git.git
16 years agocompat: introduce on_disk_bytes()
Junio C Hamano [Mon, 18 Aug 2008 19:57:16 +0000 (21:57 +0200)]
compat: introduce on_disk_bytes()

Some platforms do not have st_blocks member in "struct stat"; mingw
already emulates it by rounding it up to closest 512-byte blocks (even
though it could overcount when a file has holes).

The reason to use the member is only to figure out how many kilobytes the
files occupy on-disk, so give a helper function in git-compat-util.h to
compute this value.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
Acked-by: Johannes Sixt <johannes.sixt@telecom.at>
16 years agoStart 1.6.1 cycle
Junio C Hamano [Sun, 17 Aug 2008 22:48:51 +0000 (15:48 -0700)]
Start 1.6.1 cycle

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoStart 1.6.0.X maintenance series
Junio C Hamano [Sun, 17 Aug 2008 22:44:11 +0000 (15:44 -0700)]
Start 1.6.0.X maintenance series

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoGIT 1.6.0 v1.6.0
Junio C Hamano [Sun, 17 Aug 2008 18:42:10 +0000 (11:42 -0700)]
GIT 1.6.0

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge git-gui 0.11.0
Junio C Hamano [Sun, 17 Aug 2008 18:40:56 +0000 (11:40 -0700)]
Merge git-gui 0.11.0

16 years agoMerge branch 'ak/p4'
Junio C Hamano [Sun, 17 Aug 2008 17:53:57 +0000 (10:53 -0700)]
Merge branch 'ak/p4'

* ak/p4:
  Utilise our new p4_read_pipe and p4_write_pipe wrappers
  Add p4 read_pipe and write_pipe wrappers
  Put in the two other configuration elements found in the source
  Put some documentation in about the parameters that have been added
  Move git-p4.syncFromOrigin into a configuration parameters section
  Consistently use 'git-p4' for the configuration entries
  If the user has configured various parameters, use them.
  Switch to using 'p4_build_cmd'
  If we are in verbose mode, output what we are about to run (or return)
  Add a single command that will be used to construct the 'p4' command
  Utilise the new 'p4_system' function.
  Have a command that specifically invokes 'p4' (via system)
  Utilise the new 'p4_read_pipe_lines' command
  Create a specific version of the read_pipe_lines command for p4 invocations

Conflicts:
contrib/fast-import/git-p4

16 years agogit-p4: chdir now properly sets PWD environment variable in msysGit
Robert Blum [Fri, 1 Aug 2008 19:50:03 +0000 (12:50 -0700)]
git-p4: chdir now properly sets PWD environment variable in msysGit

P4 on Windows expects the PWD environment variable to be set to the
current working dir, but os.chdir in python doesn't do so.

Signed-off-by: Robert Blum <rob.blum@gmail.com>
Acked-by: Simon Hausmann <simon@lst.de>
Acked-by: Han-Wen Nienhuys <hanwen@xs4all.nl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoImprove error output of git-rebase
Stephan Beyer [Sun, 17 Aug 2008 04:25:43 +0000 (06:25 +0200)]
Improve error output of git-rebase

"git rebase" without arguments on initial startup showed:

fatal: Needed a single revision
invalid upstream

This patch makes it show the ordinary usage string.

If .git/rebase-merge or .git/rebase-apply/rebasing exists, git-rebase
will die with a message saying that a rebase is in progress and the user
should try --skip/--abort/--continue.

If .git/rebase-apply/applying exists, git-rebase will die with a message
saying that git-am is in progress, regardless how many arguments are
given.

If no arguments are given and .git/rebase-apply/ exists, but neither a
rebasing nor applying file is in that directory, git-rebase dies with a
message saying that rebase-apply exists and no arguments were given.

