git/git.git
14 years agoMerge branch 'jc/grep'
Junio C Hamano [Mon, 10 Jul 2006 06:49:03 +0000 (23:49 -0700)]
Merge branch 'jc/grep'

* jc/grep:
  git-grep: boolean expression on pattern matching.

14 years agoMerge branch 'jc/mailinfo'
Junio C Hamano [Mon, 10 Jul 2006 06:49:00 +0000 (23:49 -0700)]
Merge branch 'jc/mailinfo'

* jc/mailinfo:
  mailinfo: assume input is latin-1 on the header as we do for the body

14 years agoMerge branch 'ew/diff'
Junio C Hamano [Mon, 10 Jul 2006 06:47:39 +0000 (23:47 -0700)]
Merge branch 'ew/diff'

* ew/diff:
  templates/hooks--update: replace diffstat calls with git diff --stat
  diff: do not use configuration magic at the core-level
  Update diff-options and config documentation.
  diff.c: --no-color to defeat diff.color configuration.
  diff.c: respect diff.renames config option

14 years agoMerge branch 'ew/svn'
Junio C Hamano [Mon, 10 Jul 2006 06:37:19 +0000 (23:37 -0700)]
Merge branch 'ew/svn'

* ew/svn:
  Fix some doubled word typos
  Typofix in Makefile comment.
  Makefile: export NO_SVN_TESTS
  git-svn: migrate out of contrib (follow-up)
  git-svn: migrate out of contrib

14 years agoFix typos involving the word 'commit'
Alp Toker [Sun, 9 Jul 2006 09:36:24 +0000 (10:36 +0100)]
Fix typos involving the word 'commit'

Signed-off-by: Alp Toker <alp@atoker.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoFix some doubled word typos
Alp Toker [Sun, 9 Jul 2006 09:36:23 +0000 (10:36 +0100)]
Fix some doubled word typos

Signed-off-by: Alp Toker <alp@atoker.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoFix some doubled word typos
Alp Toker [Sun, 9 Jul 2006 09:36:23 +0000 (10:36 +0100)]
Fix some doubled word typos

Signed-off-by: Alp Toker <alp@atoker.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years ago"git -p cmd" to page anywhere
Junio C Hamano [Tue, 6 Jun 2006 23:58:40 +0000 (16:58 -0700)]
"git -p cmd" to page anywhere

This allows you to say:

git -p diff v2.6.16-rc5..

and the command pipes the output of any git command to your pager.

[jc: this resurrects a month old RFC patch with improvement
 suggested by Linus to call it --paginate instead of --less.]

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agochange ent to tree in git-diff documentation
Matthias Lederhofer [Sun, 9 Jul 2006 09:59:39 +0000 (11:59 +0200)]
change ent to tree in git-diff documentation

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoTypofix in Makefile comment.
Pavel Roskin [Sun, 9 Jul 2006 09:44:58 +0000 (02:44 -0700)]
Typofix in Makefile comment.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoAssorted typo fixes
Pavel Roskin [Sun, 9 Jul 2006 07:44:30 +0000 (03:44 -0400)]
Assorted typo fixes

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMerge branch 'sf/diff'
Junio C Hamano [Sun, 9 Jul 2006 07:52:36 +0000 (00:52 -0700)]
Merge branch 'sf/diff'

14 years agoDisable color detection during format-patch
Ryan Anderson [Sun, 9 Jul 2006 06:28:21 +0000 (02:28 -0400)]
Disable color detection during format-patch

Signed-off-by: Ryan Anderson <ryan@michonline.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-cvsexportcommit can't handle merge commits correctly
Peter Baumann [Fri, 7 Jul 2006 10:55:41 +0000 (12:55 +0200)]
git-cvsexportcommit can't handle merge commits correctly

git-cvsexportcommit should check if the parent (supplied on the cmdline) to use
for a merge commit is one of the real parents of the merge.

But it errors out if the _first_ parent doesn't match and never checks
the other parents.

Signed-off-by: Peter Baumann <siprbaum@stud.informatik.uni-erlangen.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoUsing 'perl' in *.sh
Michal Rokos [Sat, 8 Jul 2006 15:32:04 +0000 (17:32 +0200)]
Using 'perl' in *.sh

Some GIT's shell script are using bare 'perl' for perl invocation.
Use @@PERL@@ symbol and replace it with PERL_PATH_SQ everywhere.

Signed-off-by: Michal Rokos <michal.rokos@nextsoft.cz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agosed -e '/RE/r rfile/' needs space in 'r rfile'
Michal Rokos [Sat, 8 Jul 2006 15:27:10 +0000 (17:27 +0200)]
sed -e '/RE/r rfile/' needs space in 'r rfile'

Some implementations of sed (like HP-UX one) mandate a space between 'r'
and 'rfile'.

