Documentation / SubmittingPatcheson commit revision.c: explain what tree_difference does (0a4ba7f)
   1Checklist (and a short version for the impatient):
   2
   3        - make commits of logical units
   4        - check for unnecessary whitespace with "git diff --check"
   5          before committing
   6        - do not check in commented out code or unneeded files
   7        - provide a meaningful commit message
   8        - the first line of the commit message should be a short
   9          description and should skip the full stop
  10        - if you want your work included in git.git, add a
  11          "Signed-off-by: Your Name <your@email.com>" line to the
  12          commit message (or just use the option "-s" when
  13          committing) to confirm that you agree to the Developer's
  14          Certificate of Origin
  15        - do not PGP sign your patch
  16        - use "git format-patch -M" to create the patch
  17        - do not attach your patch, but read in the mail
  18          body, unless you cannot teach your mailer to
  19          leave the formatting of the patch alone.
  20        - be careful doing cut & paste into your mailer, not to
  21          corrupt whitespaces.
  22        - provide additional information (which is unsuitable for
  23          the commit message) between the "---" and the diffstat
  24        - send the patch to the list _and_ the maintainer
  25
  26Long version:
  27
  28I started reading over the SubmittingPatches document for Linux
  29kernel, primarily because I wanted to have a document similar to
  30it for the core GIT to make sure people understand what they are
  31doing when they write "Signed-off-by" line.
  32
  33But the patch submission requirements are a lot more relaxed
  34here on the technical/contents front, because the core GIT is
  35thousand times smaller ;-).  So here is only the relevant bits.
  36
  37
  38(1) Make separate commits for logically separate changes.
  39
  40Unless your patch is really trivial, you should not be sending
  41out a patch that was generated between your working tree and
  42your commit head.  Instead, always make a commit with complete
  43commit message and generate a series of patches from your
  44repository.  It is a good discipline.
  45
  46Describe the technical detail of the change(s).
  47
  48If your description starts to get too long, that's a sign that you
  49probably need to split up your commit to finer grained pieces.
  50
  51Oh, another thing.  I am picky about whitespaces.  Make sure your
  52changes do not trigger errors with the sample pre-commit hook shipped
  53in templates/hooks--pre-commit.  To help ensure this does not happen,
  54run git diff --check on your changes before you commit.
  55
  56
  57(2) Generate your patch using git tools out of your commits.
  58
  59git based diff tools (git, Cogito, and StGIT included) generate
  60unidiff which is the preferred format.
  61
  62You do not have to be afraid to use -M option to "git diff" or
  63"git format-patch", if your patch involves file renames.  The
  64receiving end can handle them just fine.
  65
  66Please make sure your patch does not include any extra files
  67which do not belong in a patch submission.  Make sure to review
  68your patch after generating it, to ensure accuracy.  Before
  69sending out, please make sure it cleanly applies to the "master"
  70branch head.  If you are preparing a work based on "next" branch,
  71that is fine, but please mark it as such.
  72
  73
  74(3) Sending your patches.
  75
  76People on the git mailing list need to be able to read and
  77comment on the changes you are submitting.  It is important for
  78a developer to be able to "quote" your changes, using standard
  79e-mail tools, so that they may comment on specific portions of
  80your code.  For this reason, all patches should be submitted
  81"inline".  WARNING: Be wary of your MUAs word-wrap
  82corrupting your patch.  Do not cut-n-paste your patch; you can
  83lose tabs that way if you are not careful.
  84
  85It is a common convention to prefix your subject line with
  86[PATCH].  This lets people easily distinguish patches from other
  87e-mail discussions.
  88
  89"git format-patch" command follows the best current practice to
  90format the body of an e-mail message.  At the beginning of the
  91patch should come your commit message, ending with the
  92Signed-off-by: lines, and a line that consists of three dashes,
  93followed by the diffstat information and the patch itself.  If
  94you are forwarding a patch from somebody else, optionally, at
  95the beginning of the e-mail message just before the commit
  96message starts, you can put a "From: " line to name that person.
  97
  98You often want to add additional explanation about the patch,
  99other than the commit message itself.  Place such "cover letter"
 100material between the three dash lines and the diffstat.
 101
 102Do not attach the patch as a MIME attachment, compressed or not.
 103Do not let your e-mail client send quoted-printable.  Do not let
 104your e-mail client send format=flowed which would destroy
 105whitespaces in your patches. Many
 106popular e-mail applications will not always transmit a MIME
 107attachment as plain text, making it impossible to comment on
 108your code.  A MIME attachment also takes a bit more time to
 109process.  This does not decrease the likelihood of your
 110MIME-attached change being accepted, but it makes it more likely
 111that it will be postponed.
 112
 113Exception:  If your mailer is mangling patches then someone may ask
 114you to re-send them using MIME, that is OK.