Signed-off-by: Stephan Beyer <s-beyer@gmx.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agot9300: replace '!' with test_must_fail
Miklos Vajna [Sat, 16 Aug 2008 15:17:42 +0000 (17:17 +0200)]
t9300: replace '!' with test_must_fail

Signed-off-by: Miklos Vajna <vmiklos@frugalware.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge git://git.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Sun, 17 Aug 2008 06:21:07 +0000 (23:21 -0700)]
Merge git://git.kernel.org/pub/scm/gitk/gitk

* git://git.kernel.org/pub/scm/gitk/gitk:
  gitk: Allow safely calling nukefile from a run queue handler

16 years agoGit.pm: Make File::Spec and File::Temp requirement lazy
Marcus Griep [Fri, 15 Aug 2008 19:53:59 +0000 (15:53 -0400)]
Git.pm: Make File::Spec and File::Temp requirement lazy

This will ensure that the API at large is accessible to nearly
all Perl versions, while only the temp file caching API is tied to
the File::Temp and File::Spec modules being available.

Signed-off-by: Marcus Griep <marcus@griep.us>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation: document the pager.* configuration setting
Miklos Vajna [Sat, 16 Aug 2008 02:14:33 +0000 (04:14 +0200)]
Documentation: document the pager.* configuration setting

It was already documented in RelNotes-1.6.0, but not in the git-config
manual page.

Signed-off-by: Miklos Vajna <vmiklos@frugalware.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-stash: improve synopsis in help and manual page
Stephan Beyer [Sat, 16 Aug 2008 03:27:31 +0000 (05:27 +0200)]
git-stash: improve synopsis in help and manual page

"git stash -h" showed some incomplete and ugly usage information.
For example, the useful "--keep-index" option for "save" or the "--index"
option for  "apply" were not shown. Also in the documentation synopsis they
were not shown, so that there is no incentive to scroll down and even see
that such options exist.

This patch improves the git-stash synopsis in the documentation by
mentioning that further options to the stash commands and then copies
this synopsis to the usage information string of git-stash.sh.

For the latter, the dashless git command string has to be inserted on the
second and the following usage lines. The code of this is taken from
git-sh-setup so that all lines will show the command string.

Note that the "create" command is not advertised at all now, because
it was not mentioned in git-stash.txt.

Signed-off-by: Stephan Beyer <s-beyer@gmx.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMakefile: building git in cygwin 1.7.0
Eric Blake [Fri, 15 Aug 2008 15:01:03 +0000 (15:01 +0000)]
Makefile: building git in cygwin 1.7.0

On platforms with $X, make removes any leftover scripts 'a' from
earlier builds if a new binary 'a.exe' is now built.  However, on
cygwin 1.7.0, 'git' and 'git.exe' now consistently name the same file.
Test for file equality before attempting a remove, in order to avoid
nuking just-built binaries.

Signed-off-by: Eric Blake <ebb9@byu.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-am: ignore --binary option
Stephan Beyer [Fri, 8 Aug 2008 23:28:54 +0000 (01:28 +0200)]
git-am: ignore --binary option

The git-apply documentation says that --binary is a historical option.
This patch lets git-am ignore --binary and removes advertisements of this
option.

Signed-off-by: Stephan Beyer <s-beyer@gmx.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobash-completion: Add non-command git help files to bash-completion
Marcus Griep [Fri, 15 Aug 2008 17:59:28 +0000 (13:59 -0400)]
bash-completion: Add non-command git help files to bash-completion

Git allows access to the gitattributes man page via `git help attributes`,
but this is not discoverable via the bash-completion mechanism.  This
patch adds all current non-command man pages to the completion candidate
list.

Signed-off-by: Marcus Griep <marcus@griep.us>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoFix t3700 on filesystems which do not support question marks in names
Alex Riesen [Fri, 15 Aug 2008 07:32:30 +0000 (09:32 +0200)]
Fix t3700 on filesystems which do not support question marks in names

Use square brackets instead.

And the prominent example of the deficiency are, as usual, the filesystems
of Microsoft house.