Signed-off-by: Michal Rokos <michal.rokos@nextsoft.cz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoClose the index file between writing and committing
Johannes Schindelin [Sat, 8 Jul 2006 08:56:28 +0000 (10:56 +0200)]
Close the index file between writing and committing

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agotemplates/hooks--update: replace diffstat calls with git diff --stat
Eric Wong [Sat, 8 Jul 2006 08:50:02 +0000 (01:50 -0700)]
templates/hooks--update: replace diffstat calls with git diff --stat

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agodiff: do not use configuration magic at the core-level
Junio C Hamano [Sat, 8 Jul 2006 08:05:16 +0000 (01:05 -0700)]
diff: do not use configuration magic at the core-level

The Porcelainish has become so much usable as the UI that there
is not much reason people should be using the core programs by
hand anymore.  At this point we are better off making the
behaviour of the core programs predictable by keeping them
unaffected by the configuration variables.  Otherwise they will
become very hard to use as reliable building blocks.

For example, "git-commit -a" internally uses git-diff-files to
figure out the set of paths that need to be updated in the
index, and we should never allow diff.renames that happens to be
in the configuration to interfere (or slow down the process).

The UI level configuration such as showing renamed diff and
coloring are still honored by the Porcelainish ("git log" family
and "git diff"), but not by the core anymore.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agocolored diff: diff.color = auto fix
Junio C Hamano [Sat, 8 Jul 2006 00:48:02 +0000 (17:48 -0700)]
colored diff: diff.color = auto fix

Even if the standard output is connected to a tty, do not
colorize the diff if we are talking to a dumb terminal when
diff.color configuration variable is set to "auto".

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMakefile: export NO_SVN_TESTS
Johannes Schindelin [Fri, 7 Jul 2006 11:26:31 +0000 (13:26 +0200)]
Makefile: export NO_SVN_TESTS

Without this patch, it really is not sufficient to define NO_SVN_TESTS
in config.mak or the Makefile.

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoUpdate diff-options and config documentation.
Junio C Hamano [Fri, 7 Jul 2006 12:28:05 +0000 (05:28 -0700)]
Update diff-options and config documentation.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agodiff.c: --no-color to defeat diff.color configuration.
Junio C Hamano [Fri, 7 Jul 2006 12:27:24 +0000 (05:27 -0700)]
diff.c: --no-color to defeat diff.color configuration.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agodiff.c: respect diff.renames config option
Eric Wong [Fri, 7 Jul 2006 11:01:23 +0000 (04:01 -0700)]
diff.c: respect diff.renames config option

diff.renames is mentioned several times in the documentation,
but to my surprise it didn't do anything before this patch.

Also add the --no-renames option to override this from the
command-line.

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agodiff-options: Explain --text and -a
Stephan Feder [Fri, 7 Jul 2006 13:57:09 +0000 (15:57 +0200)]
diff-options: Explain --text and -a

Signed-off-by: Stephan Feder <sf@b-i-t.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoAdd -a and --text to common diff options help
Stephan Feder [Fri, 7 Jul 2006 13:57:08 +0000 (15:57 +0200)]
Add -a and --text to common diff options help

Signed-off-by: Stephan Feder <sf@b-i-t.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoTeach diff -a as shorthand for --text
Stephan Feder [Fri, 7 Jul 2006 13:57:07 +0000 (15:57 +0200)]
Teach diff -a as shorthand for --text

Signed-off-by: Stephan Feder <sf@b-i-t.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoTeach --text option to diff
Stephan Feder [Fri, 7 Jul 2006 10:33:57 +0000 (12:33 +0200)]
Teach --text option to diff

Add new item text to struct diff_options.
If set then do not try to detect binary files.

Signed-off-by: Stephan Feder <sf@b-i-t.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agosend-email: format 2822 datestring ourselves.
Jakub Narebski [Fri, 7 Jul 2006 18:57:55 +0000 (20:57 +0200)]
send-email: format 2822 datestring ourselves.

It is not worth trying to force C locale (and failing) just to
format the 2822 datestring.

This code was borrowed from /usr/bin/822-date (Ian Jackson and
Klee Dienes, both in public domain), per suggestion by Eric Wong.

Signed-off-by: Junio C Hamano <junkio@cox.net>
Acked-by: Jakub Narebski <jnareb@gmail.com>
14 years agoDo not drop data from '\0' until eol in patch output
Stephan Feder [Fri, 7 Jul 2006 10:33:44 +0000 (12:33 +0200)]
Do not drop data from '\0' until eol in patch output

The binary file detection is just a heuristic which can well fail.
Do not produce garbage patches in these cases.

