Documentation / SubmittingPatcheson commit Merge branch 'ms/subtree-fixlets' (b81827b)
   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        - the first line of the commit message should be a short
  10          description (50 characters is the soft limit, see DISCUSSION
  11          in git-commit(1)), and should skip the full stop
  12        - it is also conventional in most cases to prefix the
  13          first line with "area: " where the area is a filename
  14          or identifier for the general area of the code being
  15          modified, e.g.
  16          . archive: ustar header checksum is computed unsigned
  17          . git-cherry-pick.txt: clarify the use of revision range notation
  18          (if in doubt which identifier to use, run "git log --no-merges"
  19          on the files you are modifying to see the current conventions)
  20        - the body should provide a meaningful commit message, which:
  21          . explains the problem the change tries to solve, iow, what
  22            is wrong with the current code without the change.
  23          . justifies the way the change solves the problem, iow, why
  24            the result with the change is better.
  25          . alternate solutions considered but discarded, if any.
  26        - describe changes in imperative mood, e.g. "make xyzzy do frotz"
  27          instead of "[This patch] makes xyzzy do frotz" or "[I] changed
  28          xyzzy to do frotz", as if you are giving orders to the codebase
  29          to change its behaviour.
  30        - try to make sure your explanation can be understood without
  31          external resources. Instead of giving a URL to a mailing list
  32          archive, summarize the relevant points of the discussion.
  33        - add a "Signed-off-by: Your Name <you@example.com>" line to the
  34          commit message (or just use the option "-s" when committing)
  35          to confirm that you agree to the Developer's Certificate of Origin
  36        - make sure that you have tests for the bug you are fixing
  37        - make sure that the test suite passes after your commit
  38
  39        Patch:
  40
  41        - use "git format-patch -M" to create the patch
  42        - do not PGP sign your patch
  43        - do not attach your patch, but read in the mail
  44          body, unless you cannot teach your mailer to
  45          leave the formatting of the patch alone.
  46        - be careful doing cut & paste into your mailer, not to
  47          corrupt whitespaces.
  48        - provide additional information (which is unsuitable for
  49          the commit message) between the "---" and the diffstat
  50        - if you change, add, or remove a command line option or
  51          make some other user interface change, the associated
  52          documentation should be updated as well.
  53        - if your name is not writable in ASCII, make sure that
  54          you send off a message in the correct encoding.
  55        - send the patch to the list (git@vger.kernel.org) and the
  56          maintainer (gitster@pobox.com) if (and only if) the patch
  57          is ready for inclusion. If you use git-send-email(1),
  58          please test it first by sending email to yourself.
  59        - see below for instructions specific to your mailer
  60
  61Long version:
  62
  63I started reading over the SubmittingPatches document for Linux
  64kernel, primarily because I wanted to have a document similar to
  65it for the core GIT to make sure people understand what they are
  66doing when they write "Signed-off-by" line.
  67
  68But the patch submission requirements are a lot more relaxed
  69here on the technical/contents front, because the core GIT is
  70thousand times smaller ;-).  So here is only the relevant bits.
  71
  72(0) Decide what to base your work on.
  73
  74In general, always base your work on the oldest branch that your
  75change is relevant to.
  76
  77 - A bugfix should be based on 'maint' in general. If the bug is not
  78   present in 'maint', base it on 'master'. For a bug that's not yet
  79   in 'master', find the topic that introduces the regression, and
  80   base your work on the tip of the topic.
  81
  82 - A new feature should be based on 'master' in general. If the new
  83   feature depends on a topic that is in 'pu', but not in 'master',
  84   base your work on the tip of that topic.
  85
  86 - Corrections and enhancements to a topic not yet in 'master' should
  87   be based on the tip of that topic. If the topic has not been merged
  88   to 'next', it's alright to add a note to squash minor corrections
  89   into the series.
  90
  91 - In the exceptional case that a new feature depends on several topics
  92   not in 'master', start working on 'next' or 'pu' privately and send
  93   out patches for discussion. Before the final merge, you may have to
  94   wait until some of the dependent topics graduate to 'master', and
  95   rebase your work.
  96
  97To find the tip of a topic branch, run "git log --first-parent
  98master..pu" and look for the merge commit. The second parent of this
  99commit is the tip of the topic branch.
 100
 101(1) Make separate commits for logically separate changes.