Signed-off-by: Alex Riesen <raa.lkml@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoUtilise our new p4_read_pipe and p4_write_pipe wrappers
Anand Kumria [Thu, 14 Aug 2008 22:40:39 +0000 (23:40 +0100)]
Utilise our new p4_read_pipe and p4_write_pipe wrappers

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoAdd p4 read_pipe and write_pipe wrappers
Anand Kumria [Thu, 14 Aug 2008 22:40:38 +0000 (23:40 +0100)]
Add p4 read_pipe and write_pipe wrappers

Two additional wrappers to cover 3 places where we utilise p4 in piped
form.  Found by Tor Arvid Lund.

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobash completion: Add '--merge' long option for 'git log'
Lee Marlow [Thu, 14 Aug 2008 22:41:11 +0000 (16:41 -0600)]
bash completion: Add '--merge' long option for 'git log'

Signed-off-by: Lee Marlow <lee.marlow@gmail.com>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobash completion: Add completion for 'git mergetool'
Lee Marlow [Thu, 14 Aug 2008 22:41:10 +0000 (16:41 -0600)]
bash completion: Add completion for 'git mergetool'

The --tool= long option to "git mergetool" can be completed with:

kdiff3 tkdiff meld xxdiff emerge
vimdiff gvimdiff ecmerge opendiff

Signed-off-by: Lee Marlow <lee.marlow@gmail.com>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit format-patch documentation: clarify what --cover-letter does
Matt McCutchen [Thu, 14 Aug 2008 17:37:41 +0000 (13:37 -0400)]
git format-patch documentation: clarify what --cover-letter does

Signed-off-by: Matt McCutchen <matt@mattmccutchen.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobash completion: 'git apply' should use 'fix' not 'strip'
Eric Raible [Thu, 14 Aug 2008 17:12:54 +0000 (10:12 -0700)]
bash completion: 'git apply' should use 'fix' not 'strip'

Bring completion up to date with the man page.

Signed-off-by: Eric Raible <raible@gmail.com>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint'
Junio C Hamano [Thu, 14 Aug 2008 02:22:29 +0000 (19:22 -0700)]
Merge branch 'maint'

* maint:
  t5304-prune: adjust file mtime based on system time rather than file mtime
  Fix escaping of glob special characters in pathspecs

16 years agot5304-prune: adjust file mtime based on system time rather than file mtime
Brandon Casey [Thu, 14 Aug 2008 00:49:30 +0000 (19:49 -0500)]
t5304-prune: adjust file mtime based on system time rather than file mtime

test-chmtime can adjust the mtime of a file based on the file's mtime, or
based on the system time. For files accessed over NFS, the file's mtime is
set by the NFS server, and as such may vary a great deal from the NFS
client's system time if the clocks of the client and server are out of
sync. Since these tests are testing the expire feature of git-prune, an
incorrect mtime could cause a file to be expired or not expired incorrectly
and produce a test failure.

Avoid this NFS pitfall by modifying the calls to test-chmtime so that the
mtime is adjusted based on the system time, rather than the file's mtime.

Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agotest-parse-options: use appropriate cast in length_callback
Brandon Casey [Thu, 14 Aug 2008 00:48:57 +0000 (19:48 -0500)]
test-parse-options: use appropriate cast in length_callback

OPT_CALLBACK() is passed &integer which is now an "int" rather than
"unsigned long". Update the length_callback function.

Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoFix escaping of glob special characters in pathspecs
Kevin Ballard [Wed, 13 Aug 2008 22:34:34 +0000 (15:34 -0700)]
Fix escaping of glob special characters in pathspecs

match_one implements an optimized pathspec match where it only uses
fnmatch if it detects glob special characters in the pattern. Unfortunately
it didn't treat \ as a special character, so attempts to escape a glob
special character would fail even though fnmatch() supports it.

Signed-off-by: Kevin Ballard <kevin@sb.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agorebase -i -p: fix parent rewriting
Thomas Rast [Wed, 13 Aug 2008 21:41:24 +0000 (23:41 +0200)]
rebase -i -p: fix parent rewriting

The existing parent rewriting did not handle the case where a previous
commit was amended (via edit or squash).  Fix by always putting the
new sha1 of the last commit into the $REWRITTEN map.