Signed-off-by: Stephan Feder <sf@b-i-t.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agobuiltin-log: respect diff configuration options
Eric Wong [Fri, 7 Jul 2006 10:10:45 +0000 (03:10 -0700)]
builtin-log: respect diff configuration options

The log commands are all capable of generating diffs, so we
should respect those configuration options for diffs here.

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-svn: migrate out of contrib (follow-up)
Junio C Hamano [Fri, 7 Jul 2006 10:17:16 +0000 (03:17 -0700)]
git-svn: migrate out of contrib (follow-up)

Check for SVN::Core so test 910[45] don't fail if the user
doesn't have those installed.

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agorev-parse documentation: talk about range notation.
Junio C Hamano [Fri, 7 Jul 2006 05:37:51 +0000 (22:37 -0700)]
rev-parse documentation: talk about range notation.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoshow-branch: match documentation and usage
Junio C Hamano [Fri, 7 Jul 2006 02:21:47 +0000 (19:21 -0700)]
show-branch: match documentation and usage

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMerge branch 'js/merge-base'
Junio C Hamano [Fri, 7 Jul 2006 02:26:13 +0000 (19:26 -0700)]
Merge branch 'js/merge-base'

14 years agobuiltin-rev-parse.c: constness tightening
Junio C Hamano [Thu, 6 Jul 2006 07:16:35 +0000 (00:16 -0700)]
builtin-rev-parse.c: constness tightening

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-svn: migrate out of contrib
Eric Wong [Thu, 6 Jul 2006 07:14:16 +0000 (00:14 -0700)]
git-svn: migrate out of contrib

Allow NO_SVN_TESTS to be defined to skip git-svn tests.  These
tests are time-consuming due to SVN being slow, and even more so
if SVN Perl libraries are not available.

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMerge branch 'ew/instaweb'
Junio C Hamano [Fri, 7 Jul 2006 00:01:00 +0000 (17:01 -0700)]
Merge branch 'ew/instaweb'

* ew/instaweb:
  instaweb: fix unportable ';' usage in sed
  Makefile: replace ugly and unportable sed invocation
  Add git-instaweb, instantly browse the working repo with gitweb
  gitweb: Declare global variables with "our"
  gitweb: Enable tree (directory) history display
  gitweb: optimize per-file history generation

14 years agocore.compression documentation formatting fix.
Joachim Berdal Haga [Thu, 6 Jul 2006 20:35:54 +0000 (22:35 +0200)]
core.compression documentation formatting fix.

I didn't notice earlier that two colons are required for the
asciidoc entry.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agomailinfo: assume input is latin-1 on the header as we do for the body
Junio C Hamano [Wed, 5 Jul 2006 21:17:49 +0000 (14:17 -0700)]
mailinfo: assume input is latin-1 on the header as we do for the body

When the input mbox does not identify what encoding it is in,
and already have RFC2047 stripped away, we cannot tell what
encoding the header text is in.  For body text, when the message
does not say what charset it is in, we fall back to assume
latin-1 input when converting to utf8.  This should be done
consistently to the header as well.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-reset: complain and exit upon seeing an unknown parameter.
Junio C Hamano [Thu, 6 Jul 2006 02:44:03 +0000 (19:44 -0700)]
git-reset: complain and exit upon seeing an unknown parameter.

The check to use "rev-parse --verify" was defeated by the use of
"--default HEAD".  "git reset --hard bogus-committish" just
defaulted to reset to the HEAD without complaining.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-grep: boolean expression on pattern matching.
Junio C Hamano [Fri, 30 Jun 2006 10:04:05 +0000 (03:04 -0700)]
git-grep: boolean expression on pattern matching.

This extends the behaviour of git-grep when multiple -e options
are given.  So far, we allowed multiple -e to behave just like
regular grep with multiple -e, i.e. the patterns are OR'ed
together.

With this change, you can also have multiple patterns AND'ed
together, or form boolean expressions, like this (the
parentheses are quoted from the shell in this example):

$ git grep -e _PATTERN --and \( -e atom -e token \)

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMerge branch 'lt/gitweb'
Junio C Hamano [Wed, 5 Jul 2006 23:40:15 +0000 (16:40 -0700)]
Merge branch 'lt/gitweb'

* lt/gitweb:
  gitweb: Declare global variables with "our"
  gitweb: Enable tree (directory) history display
  gitweb: optimize per-file history generation

14 years agoMerge branch 'jc/fmt-merge-msg-test'
Junio C Hamano [Wed, 5 Jul 2006 23:36:46 +0000 (16:36 -0700)]
Merge branch 'jc/fmt-merge-msg-test'

* jc/fmt-merge-msg-test:
  t6200: fmt-merge-msg test.