 102
 103Unless your patch is really trivial, you should not be sending
 104out a patch that was generated between your working tree and
 105your commit head.  Instead, always make a commit with complete
 106commit message and generate a series of patches from your
 107repository.  It is a good discipline.
 108
 109Give an explanation for the change(s) that is detailed enough so
 110that people can judge if it is good thing to do, without reading
 111the actual patch text to determine how well the code does what
 112the explanation promises to do.
 113
 114If your description starts to get too long, that's a sign that you
 115probably need to split up your commit to finer grained pieces.
 116That being said, patches which plainly describe the things that
 117help reviewers check the patch, and future maintainers understand
 118the code, are the most beautiful patches.  Descriptions that summarise
 119the point in the subject well, and describe the motivation for the
 120change, the approach taken by the change, and if relevant how this
 121differs substantially from the prior version, are all good things
 122to have.
 123
 124Oh, another thing.  I am picky about whitespaces.  Make sure your
 125changes do not trigger errors with the sample pre-commit hook shipped
 126in templates/hooks--pre-commit.  To help ensure this does not happen,
 127run git diff --check on your changes before you commit.
 128
 129
 130(2) Generate your patch using git tools out of your commits.
 131
 132git based diff tools generate unidiff which is the preferred format.
 133
 134You do not have to be afraid to use -M option to "git diff" or
 135"git format-patch", if your patch involves file renames.  The
 136receiving end can handle them just fine.
 137
 138Please make sure your patch does not include any extra files
 139which do not belong in a patch submission.  Make sure to review
 140your patch after generating it, to ensure accuracy.  Before
 141sending out, please make sure it cleanly applies to the "master"
 142branch head.  If you are preparing a work based on "next" branch,
 143that is fine, but please mark it as such.
 144
 145
 146(3) Sending your patches.
 147
 148People on the git mailing list need to be able to read and
 149comment on the changes you are submitting.  It is important for
 150a developer to be able to "quote" your changes, using standard
 151e-mail tools, so that they may comment on specific portions of
 152your code.  For this reason, all patches should be submitted
 153"inline".  WARNING: Be wary of your MUAs word-wrap
 154corrupting your patch.  Do not cut-n-paste your patch; you can
 155lose tabs that way if you are not careful.
 156
 157It is a common convention to prefix your subject line with
 158[PATCH].  This lets people easily distinguish patches from other
 159e-mail discussions.  Use of additional markers after PATCH and
 160the closing bracket to mark the nature of the patch is also
 161encouraged.  E.g. [PATCH/RFC] is often used when the patch is
 162not ready to be applied but it is for discussion, [PATCH v2],
 163[PATCH v3] etc. are often seen when you are sending an update to
 164what you have previously sent.
 165
 166"git format-patch" command follows the best current practice to
 167format the body of an e-mail message.  At the beginning of the
 168patch should come your commit message, ending with the
 169Signed-off-by: lines, and a line that consists of three dashes,
 170followed by the diffstat information and the patch itself.  If
 171you are forwarding a patch from somebody else, optionally, at
 172the beginning of the e-mail message just before the commit
 173message starts, you can put a "From: " line to name that person.
 174
 175You often want to add additional explanation about the patch,
 176other than the commit message itself.  Place such "cover letter"
 177material between the three dash lines and the diffstat. Git-notes
 178can also be inserted using the `--notes` option.
 179
 180Do not attach the patch as a MIME attachment, compressed or not.
 181Do not let your e-mail client send quoted-printable.  Do not let
 182your e-mail client send format=flowed which would destroy
 183whitespaces in your patches. Many
 184popular e-mail applications will not always transmit a MIME
 185attachment as plain text, making it impossible to comment on
 186your code.  A MIME attachment also takes a bit more time to
 187process.  This does not decrease the likelihood of your
 188MIME-attached change being accepted, but it makes it more likely
 189that it will be postponed.
 190
 191Exception:  If your mailer is mangling patches then someone may ask
 192you to re-send them using MIME, that is OK.
 193
 194Do not PGP sign your patch, at least for now.  Most likely, your
 195maintainer or other people on the list would not have your PGP
 196key and would not bother obtaining it anyway.  Your patch is not
 197judged by who you are; a good patch from an unknown origin has a
 198far better chance of being accepted than a patch from a known,
 199respected origin that is done poorly or does incorrect things.
 200
 201If you really really really really want to do a PGP signed
 202patch, format it as "multipart/signed", not a text/plain message
 203that starts with '-----BEGIN PGP SIGNED MESSAGE-----'.  That is
 204not a text/plain, it's something else.