Signed-off-by: Thomas Rast <trast@student.ethz.ch>
16 years agorebase -i -p: handle index and workdir correctly
Thomas Rast [Wed, 13 Aug 2008 21:41:23 +0000 (23:41 +0200)]
rebase -i -p: handle index and workdir correctly

'git rebase -i -p' forgot to update the index and working directory
during fast forwards.  Fix this.  Makes 'GIT_EDITOR=true rebase -i -p
<ancestor>' a no-op again.

Also, it attempted to do a fast forward even if it was instructed not
to commit (via -n).  Fall back to the cherry-pick code path and let
that handle the issue for us.

Signed-off-by: Thomas Rast <trast@student.ethz.ch>
16 years agoGIT 1.6.0-rc3 v1.6.0-rc3
Junio C Hamano [Wed, 13 Aug 2008 04:42:22 +0000 (21:42 -0700)]
GIT 1.6.0-rc3

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint'
Junio C Hamano [Wed, 13 Aug 2008 05:46:22 +0000 (22:46 -0700)]
Merge branch 'maint'

* maint:
  Do not talk about "diff" in rev-list documentation.

16 years agoDo not talk about "diff" in rev-list documentation.
Junio C Hamano [Mon, 11 Aug 2008 18:46:56 +0000 (11:46 -0700)]
Do not talk about "diff" in rev-list documentation.

Since 8c02eee (git-rev-list(1): group options; reformat; document more
options, 2006-09-01), git-rev-list documentation talks as if it supports
any kind of diff output.  It doesn't.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge git://git.bogomips.org/git-svn
Junio C Hamano [Wed, 13 Aug 2008 04:41:29 +0000 (21:41 -0700)]
Merge git://git.bogomips.org/git-svn

* git://git.bogomips.org/git-svn:
  git-svn: Reduce temp file usage when dealing with non-links
  git-svn: Make it incrementally faster by minimizing temp files
  Git.pm: Add faculties to allow temp files to be cached

16 years agogit-svn: Reduce temp file usage when dealing with non-links
Marcus Griep [Tue, 12 Aug 2008 16:45:39 +0000 (12:45 -0400)]
git-svn: Reduce temp file usage when dealing with non-links

Currently, in sub 'close_file', git-svn creates a temporary file and
copies the contents of the blob to be written into it. This is useful
for symlinks because svn stores symlinks in the form:

link $FILE_PATH

Git creates a blob only out of '$FILE_PATH' and uses file mode to
indicate that the blob should be interpreted as a symlink.

As git-hash-object is invoked with --stdin-paths, a duplicate of the
link from svn must be created that leaves off the first five bytes,
i.e. 'link '. However, this is wholly unnecessary for normal blobs,
though, as we already have a temp file with their contents. Copying
the entire file gains nothing, and effectively requires a file to be
written twice before making it into the object db.

This patch corrects that issue, holding onto the substr-like
duplication for symlinks, but skipping it altogether for normal blobs
by reusing the existing temp file.

Signed-off-by: Marcus Griep <marcus@griep.us>
Acked-by: Eric Wong <normalperson@yhbt.net>
16 years agogit-svn: Make it incrementally faster by minimizing temp files
Marcus Griep [Tue, 12 Aug 2008 16:00:53 +0000 (12:00 -0400)]
git-svn: Make it incrementally faster by minimizing temp files

Currently, git-svn would create a temp file on four occasions:
1. Reading a blob out of the object db
2. Creating a delta from svn
3. Hashing and writing a blob into the object db
4. Reading a blob out of the object db (in another place in code)

Any time git-svn did the above, it would dutifully create and then
delete said temp file.  Unfortunately, this means that between 2-4
temporary files are created/deleted per file 'add/modify'-ed in
svn (O(n)).  This causes significant overhead and helps the inode
counter to spin beautifully.

By its nature, git-svn is a serial beast.  Thus, reusing a temp file
does not pose significant problems.  "truncate and seek" takes much
less time than "unlink and create".  This patch centralizes the
tempfile creation and holds onto the tempfile until they are deleted
on exit.  This significantly reduces file overhead, now requiring
at most three (3) temp files per run (O(1)).