 115
 116Do not PGP sign your patch, at least for now.  Most likely, your
 117maintainer or other people on the list would not have your PGP
 118key and would not bother obtaining it anyway.  Your patch is not
 119judged by who you are; a good patch from an unknown origin has a
 120far better chance of being accepted than a patch from a known,
 121respected origin that is done poorly or does incorrect things.
 122
 123If you really really really really want to do a PGP signed
 124patch, format it as "multipart/signed", not a text/plain message
 125that starts with '-----BEGIN PGP SIGNED MESSAGE-----'.  That is
 126not a text/plain, it's something else.
 127
 128Note that your maintainer does not necessarily read everything
 129on the git mailing list.  If your patch is for discussion first,
 130send it "To:" the mailing list, and optionally "cc:" him.  If it
 131is trivially correct or after the list reached a consensus, send
 132it "To:" the maintainer and optionally "cc:" the list.
 133
 134Also note that your maintainer does not actively involve himself in
 135maintaining what are in contrib/ hierarchy.  When you send fixes and
 136enhancements to them, do not forget to "cc: " the person who primarily
 137worked on that hierarchy in contrib/.
 138
 139
 140(4) Sign your work
 141
 142To improve tracking of who did what, we've borrowed the
 143"sign-off" procedure from the Linux kernel project on patches
 144that are being emailed around.  Although core GIT is a lot
 145smaller project it is a good discipline to follow it.
 146
 147The sign-off is a simple line at the end of the explanation for
 148the patch, which certifies that you wrote it or otherwise have
 149the right to pass it on as a open-source patch.  The rules are
 150pretty simple: if you can certify the below:
 151
 152        Developer's Certificate of Origin 1.1
 153
 154        By making a contribution to this project, I certify that:
 155
 156        (a) The contribution was created in whole or in part by me and I
 157            have the right to submit it under the open source license
 158            indicated in the file; or
 159
 160        (b) The contribution is based upon previous work that, to the best
 161            of my knowledge, is covered under an appropriate open source
 162            license and I have the right under that license to submit that
 163            work with modifications, whether created in whole or in part
 164            by me, under the same open source license (unless I am
 165            permitted to submit under a different license), as indicated
 166            in the file; or
 167
 168        (c) The contribution was provided directly to me by some other
 169            person who certified (a), (b) or (c) and I have not modified
 170            it.
 171
 172        (d) I understand and agree that this project and the contribution
 173            are public and that a record of the contribution (including all
 174            personal information I submit with it, including my sign-off) is
 175            maintained indefinitely and may be redistributed consistent with
 176            this project or the open source license(s) involved.
 177
 178then you just add a line saying
 179
 180        Signed-off-by: Random J Developer <random@developer.example.org>
 181
 182This line can be automatically added by git if you run the git-commit
 183command with the -s option.
 184
 185Some people also put extra tags at the end.  They'll just be ignored for
 186now, but you can do this to mark internal company procedures or just
 187point out some special detail about the sign-off.
 188
 189
 190------------------------------------------------
 191MUA specific hints
 192
 193Some of patches I receive or pick up from the list share common
 194patterns of breakage.  Please make sure your MUA is set up
 195properly not to corrupt whitespaces.  Here are two common ones
 196I have seen:
 197
 198* Empty context lines that do not have _any_ whitespace.
 199
 200* Non empty context lines that have one extra whitespace at the
 201  beginning.
 202
 203One test you could do yourself if your MUA is set up correctly is:
 204
 205* Send the patch to yourself, exactly the way you would, except
 206  To: and Cc: lines, which would not contain the list and
 207  maintainer address.
 208
 209* Save that patch to a file in UNIX mailbox format.  Call it say
 210  a.patch.
 211
 212* Try to apply to the tip of the "master" branch from the
 213  git.git public repository:
 214
 215    $ git fetch http://kernel.org/pub/scm/git/git.git master:test-apply
 216    $ git checkout test-apply
 217    $ git reset --hard
 218    $ git applymbox a.patch
 219
 220If it does not apply correctly, there can be various reasons.
 221
 222* Your patch itself does not apply cleanly.  That is _bad_ but
 223  does not have much to do with your MUA.  Please rebase the
 224  patch appropriately.
 225
 226* Your MUA corrupted your patch; applymbox would complain that
 227  the patch does not apply.  Look at .dotest/ subdirectory and
 228  see what 'patch' file contains and check for the common
 229  corruption patterns mentioned above.
 230
 231* While you are at it, check what are in 'info' and
 232  'final-commit' files as well.  If what is in 'final-commit' is
 233  not exactly what you would want to see in the commit log
 234  message, it is very likely that your maintainer would end up
 235  hand editing the log message when he applies your patch.
 236  Things like "Hi, this is my first patch.\n", if you really
 237  want to put in the patch e-mail, should come after the
 238  three-dash line that signals the end of the commit message.