14 years agoMerge branch 'jc/sha1'
Junio C Hamano [Wed, 5 Jul 2006 23:36:25 +0000 (16:36 -0700)]
Merge branch 'jc/sha1'

* jc/sha1:
  A better-scheduled PPC SHA-1 implementation.
  test-sha1: test hashing large buffer
  Makefile: add framework to verify and bench sha1 implementations.

14 years agoMerge branch 'jc/diff-test'
Junio C Hamano [Wed, 5 Jul 2006 23:33:50 +0000 (16:33 -0700)]
Merge branch 'jc/diff-test'

* jc/diff-test:
  t4013: add "diff" UI program tests.

14 years agoMerge branch 'th/diff'
Junio C Hamano [Wed, 5 Jul 2006 23:31:24 +0000 (16:31 -0700)]
Merge branch 'th/diff'

* th/diff:
  builtin-diff: turn recursive on when defaulting to --patch format.
  t4013: note improvements brought by the new output code.
  t4013: add format-patch tests.
  format-patch: fix diff format option implementation
  combine-diff.c: type sanity.
  t4013 test updates for new output code.
  Fix some more diff options changes.
  Fix diff-tree -s
  log --raw: Don't descend into subdirectories by default
  diff-tree: Use ---\n as a message separator
  Print empty line between raw, stat, summary and patch
  t4013: add more tests around -c and --cc
  whatchanged: Default to DIFF_FORMAT_RAW
  Don't xcalloc() struct diffstat_t
  Add msg_sep to diff_options
  DIFF_FORMAT_RAW is not default anymore
  Set default diff output format after parsing command line
  Make --raw option available for all diff commands
  Merge with_raw, with_stat and summary variables to output_format
  t4013: add tests for diff/log family output options.

14 years agoMerge branch 'jc/grepfix'
Junio C Hamano [Wed, 5 Jul 2006 23:25:32 +0000 (16:25 -0700)]
Merge branch 'jc/grepfix'

* jc/grepfix:
  git-grep: use a bit more specific error messages.
  git-grep: fix exit code when we use external grep.
  git-grep: fix parsing of pathspec separator '--'

14 years agoMerge branch 'js/fmt-merge-msg'
Junio C Hamano [Wed, 5 Jul 2006 23:23:46 +0000 (16:23 -0700)]
Merge branch 'js/fmt-merge-msg'

* js/fmt-merge-msg:
  Make git-fmt-merge-msg a builtin

14 years agoFix print-log and diff compatibility with recent vc versions
Ville Skytt\e,Ad\e(B [Tue, 4 Jul 2006 22:35:52 +0000 (01:35 +0300)]
Fix print-log and diff compatibility with recent vc versions

Here's a patch that fixes print-log and diff compatibility with recent
vc versions, such as current GNU Emacs CVS.

Signed-off-by: Ville Skytt\e,Ad\e(B <scop@xemacs.org>
14 years agogit-svn: avoid fetching files outside of the URL we're tracking
Eric Wong [Wed, 5 Jul 2006 12:14:00 +0000 (05:14 -0700)]
git-svn: avoid fetching files outside of the URL we're tracking

Thanks to Santi <sbejar@gmail.com> for the bug report and explanation:
> /path/to/repository/project/file
> /path/to/repository/project-2/file
<...>
> you end up with a project with the following files:
>
> file
> -2/file

Signed-off-by: Eric Wong <normalperson@yhbt.net>
14 years agoRe-fix clear_commit_marks().
Junio C Hamano [Wed, 5 Jul 2006 00:45:22 +0000 (17:45 -0700)]
Re-fix clear_commit_marks().

Fix clear_commit_marks() enough to be usable in
get_merge_bases(), and retire now unused clear_object_marks().

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoImprove git-peek-remote
Linus Torvalds [Tue, 4 Jul 2006 19:29:10 +0000 (12:29 -0700)]
Improve git-peek-remote

This makes git-peek-remote able to basically do everything that
git-ls-remote does (but obviously just for the native protocol, so no
http[s]: or rsync: support).

The default behaviour is the same, but you can now give a mixture of
"--refs", "--tags" and "--heads" flags, where "--refs" forces
git-peek-remote to only show real refs (ie none of the fakey tag lookups,
but also not the special pseudo-refs like HEAD and MERGE_HEAD).

The "--tags" and "--heads" flags respectively limit the output to just
regular tags and heads, of course.

You can still also ask to limit them by name too.

You can combine the flags, so

git peek-remote --refs --tags .

will show all local _true_ tags, without the generated tag lookups
(compare the output without the "--refs" flag).

And "--tags --heads" will show both tags and heads, but will avoid (for
example) any special refs outside of the standard locations.

I'm also planning on adding a "--ignore-local" flag that allows us to ask
it to ignore any refs that we already have in the local tree, but that's
an independent thing.