Signed-off-by: Marcus Griep <marcus@griep.us>
Acked-by: Eric Wong <normalperson@yhbt.net>
16 years agoGit.pm: Add faculties to allow temp files to be cached
Marcus Griep [Tue, 12 Aug 2008 16:00:18 +0000 (12:00 -0400)]
Git.pm: Add faculties to allow temp files to be cached

This patch offers a generic interface to allow temp files to be
cached while using an instance of the 'Git' package. If many
temp files are created and destroyed during the execution of a
program, this caching mechanism can help reduce the amount of
files created and destroyed by the filesystem.

The temp_acquire method provides a weak guarantee that a temp
file will not be stolen by subsequent requests. If a file is
locked when another acquire request is made, a simple error is
thrown.

Signed-off-by: Marcus Griep <marcus@griep.us>
Acked-by: Eric Wong <normalperson@yhbt.net>
16 years agoDocumentation: rev-list-options: Rewrite simplification descriptions for clarity
Thomas Rast [Mon, 11 Aug 2008 23:55:36 +0000 (01:55 +0200)]
Documentation: rev-list-options: Rewrite simplification descriptions for clarity

This completely rewrites the documentation of --full-history with lots
of examples.

Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoTeach git diff about BibTeX head hunk patterns
Gustaf Hendeby [Tue, 12 Aug 2008 14:24:26 +0000 (16:24 +0200)]
Teach git diff about BibTeX head hunk patterns

All BibTeX entries starts with an @ followed by an entry type.  Since
there are many entry types and own can be defined, the pattern matches
legal entry type names instead of just the default types (which would
be a long list).  The pattern also matches strings and comments since
they will also be useful to position oneself in a bib-file.

Signed-off-by: Gustaf Hendeby <hendeby@isy.liu.se>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogitattributes: Document built in hunk header patterns
Gustaf Hendeby [Tue, 12 Aug 2008 14:24:25 +0000 (16:24 +0200)]
gitattributes: Document built in hunk header patterns

Since the hunk header pattern text was written patterns for Ruby and
Pascal/Delphi have been added.  For users to be able to find them they
should be documented not only in code.

Signed-off-by: Gustaf Hendeby <hendeby@isy.liu.se>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-daemon: SysV needs the signal handler reinstated.
Stephen R. van den Berg [Tue, 12 Aug 2008 19:36:13 +0000 (21:36 +0200)]
git-daemon: SysV needs the signal handler reinstated.

Fixes the bug on (amongst others) Solaris that only the first
child ever is reaped.

Signed-off-by: Stephen R. van den Berg <srb@cuci.nl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agodiff --check: do not unconditionally complain about trailing empty lines
Junio C Hamano [Tue, 12 Aug 2008 05:15:28 +0000 (22:15 -0700)]
diff --check: do not unconditionally complain about trailing empty lines

Recently "git diff --check" learned to detect new trailing blank lines
just like "git apply --whitespace" does.  However this check should not
trigger unconditionally.  This patch makes it honor the whitespace
settings from core.whitespace and gitattributes.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint'
Junio C Hamano [Tue, 12 Aug 2008 02:24:28 +0000 (19:24 -0700)]
Merge branch 'maint'

* maint:
  git-bisect: fix wrong usage of read(1)

16 years agoPut in the two other configuration elements found in the source
Anand Kumria [Sun, 10 Aug 2008 18:26:35 +0000 (19:26 +0100)]
Put in the two other configuration elements found in the source

I am not entirely clear what these parameters do but felt it
 useful to call them out in the documentation.

