Make git-prune submodule aware (and fix a SEGFAULT in the process)
[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 <>" 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
18 Patch:
20 - use "git format-patch -M" to create the patch
21 - send your patch to <>. If you use
22 git-send-email(1), please test it first by sending
23 email to yourself.
24 - do not PGP sign your patch
25 - do not attach your patch, but read in the mail
26 body, unless you cannot teach your mailer to
27 leave the formatting of the patch alone.
28 - be careful doing cut & paste into your mailer, not to
29 corrupt whitespaces.
30 - provide additional information (which is unsuitable for
31 the commit message) between the "---" and the diffstat
32 - send the patch to the list _and_ the maintainer
33 - if you change, add, or remove a command line option or
34 make some other user interface change, the associated
35 documentation should be updated as well.
37 Long version:
39 I started reading over the SubmittingPatches document for Linux
40 kernel, primarily because I wanted to have a document similar to
41 it for the core GIT to make sure people understand what they are
42 doing when they write "Signed-off-by" line.
44 But the patch submission requirements are a lot more relaxed
45 here on the technical/contents front, because the core GIT is
46 thousand times smaller ;-). So here is only the relevant bits.
49 (1) Make separate commits for logically separate changes.
51 Unless your patch is really trivial, you should not be sending
52 out a patch that was generated between your working tree and
53 your commit head. Instead, always make a commit with complete
54 commit message and generate a series of patches from your
55 repository. It is a good discipline.
57 Describe the technical detail of the change(s).
59 If your description starts to get too long, that's a sign that you
60 probably need to split up your commit to finer grained pieces.
62 Oh, another thing. I am picky about whitespaces. Make sure your
63 changes do not trigger errors with the sample pre-commit hook shipped
64 in templates/hooks--pre-commit. To help ensure this does not happen,
65 run git diff --check on your changes before you commit.
68 (1a) Try to be nice to older C compilers
70 We try to support wide range of C compilers to compile
71 git with. That means that you should not use C99 initializers, even
72 if a lot of compilers grok it.
74 Also, variables have to be declared at the beginning of the block
75 (you can check this with gcc, using the -Wdeclaration-after-statement
76 option).
78 Another thing: NULL pointers shall be written as NULL, not as 0.
81 (2) Generate your patch using git tools out of your commits.
83 git based diff tools (git, Cogito, and StGIT included) generate
84 unidiff which is the preferred format.
86 You do not have to be afraid to use -M option to "git diff" or
87 "git format-patch", if your patch involves file renames. The
88 receiving end can handle them just fine.
90 Please make sure your patch does not include any extra files
91 which do not belong in a patch submission. Make sure to review
92 your patch after generating it, to ensure accuracy. Before
93 sending out, please make sure it cleanly applies to the "master"
94 branch head. If you are preparing a work based on "next" branch,
95 that is fine, but please mark it as such.
98 (3) Sending your patches.
100 People on the git mailing list need to be able to read and
101 comment on the changes you are submitting. It is important for
102 a developer to be able to "quote" your changes, using standard
103 e-mail tools, so that they may comment on specific portions of
104 your code. For this reason, all patches should be submitted
105 "inline". WARNING: Be wary of your MUAs word-wrap
106 corrupting your patch. Do not cut-n-paste your patch; you can
107 lose tabs that way if you are not careful.
109 It is a common convention to prefix your subject line with
110 [PATCH]. This lets people easily distinguish patches from other
111 e-mail discussions.
113 "git format-patch" command follows the best current practice to
114 format the body of an e-mail message. At the beginning of the
115 patch should come your commit message, ending with the
116 Signed-off-by: lines, and a line that consists of three dashes,
117 followed by the diffstat information and the patch itself. If
118 you are forwarding a patch from somebody else, optionally, at
119 the beginning of the e-mail message just before the commit
120 message starts, you can put a "From: " line to name that person.
122 You often want to add additional explanation about the patch,
123 other than the commit message itself. Place such "cover letter"
124 material between the three dash lines and the diffstat.
126 Do not attach the patch as a MIME attachment, compressed or not.
127 Do not let your e-mail client send quoted-printable. Do not let
128 your e-mail client send format=flowed which would destroy
129 whitespaces in your patches. Many
130 popular e-mail applications will not always transmit a MIME
131 attachment as plain text, making it impossible to comment on
132 your code. A MIME attachment also takes a bit more time to
133 process. This does not decrease the likelihood of your
134 MIME-attached change being accepted, but it makes it more likely
135 that it will be postponed.
