Documentation / SubmittingPatcheson commit git-cvsserver runs hooks/post-receive (cdf6328)
   1Checklist (and a short version for the impatient):
   2
   3        Commits:
   4
   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 <your@email.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
  19
  20        Patch:
  21
  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 you use
  38          git-send-email(1), please test it first by sending
  39          email to yourself.
  40
  41Long version:
  42
  43I started reading over the SubmittingPatches document for Linux
  44kernel, primarily because I wanted to have a document similar to
  45it for the core GIT to make sure people understand what they are
  46doing when they write "Signed-off-by" line.
  47
  48But the patch submission requirements are a lot more relaxed
  49here on the technical/contents front, because the core GIT is
  50thousand times smaller ;-).  So here is only the relevant bits.
  51
  52
  53(1) Make separate commits for logically separate changes.
  54
  55Unless your patch is really trivial, you should not be sending
  56out a patch that was generated between your working tree and
  57your commit head.  Instead, always make a commit with complete
  58commit message and generate a series of patches from your
  59repository.  It is a good discipline.
  60
  61Describe the technical detail of the change(s).
  62
  63If your description starts to get too long, that's a sign that you
  64probably need to split up your commit to finer grained pieces.
  65
  66Oh, another thing.  I am picky about whitespaces.  Make sure your
  67changes do not trigger errors with the sample pre-commit hook shipped
  68in templates/hooks--pre-commit.  To help ensure this does not happen,
  69run git diff --check on your changes before you commit.
  70
  71
  72(1a) Try to be nice to older C compilers
  73
  74We try to support wide range of C compilers to compile
  75git with. That means that you should not use C99 initializers, even
  76if a lot of compilers grok it.
  77
  78Also, variables have to be declared at the beginning of the block
  79(you can check this with gcc, using the -Wdeclaration-after-statement
  80option).
  81
  82Another thing: NULL pointers shall be written as NULL, not as 0.
  83
  84
  85(2) Generate your patch using git tools out of your commits.
  86
  87git based diff tools (git, Cogito, and StGIT included) generate
  88unidiff which is the preferred format.
  89
  90You 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
  92receiving end can handle them just fine.
  93
  94Please make sure your patch does not include any extra files
  95which do not belong in a patch submission.  Make sure to review
  96your patch after generating it, to ensure accuracy.  Before
  97sending out, please make sure it cleanly applies to the "master"
  98branch head.  If you are preparing a work based on "next" branch,
  99that is fine, but please mark it as such.
 100
 101
 102(3) Sending your patches.
 103
 104People on the git mailing list need to be able to read and
 105comment on the changes you are submitting.  It is important for
 106a developer to be able to "quote" your changes, using standard
 107e-mail tools, so that they may comment on specific portions of
 108your code.  For this reason, all patches should be submitted
 109"inline".  WARNING: Be wary of your MUAs word-wrap
 110corrupting your patch.  Do not cut-n-paste your patch; you can
 111lose tabs that way if you are not careful.
 112
 113It is a common convention to prefix your subject line with
 114[PATCH].  This lets people easily distinguish patches from other
 115e-mail discussions.
 116
 117"git format-patch" command follows the best current practice to
 118format the body of an e-mail message.  At the beginning of the
 119patch should come your commit message, ending with the
 120Signed-off-by: lines, and a line that consists of three dashes,
 121followed by the diffstat information and the patch itself.  If
 122you are forwarding a patch from somebody else, optionally, at
 123the beginning of the e-mail message just before the commit
 124message starts, you can put a "From: " line to name that person.
 125
 126You often want to add additional explanation about the patch,
 127other than the commit message itself.  Place such "cover letter"
 128material between the three dash lines and the diffstat.
 129
 130Do not attach the patch as a MIME attachment, compressed or not.
 131Do not let your e-mail client send quoted-printable.  Do not let
 132your e-mail client send format=flowed which would destroy
 133whitespaces in your patches. Many
 134popular e-mail applications will not always transmit a MIME
 135attachment as plain text, making it impossible to comment on
 136your code.  A MIME attachment also takes a bit more time to
 137process.  This does not decrease the likelihood of your
 138MIME-attached change being accepted, but it makes it more likely
 139that it will be postponed.
 140
 141Exception:  If your mailer is mangling patches then someone may ask
 142you to re-send them using MIME, that is OK.
 143
 144Do not PGP sign your patch, at least for now.  Most likely, your
 145maintainer or other people on the list would not have your PGP
 146key and would not bother obtaining it anyway.  Your patch is not
 147judged by who you are; a good patch from an unknown origin has a
 148far better chance of being accepted than a patch from a known,
 149respected origin that is done poorly or does incorrect things.