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoPut some documentation in about the parameters that have been added
Anand Kumria [Sun, 10 Aug 2008 18:26:34 +0000 (19:26 +0100)]
Put some documentation in about the parameters that have been added

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMove git-p4.syncFromOrigin into a configuration parameters section
Anand Kumria [Sun, 10 Aug 2008 18:26:33 +0000 (19:26 +0100)]
Move git-p4.syncFromOrigin into a configuration parameters section

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoConsistently use 'git-p4' for the configuration entries
Anand Kumria [Sun, 10 Aug 2008 18:26:32 +0000 (19:26 +0100)]
Consistently use 'git-p4' for the configuration entries

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoIf the user has configured various parameters, use them.
Anand Kumria [Sun, 10 Aug 2008 18:26:31 +0000 (19:26 +0100)]
If the user has configured various parameters, use them.

Some repositories require authentication and access to certain
 hosts. Allow git-p4 to pull this information from the configuration

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoSwitch to using 'p4_build_cmd'
Anand Kumria [Sun, 10 Aug 2008 18:26:30 +0000 (19:26 +0100)]
Switch to using 'p4_build_cmd'

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoIf we are in verbose mode, output what we are about to run (or return)
Anand Kumria [Sun, 10 Aug 2008 18:26:29 +0000 (19:26 +0100)]
If we are in verbose mode, output what we are about to run (or return)

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoAdd a single command that will be used to construct the 'p4' command
Anand Kumria [Sun, 10 Aug 2008 18:26:28 +0000 (19:26 +0100)]
Add a single command that will be used to construct the 'p4' command

Rather than having three locations where the 'p4' command is built up,
 refactor this into the one place. This will, eventually, allow us to
 have one place where we modify the evironment or pass extra
 command-line options to the 'p4' binary.

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoUtilise the new 'p4_system' function.
Anand Kumria [Sun, 10 Aug 2008 18:26:27 +0000 (19:26 +0100)]
Utilise the new 'p4_system' function.

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoHave a command that specifically invokes 'p4' (via system)
Anand Kumria [Sun, 10 Aug 2008 18:26:26 +0000 (19:26 +0100)]
Have a command that specifically invokes 'p4' (via system)

Similiar to our 'p4_read_pipe_lines' command, we can isolate
 specific changes to the invocation method in the one location
 with this change.

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoUtilise the new 'p4_read_pipe_lines' command
Anand Kumria [Sun, 10 Aug 2008 18:26:25 +0000 (19:26 +0100)]
Utilise the new 'p4_read_pipe_lines' command

Now that we have the new command, we can utilise it and then
 eventually, isolate any changes required to the one place.

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoCreate a specific version of the read_pipe_lines command for p4 invocations
Anand Kumria [Sun, 10 Aug 2008 18:26:24 +0000 (19:26 +0100)]
Create a specific version of the read_pipe_lines command for p4 invocations

This will make it easier to isolate changes to how 'p4' is invoked
 (whether with parameters or not, etc.).

Signed-off-by: Anand Kumria <wildfire@progsoc.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation: rev-list-options: Fix a typo
Thomas Rast [Mon, 11 Aug 2008 23:55:35 +0000 (01:55 +0200)]
Documentation: rev-list-options: Fix a typo

Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoAdjust for the new way of enabling the default post-update hook
Petr Baudis [Mon, 11 Aug 2008 22:34:46 +0000 (00:34 +0200)]
Adjust for the new way of enabling the default post-update hook

The post-update hook, which is required to be enabled in order for
the repository to be accessible over HTTP, is not enabled by
chmod a+x anymore, but instead by dropping the .sample suffix.

This patch emphasizes this change in the release notes (since
I believe this is rather noticeable backwards-incompatible change).
It also adjusts the documentation which still described the old way
and fixes t/t5540-http-push.sh, which was broken for 1.5 month
but apparently noone ever runs this test.

Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-bisect: fix wrong usage of read(1)
Francis Moreau [Mon, 11 Aug 2008 17:37:46 +0000 (19:37 +0200)]
git-bisect: fix wrong usage of read(1)

Signed-off-by: Francis Moreau <francis.moro@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoFix typo in comments of longest_ancestor_length()
Nguyễn Thái Ngọc Duy [Sun, 10 Aug 2008 15:26:23 +0000 (22:26 +0700)]
Fix typo in comments of longest_ancestor_length()

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint'
Junio C Hamano [Mon, 11 Aug 2008 21:28:35 +0000 (14:28 -0700)]
Merge branch 'maint'

* maint:
  Re-fix rev-list-options documentation

16 years agoRe-fix rev-list-options documentation
Junio C Hamano [Mon, 11 Aug 2008 21:24:51 +0000 (14:24 -0700)]
Re-fix rev-list-options documentation

18a2197 (Documentation: rev-list-options: Fix -g paragraph formatting,
2008-08-10) introduced the third paragraph that is continued, but it seems
to confuse docbook toolchain on FC9 machines.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint'
Junio C Hamano [Mon, 11 Aug 2008 07:53:31 +0000 (00:53 -0700)]
Merge branch 'maint'

* maint:
  Documentation: fix invalid reference to 'mybranch' in user manual
  Fix deleting reflog entries from HEAD reflog
  reflog test: add more tests for 'reflog delete'
  Documentation: rev-list-options: Fix -g paragraph formatting

Conflicts:
Documentation/user-manual.txt

16 years agoDocumentation: fix invalid reference to 'mybranch' in user manual
Ivan Stankovic [Sun, 10 Aug 2008 16:22:14 +0000 (18:22 +0200)]
Documentation: fix invalid reference to 'mybranch' in user manual

Signed-off-by: Ivan Stankovic <pokemon@fly.srk.fer.hr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoFix deleting reflog entries from HEAD reflog
Junio C Hamano [Mon, 11 Aug 2008 06:21:25 +0000 (23:21 -0700)]
Fix deleting reflog entries from HEAD reflog

dwim_ref() used to resolve HEAD symbolic ref to its target (i.e. current
branch).  This incorrectly removed the reflog entry from the current
branch when 'git reflog delete HEAD@{1}' was asked for.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoreflog test: add more tests for 'reflog delete'
Pieter de Bie [Sat, 9 Aug 2008 23:33:29 +0000 (01:33 +0200)]
reflog test: add more tests for 'reflog delete'

This adds more tests for 'reflog delete' and marks it as
broken, as currently a call to 'git reflog delete HEAD@{1}'
deletes entries in the currently checked out branch's log,
not the HEAD log.

Noticed by John Wiegley

Signed-off-by: Pieter de Bie <pdebie@ai.rug.nl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation: rev-list-options: Fix -g paragraph formatting
Thomas Rast [Sun, 10 Aug 2008 13:04:34 +0000 (15:04 +0200)]
Documentation: rev-list-options: Fix -g paragraph formatting

- Add an escape to @{now}.  Without the escape, the brace does
  something magic and eats half the sentence up to the closing brace
  at 'timestamp}'.

- Join the last paragraph with a '+'.

Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogitk: Allow safely calling nukefile from a run queue handler
Alexander Gavrilov [Sat, 9 Aug 2008 10:41:50 +0000 (14:41 +0400)]
gitk: Allow safely calling nukefile from a run queue handler

Originally dorunq assumed that the queue entry remained first
in the queue after the script eval, and blindly removed it.
However, if the handler calls nukefile, it may not be the
case anymore, and a random queue entry gets dropped instead.

This makes dorunq remove the entry before calling the
script, and adds a global variable to allow other functions
to determine if they are called from within a dorunq handler.

Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
16 years agogit-gui 0.11 gitgui-0.11.0
Shawn O. Pearce [Sun, 10 Aug 2008 07:02:08 +0000 (00:02 -0700)]
git-gui 0.11

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
16 years agoUpdate draft RelNotes for 1.6.0
Junio C Hamano [Sat, 9 Aug 2008 22:11:48 +0000 (15:11 -0700)]
Update draft RelNotes for 1.6.0

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint'
Junio C Hamano [Sat, 9 Aug 2008 08:40:08 +0000 (01:40 -0700)]
Merge branch 'maint'

* maint:
  asciidoc markup fixes
  Fail properly when cloning from invalid HTTP URL

Conflicts:
Documentation/git-push.txt

16 years agomailinfo: fix MIME multi-part message boundary handling
Junio C Hamano [Sat, 9 Aug 2008 08:17:24 +0000 (01:17 -0700)]
mailinfo: fix MIME multi-part message boundary handling

After finding a MIME multi-part message boundary line, the handle_body()
function is supposed to first flush any accumulated contents from the
previous part to the output stream.  However, the code mistakenly output
the boundary line it found.