 239
 240
 241Pine
 242----
 243
 244(Johannes Schindelin)
 245
 246I don't know how many people still use pine, but for those poor
 247souls it may be good to mention that the quell-flowed-text is
 248needed for recent versions.
 249
 250... the "no-strip-whitespace-before-send" option, too. AFAIK it
 251was introduced in 4.60.
 252
 253(Linus Torvalds)
 254
 255And 4.58 needs at least this.
 256
 257---
 258diff-tree 8326dd8350be64ac7fc805f6563a1d61ad10d32c (from e886a61f76edf5410573e92e38ce22974f9c40f1)
 259Author: Linus Torvalds <torvalds@g5.osdl.org>
 260Date:   Mon Aug 15 17:23:51 2005 -0700
 261
 262    Fix pine whitespace-corruption bug
 263
 264    There's no excuse for unconditionally removing whitespace from
 265    the pico buffers on close.
 266
 267diff --git a/pico/pico.c b/pico/pico.c
 268--- a/pico/pico.c
 269+++ b/pico/pico.c
 270@@ -219,7 +219,9 @@ PICO *pm;
 271            switch(pico_all_done){      /* prepare for/handle final events */
 272              case COMP_EXIT :          /* already confirmed */
 273                packheader();
 274+#if 0
 275                stripwhitespace();
 276+#endif
 277                c |= COMP_EXIT;
 278                break;
 279 
 280
 281(Daniel Barkalow)
 282
 283> A patch to SubmittingPatches, MUA specific help section for
 284> users of Pine 4.63 would be very much appreciated.
 285
 286Ah, it looks like a recent version changed the default behavior to do the
 287right thing, and inverted the sense of the configuration option. (Either
 288that or Gentoo did it.) So you need to set the
 289"no-strip-whitespace-before-send" option, unless the option you have is
 290"strip-whitespace-before-send", in which case you should avoid checking
 291it.
 292
 293
 294Thunderbird
 295-----------
 296
 297(A Large Angry SCM)
 298
 299Here are some hints on how to successfully submit patches inline using
 300Thunderbird.
 301
 302This recipe appears to work with the current [*1*] Thunderbird from Suse.
 303
 304The following Thunderbird extensions are needed:
 305        AboutConfig 0.5
 306                http://aboutconfig.mozdev.org/
 307        External Editor 0.7.2
 308                http://globs.org/articles.php?lng=en&pg=8
 309
 3101) Prepare the patch as a text file using your method of choice.
 311
 3122) Before opening a compose window, use Edit->Account Settings to
 313uncheck the "Compose messages in HTML format" setting in the
 314"Composition & Addressing" panel of the account to be used to send the
 315patch. [*2*]
 316
 3173) In the main Thunderbird window, _before_ you open the compose window
 318for the patch, use Tools->about:config to set the following to the
 319indicated values:
 320        mailnews.send_plaintext_flowed  => false
 321        mailnews.wraplength             => 0
 322
 3234) Open a compose window and click the external editor icon.
 324
 3255) In the external editor window, read in the patch file and exit the
 326editor normally.
 327
 3286) Back in the compose window: Add whatever other text you wish to the
 329message, complete the addressing and subject fields, and press send.
 330
 3317) Optionally, undo the about:config/account settings changes made in
 332steps 2 & 3.
 333
 334
 335[Footnotes]
 336*1* Version 1.0 (20041207) from the MozillaThunderbird-1.0-5 rpm of Suse
 3379.3 professional updates.
 338
 339*2* It may be possible to do this with about:config and the following
 340settings but I haven't tried, yet.
 341        mail.html_compose                       => false
 342        mail.identity.default.compose_html      => false
 343        mail.identity.id?.compose_html          => false
 344
 345
 346Gnus
 347----
 348
 349'|' in the *Summary* buffer can be used to pipe the current
 350message to an external program, and this is a handy way to drive
 351"git am".  However, if the message is MIME encoded, what is
 352piped into the program is the representation you see in your
 353*Article* buffer after unwrapping MIME.  This is often not what
 354you would want for two reasons.  It tends to screw up non ASCII
 355characters (most notably in people's names), and also
 356whitespaces (fatal in patches).  Running 'C-u g' to display the
 357message in raw form before using '|' to run the pipe can work
 358this problem around.
 359
 360
 361KMail
 362-----
 363
 364This should help you to submit patches inline using KMail.
 365
 3661) Prepare the patch as a text file.
 367
 3682) Click on New Mail.
 369
 3703) Go under "Options" in the Composer window and be sure that
 371"Word wrap" is not set.
 372
 3734) Use Message -> Insert file... and insert the patch.
 374
 3755) Back in the compose window: add whatever other text you wish to the
 376message, complete the addressing and subject fields, and press send.