All this is obviously gearing up to making "git fetch" cheaper.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agorev-list: free commit_list in ... handler
Rene Scharfe [Tue, 4 Jul 2006 19:22:20 +0000 (21:22 +0200)]
rev-list: free commit_list in ... handler

Johannes noticed the missing call to free_commit_list() in the
patch from Santi to add ... support to rev-parse.  Turns out I
forgot it too in rev-list.  This patch is against the next branch
(3b1d06a).

Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-grep: use a bit more specific error messages.
Junio C Hamano [Tue, 4 Jul 2006 09:44:48 +0000 (02:44 -0700)]
git-grep: use a bit more specific error messages.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-grep: fix exit code when we use external grep.
Junio C Hamano [Tue, 4 Jul 2006 09:43:40 +0000 (02:43 -0700)]
git-grep: fix exit code when we use external grep.

Upon hit, we should exit with status 0.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-grep: fix parsing of pathspec separator '--'
Junio C Hamano [Tue, 4 Jul 2006 09:31:50 +0000 (02:31 -0700)]
git-grep: fix parsing of pathspec separator '--'

We used to misparse

git grep -e foo -- '*.sh'

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoTeach rev-parse the ... syntax.
Santi Béjar [Tue, 4 Jul 2006 09:02:22 +0000 (11:02 +0200)]
Teach rev-parse the ... syntax.

[jc: moved the difference code around into its own function.]

Signed-off-by: Santi Béjar <sbejar@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agot8001-annotate: fix a bash-ism in this test
Eric Wong [Tue, 4 Jul 2006 08:04:24 +0000 (01:04 -0700)]
t8001-annotate: fix a bash-ism in this test

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMake git-fmt-merge-msg a builtin
Johannes Schindelin [Mon, 3 Jul 2006 15:18:43 +0000 (17:18 +0200)]
Make git-fmt-merge-msg a builtin

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agot6200: fmt-merge-msg test.
Junio C Hamano [Tue, 4 Jul 2006 01:09:54 +0000 (18:09 -0700)]
t6200: fmt-merge-msg test.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agosend-email: do not barf when Term::ReadLine does not like your terminal
Junio C Hamano [Sun, 2 Jul 2006 23:03:59 +0000 (16:03 -0700)]
send-email: do not barf when Term::ReadLine does not like your terminal

As long as we do not need to readline from the terminal, we
should not barf when starting up the program.  Without this
patch, t9001 test on Cygwin occasionally died with the following
error message:

Unable to get Terminal Size. The TIOCGWINSZ ioctl didn't work. The COLUMNS and LINES environment variables didn't work. The resize program didn't work. at /usr/lib/perl5/vendor_perl/5.8/cygwin/Term/ReadKey.pm line 362.
Compilation failed in require at /usr/lib/perl5/vendor_perl/5.8/Term/ReadLine/Perl.pm line 58.

Acked-by: Ryan Anderson <ryan@michonline.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoboolean: accept yes and no as well
Junio C Hamano [Tue, 4 Jul 2006 01:48:23 +0000 (18:48 -0700)]
boolean: accept yes and no as well

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoannotate: Correct most merge following to annotate correctly.
Ryan Anderson [Tue, 4 Jul 2006 01:30:02 +0000 (21:30 -0400)]
annotate: Correct most merge following to annotate correctly.

There is still a bug involving octopus merges, somewhere, but this gets normal
merges correct, so it's still an improvement over the existing version.

Signed-off-by: Ryan Anderson <ryan@michonline.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoannotate: Support annotation of files on other revisions.
Ryan Anderson [Tue, 4 Jul 2006 01:30:01 +0000 (21:30 -0400)]
annotate: Support annotation of files on other revisions.

This is a bug fix, and cleans up one or two other things spotted during the
course of tracking down the main bug here.

Also, the test-suite is updated to reflect this case.

Signed-off-by: Ryan Anderson <ryan@michonline.com>
(cherry picked from 2f7554b4db3ab2c2d3866b160245c91c9236fc9a commit)
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMake zlib compression level configurable, and change default.
Joachim B Haga [Mon, 3 Jul 2006 20:11:47 +0000 (22:11 +0200)]
Make zlib compression level configurable, and change default.

With the change in default, "git add ." on kernel dir is about
twice as fast as before, with only minimal (0.5%) change in
object size. The speed difference is even more noticeable
when committing large files, which is now up to 8 times faster.

The configurability is through setting core.compression = [-1..9]
which maps to the zlib constants; -1 is the default, 0 is no
compression, and 1..9 are various speed/size tradeoffs, 9
being slowest.