 205
 206Unless your patch is a very trivial and an obviously correct one,
 207first send it with "To:" set to the mailing list, with "cc:" listing
 208people who are involved in the area you are touching (the output from
 209"git blame $path" and "git shortlog --no-merges $path" would help to
 210identify them), to solicit comments and reviews.  After the list
 211reached a consensus that it is a good idea to apply the patch, re-send
 212it with "To:" set to the maintainer and optionally "cc:" the list for
 213inclusion.  Do not forget to add trailers such as "Acked-by:",
 214"Reviewed-by:" and "Tested-by:" after your "Signed-off-by:" line as
 215necessary.
 216
 217
 218(4) Sign your work
 219
 220To improve tracking of who did what, we've borrowed the
 221"sign-off" procedure from the Linux kernel project on patches
 222that are being emailed around.  Although core GIT is a lot
 223smaller project it is a good discipline to follow it.
 224
 225The sign-off is a simple line at the end of the explanation for
 226the patch, which certifies that you wrote it or otherwise have
 227the right to pass it on as a open-source patch.  The rules are
 228pretty simple: if you can certify the below:
 229
 230        Developer's Certificate of Origin 1.1
 231
 232        By making a contribution to this project, I certify that:
 233
 234        (a) The contribution was created in whole or in part by me and I
 235            have the right to submit it under the open source license
 236            indicated in the file; or
 237
 238        (b) The contribution is based upon previous work that, to the best
 239            of my knowledge, is covered under an appropriate open source
 240            license and I have the right under that license to submit that
 241            work with modifications, whether created in whole or in part
 242            by me, under the same open source license (unless I am
 243            permitted to submit under a different license), as indicated
 244            in the file; or
 245
 246        (c) The contribution was provided directly to me by some other
 247            person who certified (a), (b) or (c) and I have not modified
 248            it.
 249
 250        (d) I understand and agree that this project and the contribution
 251            are public and that a record of the contribution (including all
 252            personal information I submit with it, including my sign-off) is
 253            maintained indefinitely and may be redistributed consistent with
 254            this project or the open source license(s) involved.
 255
 256then you just add a line saying
 257
 258        Signed-off-by: Random J Developer <random@developer.example.org>
 259
 260This line can be automatically added by git if you run the git-commit
 261command with the -s option.
 262
 263Notice that you can place your own Signed-off-by: line when
 264forwarding somebody else's patch with the above rules for
 265D-C-O.  Indeed you are encouraged to do so.  Do not forget to
 266place an in-body "From: " line at the beginning to properly attribute
 267the change to its true author (see (2) above).
 268
 269Also notice that a real name is used in the Signed-off-by: line. Please
 270don't hide your real name.
 271
 272If you like, you can put extra tags at the end:
 273
 2741. "Reported-by:" is used to credit someone who found the bug that
 275   the patch attempts to fix.
 2762. "Acked-by:" says that the person who is more familiar with the area
 277   the patch attempts to modify liked the patch.
 2783. "Reviewed-by:", unlike the other tags, can only be offered by the
 279   reviewer and means that she is completely satisfied that the patch
 280   is ready for application.  It is usually offered only after a
 281   detailed review.
 2824. "Tested-by:" is used to indicate that the person applied the patch
 283   and found it to have the desired effect.
 284
 285You can also create your own tag or use one that's in common usage
 286such as "Thanks-to:", "Based-on-patch-by:", or "Mentored-by:".
 287
 288------------------------------------------------
 289An ideal patch flow
 290
 291Here is an ideal patch flow for this project the current maintainer
 292suggests to the contributors:
 293
 294 (0) You come up with an itch.  You code it up.
 295
 296 (1) Send it to the list and cc people who may need to know about
 297     the change.
 298
 299     The people who may need to know are the ones whose code you
 300     are butchering.  These people happen to be the ones who are
 301     most likely to be knowledgeable enough to help you, but
 302     they have no obligation to help you (i.e. you ask for help,
 303     don't demand).  "git log -p -- $area_you_are_modifying" would
 304     help you find out who they are.
 305
 306 (2) You get comments and suggestions for improvements.  You may
 307     even get them in a "on top of your change" patch form.
 308
 309 (3) Polish, refine, and re-send to the list and the people who
 310     spend their time to improve your patch.  Go back to step (2).
