitespace-before-send" option, too. AFAIK it
328 was introduced in 4.60.
330 (Linus Torvalds)
332 And 4.58 needs at least this.
334 ---
335 diff-tree 8326dd8350be64ac7fc805f6563a1d61ad10d32c (from e886a61f76edf5410573e92e38ce22974f9c40f1)
336 Author: Linus Torvalds <torvalds@g5.osdl.org>
337 Date: Mon Aug 15 17:23:51 2005 -0700
339 Fix pine whitespace-corruption bug
341 There's no excuse for unconditionally removing whitespace from
342 the pico buffers on close.
344 diff --git a/pico/pico.c b/pico/pico.c
345 --- a/pico/pico.c
346 +++ b/pico/pico.c
347 @@ -219,7 +219,9 @@ PICO *pm;
348 switch(pico_all_done){ /* prepare for/handle final events */
349 case COMP_EXIT : /* already confirmed */
350 packheader();
351 +#if 0
352 stripwhitespace();
353 +#endif
354 c |= COMP_EXIT;
355 break;
358 (Daniel Barkalow)
360 > A patch to SubmittingPatches, MUA specific help section for
361 > users of Pine 4.63 would be very much appreciated.
363 Ah, it looks like a recent version changed the default behavior to do the
364 right thing, and inverted the sense of the configuration option. (Either
365 that or Gentoo did it.) So you need to set the
366 "no-strip-whitespace-before-send" option, unless the option you have is
367 "strip-whitespace-before-send", in which case you should avoid checking
368 it.
371 Thunderbird
372 -----------
374 (A Large Angry SCM)
376 Here are some hints on how to successfully submit patches inline using
377 Thunderbird.
379 This recipe appears to work with the current [*1*] Thunderbird from Suse.
381 The following Thunderbird extensions are needed:
382 AboutConfig 0.5
383 http://aboutconfig.mozdev.org/
384 External Editor 0.7.2
385 http://globs.org/articles.php?lng=en&pg=8
387 1) Prepare the patch as a text file using your method of choice.
389 2) Before opening a compose window, use Edit->Account Settings to
390 uncheck the "Compose messages in HTML format" setting in the
391 "Composition & Addressing" panel of the account to be used to send the
392 patch. [*2*]
394 3) In the main Thunderbird window, _before_ you open the compose window
395 for the patch, use Tools->about:config to set the following to the
396 indicated values:
397 mailnews.send_plaintext_flowed => false
398 mailnews.wraplength => 0
400 4) Open a compose window and click the external editor icon.
402 5) In the external editor window, read in the patch file and exit the
403 editor normally.
405 6) Back in the compose window: Add whatever other text you wish to the
406 message, complete the addressing and subject fields, and press send.
408 7) Optionally, undo the about:config/account settings changes made in
409 steps 2 & 3.
412 [Footnotes]
413 *1* Version 1.0 (20041207) from the MozillaThunderbird-1.0-5 rpm of Suse
414 9.3 professional updates.
416 *2* It may be possible to do this with about:config and the following
417 settings but I haven't tried, yet.
418 mail.html_compose => false
419 mail.identity.default.compose_html => false
420 mail.identity.id?.compose_html => false
422 (Lukas Sandström)
424 There is a script in contrib/thunderbird-patch-inline which can help
425 you include patches with Thunderbird in an easy way. To use it, do the
426 steps above and then use the script as the external editor.
428 Gnus
429 ----
431 '|' in the *Summary* buffer can be used to pipe the current
432 message to an external program, and this is a handy way to drive
433 "git am". However, if the message is MIME encoded, what is
434 piped into the program is the representation you see in your
435 *Article* buffer after unwrapping MIME. This is often not what
436 you would want for two reasons. It tends to screw up non ASCII
437 characters (most notably in people's names), and also
438 whitespaces (fatal in patches). Running 'C-u g' to display the
439 message in raw form before using '|' to run the pipe can work
440 this problem around.
443 KMail
444 -----
446 This should help you to submit patches inline using KMail.
448 1) Prepare the patch as a text file.
450 2) Click on New Mail.
452 3) Go under "Options" in the Composer window and be sure that
453 "Word wrap" is not set.
455 4) Use Message -> Insert file... and insert the patch.
457 5) Back in the compose window: add whatever other text you wish to the
458 message, complete the addressing and subject fields, and press send.