Signed-off-by: Joachim B Haga (cjhaga@fys.uio.no)
Acked-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMerge branch 'master' into js/merge-base
Junio C Hamano [Mon, 3 Jul 2006 10:14:32 +0000 (03:14 -0700)]
Merge branch 'master' into js/merge-base

This is to pull in the object-hash clean-up from the master branch.

14 years agorevert clear-commit-marks for now.
Junio C Hamano [Mon, 3 Jul 2006 10:05:20 +0000 (03:05 -0700)]
revert clear-commit-marks for now.

Earlier change broke "git describe A B" among other things.
Revert it for now, and clean the commits smudged by
get_merge_bases using clear_object_marks() function.  For
complex commit ancestry graph, this is way cheaper as well.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoget_merge_bases: clean up even when there is no common commit.
Junio C Hamano [Mon, 3 Jul 2006 10:02:27 +0000 (03:02 -0700)]
get_merge_bases: clean up even when there is no common commit.

Actually in this case we would have traversed a lot of commits, so cleaning
things up is even more important.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMakefile: tighten git-http-{fetch,push} dependencies
Junio C Hamano [Mon, 3 Jul 2006 07:53:13 +0000 (00:53 -0700)]
Makefile: tighten git-http-{fetch,push} dependencies

Although our "git-%$X:" implicit target had dependency on
$(GITLIBS) which included xdiff/lib.a, git-http-{fetch,push} had
their own building rules and with an obsolete dependency on
$(LIB_FILES).  Update the rules to depend on $(GITLIBS), to make
parallel build work correctly.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoEmpty author may be presented by svn as an empty string or a null value.
Robin Rosenberg [Sun, 2 Jul 2006 22:21:00 +0000 (00:21 +0200)]
Empty author may be presented by svn as an empty string or a null value.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agocommit.c: do not redefine UNINTERESTING bit.
Junio C Hamano [Sun, 2 Jul 2006 18:34:17 +0000 (11:34 -0700)]
commit.c: do not redefine UNINTERESTING bit.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoinstaweb: fix unportable ';' usage in sed
Eric Wong [Sun, 2 Jul 2006 11:56:16 +0000 (04:56 -0700)]
instaweb: fix unportable ';' usage in sed

Hint taken from Johannes.  I've tested this with sed --posix on
my system with GNU sed and it works fine with and also without
it.  Further portability testing/review would be good.

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMakefile: replace ugly and unportable sed invocation
Johannes Schindelin [Sun, 2 Jul 2006 09:31:30 +0000 (11:31 +0200)]
Makefile: replace ugly and unportable sed invocation

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoFold get_merge_bases_clean() into get_merge_bases()
Rene Scharfe [Sun, 2 Jul 2006 09:49:38 +0000 (11:49 +0200)]
Fold get_merge_bases_clean() into get_merge_bases()

Change get_merge_bases() to be able to clean up after itself if
needed by adding a cleanup parameter.

We don't need to save the flags and restore them afterwards anymore;
that was a leftover from before the flags were moved out of the
range used in revision.c.  clear_commit_marks() sets them to zero,
which is enough.

Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Acked-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agorevision.c: fix "dense" under --remove-empty
Linus Torvalds [Sun, 2 Jul 2006 17:55:59 +0000 (10:55 -0700)]
revision.c: fix "dense" under --remove-empty

It had the wrong test for whether a commit was a merge. What it did was to
say that a non-merge has exactly one parent (which sounds almost right),
but the fact is, initial trees have no parent at all, but they're
obviously not merges.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agobuiltin-diff: turn recursive on when defaulting to --patch format.
Junio C Hamano [Sun, 2 Jul 2006 05:15:40 +0000 (22:15 -0700)]
builtin-diff: turn recursive on when defaulting to --patch format.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agot4013: add "diff" UI program tests.
Junio C Hamano [Sun, 2 Jul 2006 05:02:17 +0000 (22:02 -0700)]
t4013: add "diff" UI program tests.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoAdd git-instaweb, instantly browse the working repo with gitweb
Eric Wong [Sat, 1 Jul 2006 22:14:14 +0000 (15:14 -0700)]
Add git-instaweb, instantly browse the working repo with gitweb

I got tired of having to configure gitweb for every repository
I work on.  I sometimes prefer gitweb to standard GUIs like gitk
or gitview; so this lets me automatically configure gitweb to
browse my working repository and also opens my browser to it.

Updates from the original patch:

Added Apache/mod_perl2 compatibility if Dennis Stosberg's gitweb
has been applied, too: <20060621130708.Gcbc6e5c@leonov.stosberg.net>

General cleanups in shell code usage.