 150
 151If you really really really really want to do a PGP signed
 152patch, format it as "multipart/signed", not a text/plain message
 153that starts with '-----BEGIN PGP SIGNED MESSAGE-----'.  That is
 154not a text/plain, it's something else.
 155
 156Note that your maintainer does not necessarily read everything
 157on the git mailing list.  If your patch is for discussion first,
 158send it "To:" the mailing list, and optionally "cc:" him.  If it
 159is trivially correct or after the list reached a consensus, send
 160it "To:" the maintainer and optionally "cc:" the list.
 161
 162Also note that your maintainer does not actively involve himself in
 163maintaining what are in contrib/ hierarchy.  When you send fixes and
 164enhancements to them, do not forget to "cc: " the person who primarily
 165worked on that hierarchy in contrib/.
 166
 167
 168(4) Sign your work
 169
 170To improve tracking of who did what, we've borrowed the
 171"sign-off" procedure from the Linux kernel project on patches
 172that are being emailed around.  Although core GIT is a lot
 173smaller project it is a good discipline to follow it.
 174
 175The sign-off is a simple line at the end of the explanation for
 176the patch, which certifies that you wrote it or otherwise have
 177the right to pass it on as a open-source patch.  The rules are
 178pretty simple: if you can certify the below:
 179
 180        Developer's Certificate of Origin 1.1
 181
 182        By making a contribution to this project, I certify that:
 183
 184        (a) The contribution was created in whole or in part by me and I
 185            have the right to submit it under the open source license
 186            indicated in the file; or
 187
 188        (b) The contribution is based upon previous work that, to the best
 189            of my knowledge, is covered under an appropriate open source
 190            license and I have the right under that license to submit that
 191            work with modifications, whether created in whole or in part
 192            by me, under the same open source license (unless I am
 193            permitted to submit under a different license), as indicated
 194            in the file; or
 195
 196        (c) The contribution was provided directly to me by some other
 197            person who certified (a), (b) or (c) and I have not modified
 198            it.
 199
 200        (d) I understand and agree that this project and the contribution
 201            are public and that a record of the contribution (including all
 202            personal information I submit with it, including my sign-off) is
 203            maintained indefinitely and may be redistributed consistent with
 204            this project or the open source license(s) involved.
 205
 206then you just add a line saying
 207
 208        Signed-off-by: Random J Developer <random@developer.example.org>
 209
 210This line can be automatically added by git if you run the git-commit
 211command with the -s option.
 212
 213Some people also put extra tags at the end.  They'll just be ignored for
 214now, but you can do this to mark internal company procedures or just
 215point out some special detail about the sign-off.
 216
 217
 218------------------------------------------------
 219MUA specific hints
 220
 221Some of patches I receive or pick up from the list share common
 222patterns of breakage.  Please make sure your MUA is set up
 223properly not to corrupt whitespaces.  Here are two common ones
 224I have seen:
 225
 226* Empty context lines that do not have _any_ whitespace.
 227
 228* Non empty context lines that have one extra whitespace at the
 229  beginning.
 230
 231One test you could do yourself if your MUA is set up correctly is:
 232
 233* Send the patch to yourself, exactly the way you would, except
 234  To: and Cc: lines, which would not contain the list and
 235  maintainer address.
 236
 237* Save that patch to a file in UNIX mailbox format.  Call it say
 238  a.patch.
 239
 240* Try to apply to the tip of the "master" branch from the
 241  git.git public repository:
 242
 243    $ git fetch http://kernel.org/pub/scm/git/git.git master:test-apply
 244    $ git checkout test-apply
 245    $ git reset --hard
 246    $ git am a.patch
 247
 248If it does not apply correctly, there can be various reasons.
 249
 250* Your patch itself does not apply cleanly.  That is _bad_ but
 251  does not have much to do with your MUA.  Please rebase the
 252  patch appropriately.
 253
 254* Your MUA corrupted your patch; "am" would complain that
 255  the patch does not apply.  Look at .dotest/ subdirectory and
 256  see what 'patch' file contains and check for the common
 257  corruption patterns mentioned above.
 258
 259* While you are at it, check what are in 'info' and
 260  'final-commit' files as well.  If what is in 'final-commit' is
 261  not exactly what you would want to see in the commit log
 262  message, it is very likely that your maintainer would end up
 263  hand editing the log message when he applies your patch.
 264  Things like "Hi, this is my first patch.\n", if you really
 265  want to put in the patch e-mail, should come after the
 266  three-dash line that signals the end of the commit message.