137 Exception: If your mailer is mangling patches then someone may ask
138 you to re-send them using MIME, that is OK.
140 Do not PGP sign your patch, at least for now. Most likely, your
141 maintainer or other people on the list would not have your PGP
142 key and would not bother obtaining it anyway. Your patch is not
143 judged by who you are; a good patch from an unknown origin has a
144 far better chance of being accepted than a patch from a known,
145 respected origin that is done poorly or does incorrect things.
147 If you really really really really want to do a PGP signed
148 patch, format it as "multipart/signed", not a text/plain message
149 that starts with '-----BEGIN PGP SIGNED MESSAGE-----'. That is
150 not a text/plain, it's something else.
152 Note that your maintainer does not necessarily read everything
153 on the git mailing list. If your patch is for discussion first,
154 send it "To:" the mailing list, and optionally "cc:" him. If it
155 is trivially correct or after the list reached a consensus, send
156 it "To:" the maintainer and optionally "cc:" the list.
158 Also note that your maintainer does not actively involve himself in
159 maintaining what are in contrib/ hierarchy. When you send fixes and
160 enhancements to them, do not forget to "cc: " the person who primarily
161 worked on that hierarchy in contrib/.
164 (4) Sign your work
166 To improve tracking of who did what, we've borrowed the
167 "sign-off" procedure from the Linux kernel project on patches
168 that are being emailed around. Although core GIT is a lot
169 smaller project it is a good discipline to follow it.
171 The sign-off is a simple line at the end of the explanation for
172 the patch, which certifies that you wrote it or otherwise have
173 the right to pass it on as a open-source patch. The rules are
174 pretty simple: if you can certify the below:
176 Developer's Certificate of Origin 1.1
178 By making a contribution to this project, I certify that:
180 (a) The contribution was created in whole or in part by me and I
181 have the right to submit it under the open source license
182 indicated in the file; or
184 (b) The contribution is based upon previous work that, to the best
185 of my knowledge, is covered under an appropriate open source
186 license and I have the right under that license to submit that
187 work with modifications, whether created in whole or in part
188 by me, under the same open source license (unless I am
189 permitted to submit under a different license), as indicated
190 in the file; or
192 (c) The contribution was provided directly to me by some other
193 person who certified (a), (b) or (c) and I have not modified
194 it.
196 (d) I understand and agree that this project and the contribution
197 are public and that a record of the contribution (including all
198 personal information I submit with it, including my sign-off) is
199 maintained indefinitely and may be redistributed consistent with
200 this project or the open source license(s) involved.
202 then you just add a line saying
204 Signed-off-by: Random J Developer <>
206 This line can be automatically added by git if you run the git-commit
207 command with the -s option.
209 Some people also put extra tags at the end. They'll just be ignored for
210 now, but you can do this to mark internal company procedures or just
211 point out some special detail about the sign-off.
214 ------------------------------------------------
215 MUA specific hints
217 Some of patches I receive or pick up from the list share common
218 patterns of breakage. Please make sure your MUA is set up
219 properly not to corrupt whitespaces. Here are two common ones
220 I have seen:
222 * Empty context lines that do not have _any_ whitespace.
224 * Non empty context lines that have one extra whitespace at the
225 beginning.
227 One test you could do yourself if your MUA is set up correctly is:
229 * Send the patch to yourself, exactly the way you would, except
230 To: and Cc: lines, which would not contain the list and
231 maintainer address.
233 * Save that patch to a file in UNIX mailbox format. Call it say
234 a.patch.
236 * Try to apply to the tip of the "master" branch from the
237 git.git public repository:
239 $ git fetch master:test-apply
240 $ git checkout test-apply
241 $ git reset --hard
242 $ git applymbox a.patch
244 If it does not apply correctly, there can be various reasons.
246 * Your patch itself does not apply cleanly. That is _bad_ but
247 does not have much to do with your MUA. Please rebase the
248 patch appropriately.
250 * Your MUA corrupted your patch; applymbox would complain that
251 the patch does not apply. Look at .dotest/ subdirectory and
252 see what 'patch' file contains and check for the common
253 corruption patterns mentioned above.
255 * While you are at it, check what are in 'info' and
256 'final-commit' files as well. If what is in 'final-commit' is
257 not exactly what you would want to see in the commit log
258 message, it is very likely that your maintainer would end up
259 hand editing the log message when he applies your patch.