 311
 312 (4) The list forms consensus that the last round of your patch is
 313     good.  Send it to the list and cc the maintainer.
 314
 315 (5) A topic branch is created with the patch and is merged to 'next',
 316     and cooked further and eventually graduates to 'master'.
 317
 318In any time between the (2)-(3) cycle, the maintainer may pick it up
 319from the list and queue it to 'pu', in order to make it easier for
 320people play with it without having to pick up and apply the patch to
 321their trees themselves.
 322
 323------------------------------------------------
 324Know the status of your patch after submission
 325
 326* You can use Git itself to find out when your patch is merged in
 327  master. 'git pull --rebase' will automatically skip already-applied
 328  patches, and will let you know. This works only if you rebase on top
 329  of the branch in which your patch has been merged (i.e. it will not
 330  tell you if your patch is merged in pu if you rebase on top of
 331  master).
 332
 333* Read the git mailing list, the maintainer regularly posts messages
 334  entitled "What's cooking in git.git" and "What's in git.git" giving
 335  the status of various proposed changes.
 336
 337------------------------------------------------
 338MUA specific hints
 339
 340Some of patches I receive or pick up from the list share common
 341patterns of breakage.  Please make sure your MUA is set up
 342properly not to corrupt whitespaces.
 343
 344See the DISCUSSION section of git-format-patch(1) for hints on
 345checking your patch by mailing it to yourself and applying with
 346git-am(1).
 347
 348While you are at it, check the resulting commit log message from
 349a trial run of applying the patch.  If what is in the resulting
 350commit is not exactly what you would want to see, it is very
 351likely that your maintainer would end up hand editing the log
 352message when he applies your patch.  Things like "Hi, this is my
 353first patch.\n", if you really want to put in the patch e-mail,
 354should come after the three-dash line that signals the end of the
 355commit message.
 356
 357
 358Pine
 359----
 360
 361(Johannes Schindelin)
 362
 363I don't know how many people still use pine, but for those poor
 364souls it may be good to mention that the quell-flowed-text is
 365needed for recent versions.
 366
 367... the "no-strip-whitespace-before-send" option, too. AFAIK it
 368was introduced in 4.60.
 369
 370(Linus Torvalds)
 371
 372And 4.58 needs at least this.
 373
 374---
 375diff-tree 8326dd8350be64ac7fc805f6563a1d61ad10d32c (from e886a61f76edf5410573e92e38ce22974f9c40f1)
 376Author: Linus Torvalds <torvalds@g5.osdl.org>
 377Date:   Mon Aug 15 17:23:51 2005 -0700
 378
 379    Fix pine whitespace-corruption bug
 380
 381    There's no excuse for unconditionally removing whitespace from
 382    the pico buffers on close.
 383
 384diff --git a/pico/pico.c b/pico/pico.c
 385--- a/pico/pico.c
 386+++ b/pico/pico.c
 387@@ -219,7 +219,9 @@ PICO *pm;
 388            switch(pico_all_done){      /* prepare for/handle final events */
 389              case COMP_EXIT :          /* already confirmed */
 390                packheader();
 391+#if 0
 392                stripwhitespace();
 393+#endif
 394                c |= COMP_EXIT;
 395                break;
 396
 397
 398(Daniel Barkalow)
 399
 400> A patch to SubmittingPatches, MUA specific help section for
 401> users of Pine 4.63 would be very much appreciated.
 402
 403Ah, it looks like a recent version changed the default behavior to do the
 404right thing, and inverted the sense of the configuration option. (Either
 405that or Gentoo did it.) So you need to set the
 406"no-strip-whitespace-before-send" option, unless the option you have is
 407"strip-whitespace-before-send", in which case you should avoid checking
 408it.
 409
 410
 411Thunderbird, KMail, GMail
 412-------------------------
 413
 414See the MUA-SPECIFIC HINTS section of git-format-patch(1).
 415
 416Gnus
 417----
 418
 419'|' in the *Summary* buffer can be used to pipe the current
 420message to an external program, and this is a handy way to drive
 421"git am".  However, if the message is MIME encoded, what is
 422piped into the program is the representation you see in your
 423*Article* buffer after unwrapping MIME.  This is often not what
 424you would want for two reasons.  It tends to screw up non ASCII
 425characters (most notably in people's names), and also
 426whitespaces (fatal in patches).  Running 'C-u g' to display the
 427message in raw form before using '|' to run the pipe can work
 428this problem around.