 267
 268
 269Pine
 270----
 271
 272(Johannes Schindelin)
 273
 274I don't know how many people still use pine, but for those poor
 275souls it may be good to mention that the quell-flowed-text is
 276needed for recent versions.
 277
 278... the "no-strip-whitespace-before-send" option, too. AFAIK it
 279was introduced in 4.60.
 280
 281(Linus Torvalds)
 282
 283And 4.58 needs at least this.
 284
 285---
 286diff-tree 8326dd8350be64ac7fc805f6563a1d61ad10d32c (from e886a61f76edf5410573e92e38ce22974f9c40f1)
 287Author: Linus Torvalds <torvalds@g5.osdl.org>
 288Date:   Mon Aug 15 17:23:51 2005 -0700
 289
 290    Fix pine whitespace-corruption bug
 291
 292    There's no excuse for unconditionally removing whitespace from
 293    the pico buffers on close.
 294
 295diff --git a/pico/pico.c b/pico/pico.c
 296--- a/pico/pico.c
 297+++ b/pico/pico.c
 298@@ -219,7 +219,9 @@ PICO *pm;
 299            switch(pico_all_done){      /* prepare for/handle final events */
 300              case COMP_EXIT :          /* already confirmed */
 301                packheader();
 302+#if 0
 303                stripwhitespace();
 304+#endif
 305                c |= COMP_EXIT;
 306                break;
 307
 308
 309(Daniel Barkalow)
 310
 311> A patch to SubmittingPatches, MUA specific help section for
 312> users of Pine 4.63 would be very much appreciated.
 313
 314Ah, it looks like a recent version changed the default behavior to do the
 315right thing, and inverted the sense of the configuration option. (Either
 316that or Gentoo did it.) So you need to set the
 317"no-strip-whitespace-before-send" option, unless the option you have is
 318"strip-whitespace-before-send", in which case you should avoid checking
 319it.
 320
 321
 322Thunderbird
 323-----------
 324
 325(A Large Angry SCM)
 326
 327Here are some hints on how to successfully submit patches inline using
 328Thunderbird.
 329
 330This recipe appears to work with the current [*1*] Thunderbird from Suse.
 331
 332The following Thunderbird extensions are needed:
 333        AboutConfig 0.5
 334                http://aboutconfig.mozdev.org/
 335        External Editor 0.7.2
 336                http://globs.org/articles.php?lng=en&pg=8
 337
 3381) Prepare the patch as a text file using your method of choice.
 339
 3402) Before opening a compose window, use Edit->Account Settings to
 341uncheck the "Compose messages in HTML format" setting in the
 342"Composition & Addressing" panel of the account to be used to send the
 343patch. [*2*]
 344
 3453) In the main Thunderbird window, _before_ you open the compose window
 346for the patch, use Tools->about:config to set the following to the
 347indicated values:
 348        mailnews.send_plaintext_flowed  => false
 349        mailnews.wraplength             => 0
 350
 3514) Open a compose window and click the external editor icon.
 352
 3535) In the external editor window, read in the patch file and exit the
 354editor normally.
 355
 3566) Back in the compose window: Add whatever other text you wish to the
 357message, complete the addressing and subject fields, and press send.
 358
 3597) Optionally, undo the about:config/account settings changes made in
 360steps 2 & 3.
 361
 362
 363[Footnotes]
 364*1* Version 1.0 (20041207) from the MozillaThunderbird-1.0-5 rpm of Suse
 3659.3 professional updates.
 366
 367*2* It may be possible to do this with about:config and the following
 368settings but I haven't tried, yet.
 369        mail.html_compose                       => false
 370        mail.identity.default.compose_html      => false
 371        mail.identity.id?.compose_html          => false
 372
 373
 374Gnus
 375----
 376
 377'|' in the *Summary* buffer can be used to pipe the current
 378message to an external program, and this is a handy way to drive
 379"git am".  However, if the message is MIME encoded, what is
 380piped into the program is the representation you see in your
 381*Article* buffer after unwrapping MIME.  This is often not what
 382you would want for two reasons.  It tends to screw up non ASCII
 383characters (most notably in people's names), and also
 384whitespaces (fatal in patches).  Running 'C-u g' to display the
 385message in raw form before using '|' to run the pipe can work
 386this problem around.
 387
 388
 389KMail
 390-----
 391
 392This should help you to submit patches inline using KMail.
 393
 3941) Prepare the patch as a text file.
 395
 3962) Click on New Mail.
 397
 3983) Go under "Options" in the Composer window and be sure that
 399"Word wrap" is not set.
 400
 4014) Use Message -> Insert file... and insert the patch.
 402
 4035) Back in the compose window: add whatever other text you wish to the
 404message, complete the addressing and subject fields, and press send.