260 Things like "Hi, this is my first patch.\n", if you really
261 want to put in the patch e-mail, should come after the
262 three-dash line that signals the end of the commit message.
265 Pine
266 ----
268 (Johannes Schindelin)
270 I don't know how many people still use pine, but for those poor
271 souls it may be good to mention that the quell-flowed-text is
272 needed for recent versions.
274 ... the "no-strip-whitespace-before-send" option, too. AFAIK it
275 was introduced in 4.60.
277 (Linus Torvalds)
279 And 4.58 needs at least this.
281 ---
282 diff-tree 8326dd8350be64ac7fc805f6563a1d61ad10d32c (from e886a61f76edf5410573e92e38ce22974f9c40f1)
283 Author: Linus Torvalds <>
284 Date: Mon Aug 15 17:23:51 2005 -0700
286 Fix pine whitespace-corruption bug
288 There's no excuse for unconditionally removing whitespace from
289 the pico buffers on close.
291 diff --git a/pico/pico.c b/pico/pico.c
292 --- a/pico/pico.c
293 +++ b/pico/pico.c
294 @@ -219,7 +219,9 @@ PICO *pm;
295 switch(pico_all_done){ /* prepare for/handle final events */
296 case COMP_EXIT : /* already confirmed */
297 packheader();
298 +#if 0
299 stripwhitespace();
300 +#endif
301 c |= COMP_EXIT;
302 break;
305 (Daniel Barkalow)
307 > A patch to SubmittingPatches, MUA specific help section for
308 > users of Pine 4.63 would be very much appreciated.
310 Ah, it looks like a recent version changed the default behavior to do the
311 right thing, and inverted the sense of the configuration option. (Either
312 that or Gentoo did it.) So you need to set the
313 "no-strip-whitespace-before-send" option, unless the option you have is
314 "strip-whitespace-before-send", in which case you should avoid checking
315 it.
318 Thunderbird
319 -----------
321 (A Large Angry SCM)
323 Here are some hints on how to successfully submit patches inline using
324 Thunderbird.
326 This recipe appears to work with the current [*1*] Thunderbird from Suse.
328 The following Thunderbird extensions are needed:
329 AboutConfig 0.5
331 External Editor 0.7.2
334 1) Prepare the patch as a text file using your method of choice.
336 2) Before opening a compose window, use Edit->Account Settings to
337 uncheck the "Compose messages in HTML format" setting in the
338 "Composition & Addressing" panel of the account to be used to send the
339 patch. [*2*]
341 3) In the main Thunderbird window, _before_ you open the compose window
342 for the patch, use Tools->about:config to set the following to the
343 indicated values:
344 mailnews.send_plaintext_flowed => false
345 mailnews.wraplength => 0
347 4) Open a compose window and click the external editor icon.
349 5) In the external editor window, read in the patch file and exit the
350 editor normally.
352 6) Back in the compose window: Add whatever other text you wish to the
353 message, complete the addressing and subject fields, and press send.
355 7) Optionally, undo the about:config/account settings changes made in
356 steps 2 & 3.
359 [Footnotes]
360 *1* Version 1.0 (20041207) from the MozillaThunderbird-1.0-5 rpm of Suse
361 9.3 professional updates.
363 *2* It may be possible to do this with about:config and the following
364 settings but I haven't tried, yet.
365 mail.html_compose => false
366 mail.identity.default.compose_html => false
367 => false
370 Gnus
371 ----
373 '|' in the *Summary* buffer can be used to pipe the current
374 message to an external program, and this is a handy way to drive
375 "git am". However, if the message is MIME encoded, what is
376 piped into the program is the representation you see in your
377 *Article* buffer after unwrapping MIME. This is often not what
378 you would want for two reasons. It tends to screw up non ASCII
379 characters (most notably in people's names), and also
380 whitespaces (fatal in patches). Running 'C-u g' to display the
381 message in raw form before using '|' to run the pipe can work
382 this problem around.
385 KMail
386 -----
388 This should help you to submit patches inline using KMail.
390 1) Prepare the patch as a text file.
392 2) Click on New Mail.
394 3) Go under "Options" in the Composer window and be sure that
395 "Word wrap" is not set.
397 4) Use Message -> Insert file... and insert the patch.
399 5) Back in the compose window: add whatever other text you wish to the
400 message, complete the addressing and subject fields, and press send.