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogitweb: Declare global variables with "our"
Dennis Stosberg [Wed, 21 Jun 2006 13:07:08 +0000 (15:07 +0200)]
gitweb: Declare global variables with "our"

Variables declared with "my" in the file scope cannot be accessed from
subroutines with mod_perl.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogitweb: Enable tree (directory) history display
Luben Tuikov [Sat, 1 Jul 2006 02:11:18 +0000 (19:11 -0700)]
gitweb: Enable tree (directory) history display

This patch allows history display of whole trees/directories a la
"git-rev-list HEAD -- <dir or file>".  I find this useful especially
when a project lives in its own subdirectory, as opposed to being all
of the GIT repository (i.e. when a sub-project is merged into a
super-project).

Signed-off-by: Luben Tuikov <ltuikov@yahoo.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogitweb: optimize per-file history generation
Junio C Hamano [Sat, 1 Jul 2006 01:54:32 +0000 (18:54 -0700)]
gitweb: optimize per-file history generation

The rev-list command that is recent enough can filter commits
based on paths they touch, so use it instead of generating the
full list and limiting it by passing it with diff-tree --stdin.

[jc: The patch originally came from Luben Tuikov but the it was
 corrupt, but it was short enough to be applied by hand.  I
 added the --full-history to make the output compatible with the
 original while doing so.]

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogitweb: Declare global variables with "our"
Dennis Stosberg [Wed, 21 Jun 2006 13:07:08 +0000 (15:07 +0200)]
gitweb: Declare global variables with "our"

Variables declared with "my" in the file scope cannot be accessed from
subroutines with mod_perl.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogitweb: Enable tree (directory) history display
Luben Tuikov [Sat, 1 Jul 2006 02:11:18 +0000 (19:11 -0700)]
gitweb: Enable tree (directory) history display

This patch allows history display of whole trees/directories a la
"git-rev-list HEAD -- <dir or file>".  I find this useful especially
when a project lives in its own subdirectory, as opposed to being all
of the GIT repository (i.e. when a sub-project is merged into a
super-project).

Signed-off-by: Luben Tuikov <ltuikov@yahoo.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogitweb: optimize per-file history generation
Junio C Hamano [Sat, 1 Jul 2006 01:54:32 +0000 (18:54 -0700)]
gitweb: optimize per-file history generation

The rev-list command that is recent enough can filter commits
based on paths they touch, so use it instead of generating the
full list and limiting it by passing it with diff-tree --stdin.

[jc: The patch originally came from Luben Tuikov but the it was
 corrupt, but it was short enough to be applied by hand.  I
 added the --full-history to make the output compatible with the
 original while doing so.]

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit object hash cleanups v1.4.1
Linus Torvalds [Fri, 30 Jun 2006 18:20:33 +0000 (11:20 -0700)]
git object hash cleanups

This IMNSHO cleans up the object hashing.

The hash expansion is separated out into a function of its own, the hash
array (and size) names are made more obvious, and the code is generally
made to look a bit more like the object-ref hashing.

It also gets rid of "find_object()" returning an index (or negative
position if no object is found), since that is made redundant by the
simplified object rehashing. The basic operation is now "lookup_object()"
which just returns the object itself.

There's an almost unmeasurable speed increase, but more importantly, I
think the end result is more readable.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agorevision.c: --full-history fix.
Linus Torvalds [Sat, 1 Jul 2006 03:21:59 +0000 (20:21 -0700)]
revision.c: --full-history fix.

With history simplification, we still show merges that are required
to make the history _complete_, i.e. say that you had:

  a
  |
  b
 / \
c   d
|   |

and neither "a" nor "b" actually changed the file, but both "c" and "d"
did: in this case we have to leave "b" around just because otherwise there
would be no way to show the _relationship_, even if "b" itself doesn't
actually change the tree in any way what-so-ever.

It would make sense to make that further simplification if the
"--parents" flag wasn't present.  In that case the user is
literally asking for a list of commits and is not interested in
the relationship between them.

This patch also fixes a real bug.  Without this patch, the
"--parents --full-history" combination (which you'd get if you
do something like

gitk --full-history Makefile

or similar) will actually _drop_ merges where all children are identical.
That's wrong in the --full-history case, because it means that the graph
ends up missing lots of entries.

In the process, this also should make

git-rev-list --full-history Makefile

give just the _true_ list of all commits that changed Makefile (and
properly ignore merges that were identical in one parent), because now
we're not asking for "--parent", so we don't need the unnecessary merge
commits to keep the history together.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMake clear_commit_marks() clean harder
Rene Scharfe [Sat, 1 Jul 2006 23:29:58 +0000 (01:29 +0200)]
Make clear_commit_marks() clean harder

Don't care if objects have been parsed or not and don't stop when we
reach a commit that is already clean -- its parents could be dirty.

Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoAdd '...' operator for revisions
Rene Scharfe [Sat, 1 Jul 2006 23:29:37 +0000 (01:29 +0200)]
Add '...' operator for revisions

'A...B' is a shortcut for 'A B --not $(git-merge-base --all A B)'.
This XOR-like operation is called symmetric difference in set
theory.

The symbol '...' has been chosen because it's rather similar to the
existing '..' operator and the somewhat more natural caret ('^') is
already taken.

Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoAdd get_merge_bases_clean()
Rene Scharfe [Sat, 1 Jul 2006 23:29:26 +0000 (01:29 +0200)]
Add get_merge_bases_clean()

Add get_merge_bases_clean(), a wrapper for get_merge_bases() that cleans
up after doing its work and make get_merge_bases() NOT clean up.
Single-shot programs like git-merge-base can use the dirty and fast
version.

Also move the object flags used in get_merge_bases() out of the range
defined in revision.h.  This fixes the "66ae0c77...ced9456a
89719209...262a6ef7" test of the ... operator which is introduced with
the next patch.

Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoFix errno usage in connect.c
Petr Baudis [Sat, 1 Jul 2006 21:56:26 +0000 (23:56 +0200)]
Fix errno usage in connect.c

errno was used after it could've been modified by a subsequent library call.
Spotted by Morten Welinder.

Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMinor documentation fixup.
Robin Rosenberg [Sun, 2 Jul 2006 00:07:40 +0000 (02:07 +0200)]
Minor documentation fixup.

Signed-off-by: Robin Rosenberg <robin.rosenberg@dewire.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agogit-svn: allow a local target directory to be specified for init
Eric Wong [Sat, 1 Jul 2006 04:42:53 +0000 (21:42 -0700)]
git-svn: allow a local target directory to be specified for init

git-svn init url://to/the/repo local-repo

will create the local-repo dirrectory if doesn't exist yet and
populate it as expected.

Original patch by Luca Barbato, cleaned up and made to work for
the current version of git-svn by me (Eric Wong).

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agodon't load objects needlessly when repacking
Nicolas Pitre [Sat, 1 Jul 2006 02:55:30 +0000 (22:55 -0400)]
don't load objects needlessly when repacking

If no delta is attempted on some objects then it is useless to load them
in memory, neither create any delta index for them.  The best thing to
do is therefore to load and index them only when really needed.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoupload-pack.c: <sys/poll.h> includes <ctype.h> on OpenBSD 3.8
Junio C Hamano [Fri, 30 Jun 2006 18:30:29 +0000 (11:30 -0700)]
upload-pack.c: <sys/poll.h> includes <ctype.h> on OpenBSD 3.8

Merlyn reports that <sys/poll.h> on OpenBSD 3.8 includes <ctype.h>
and having our custom ctype (done in git-compat-util.h which is
included via cache.h) makes upload-pack.c uncompilable.  Try to
work it around by including the system headers first.

Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoAbstract out accesses to object hash array
Linus Torvalds [Fri, 30 Jun 2006 04:38:55 +0000 (21:38 -0700)]
Abstract out accesses to object hash array

There are a few special places where some programs accessed the object
hash array directly, which bothered me because I wanted to play with some
simple re-organizations.

So this patch makes the object hash array data structures all entirely
local to object.c, and the few users who wanted to look at it now get to
use a function to query how many object index entries there can be, and to
actually access the array.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoconsider previous pack undeltified object state only when reusing delta data
Nicolas Pitre [Fri, 30 Jun 2006 03:44:52 +0000 (23:44 -0400)]
consider previous pack undeltified object state only when reusing delta data

Without this there would never be a chance to improve packing for
previously undeltified objects.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoMerge branch 'jc/test-3402'
Junio C Hamano [Fri, 30 Jun 2006 06:47:59 +0000 (23:47 -0700)]
Merge branch 'jc/test-3402'

* jc/test-3402:
  Racy GIT (part #3)

14 years agoDo not try futile object pairs when repacking.
Linus Torvalds [Thu, 29 Jun 2006 21:04:01 +0000 (14:04 -0700)]
Do not try futile object pairs when repacking.

In the repacking window, if both objects we are looking at already came
from the same (old) pack-file, don't bother delta'ing them against each
other.

That means that we'll still always check for better deltas for (and
against!) _unpacked_ objects, but assuming incremental repacks, you'll
avoid the delta creation 99% of the time.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
14 years agoRacy GIT (part #3)
Junio C Hamano [Thu, 29 Jun 2006 21:48:22 +0000 (14:48 -0700)]
Racy GIT (part #3)

Commit 29e4d3635709778bcc808dbad0477efad82f8d7e fixed the
underlying update-index races but git-commit was not careful
enough to preserve the index file timestamp when copying the
index file.  This caused t3402 test to occasionally fail.

Signed-off-by: Junio C Hamano <junkio@cox.net>