The old code that used one global, fixed-length buffer line[] used an
alternate static buffer newline[] for keeping track of this accumulated
contents and flushed newline[] upon seeing the boundary; when 3b6121f
(git-mailinfo: use strbuf's instead of fixed buffers, 2008-07-13)
converted a fixed-length buffer in this program to use strbuf,these two
buffers were converted to "line" and "prev" (the latter of which now has a
much more sensible name) strbufs, but the code mistakenly flushed "line"
(which contains the boundary we have just found), instead of "prev".

This resulted in the first boundary to be output in front of the first
line of the message.

The rewritten implementation of handle_boundary() lost the terminating
newline; this would then result in the second line of the message to be
stuck with the first line.

The is_multipart_boundary() was designed to catch both the internal
boundary and the terminating one (the one with trailing "--"); this also
was broken with the rewrite, and the code in the handle_boundary() to
handle the terminating boundary was never triggered.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobuiltin-rm: Add a --force flag
Pieter de Bie [Fri, 8 Aug 2008 22:37:02 +0000 (00:37 +0200)]
builtin-rm: Add a --force flag

This adds a --force flag to git-rm, making it somewhat easier for
subversion people to switch.

Signed-off-by: Pieter de Bie <pdebie@ai.rug.nl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-svn: wrap long lines in a few places
Eric Wong [Fri, 8 Aug 2008 08:41:58 +0000 (01:41 -0700)]
git-svn: wrap long lines in a few places

Oops, I let a few patches slip by with long lines in them.
Extracted from an unrelated patch by: Marcus Griep <marcus@griep.us>

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-svn: Allow deep branch names by supporting multi-globs
Marcus Griep [Fri, 8 Aug 2008 08:41:57 +0000 (01:41 -0700)]
git-svn: Allow deep branch names by supporting multi-globs

Some repositories use a deep branching strategy, such as:

    branches/1.0/1.0.rc1
    branches/1.0/1.0.rc2
    branches/1.0/1.0.rtm
    branches/1.0/1.0.gold

Only allowing a single glob stiffles this.

This change allows for a single glob 'set' to accept this deep
branching strategy.

The ref glob depth must match the branch glob depth.  When using
the -b or -t options for init or clone, this is automatically
done.

For example, using the above branches:

  svn-remote.svn.branches = branches/*/*:refs/remote/*/*

gives the following branch names:

  1.0/1.0.rc1
  1.0/1.0.rc2
  1.0/1.0.rtm
  1.0/1.0.gold

[ew:
  * removed unrelated line-wrapping changes
  * fixed line-wrapping in a few more places
  * removed trailing whitespace
  * fixed bashism in test
  * removed unnecessary httpd startup in test
  * changed copyright on tests to 2008 Marcus Griep
  * added executable permissions to new tests
]

Signed-off-by: Marcus Griep <marcus@griep.us>
Acked-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoFix multi-glob assertion in git-svn
Marcus Griep [Fri, 8 Aug 2008 08:41:56 +0000 (01:41 -0700)]
Fix multi-glob assertion in git-svn

Fixes bad regex match check for multiple globs (would always return
one glob regardless of actual number).

[ew: fixed a bashism in the test and some minor line-wrapping]

Signed-off-by: Marcus Griep <marcus@griep.us>
Acked-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agofilter-branch: be more helpful when an annotated tag changes
Thomas Rast [Thu, 7 Aug 2008 23:50:31 +0000 (01:50 +0200)]
filter-branch: be more helpful when an annotated tag changes

Previously, git-filter-branch failed if it attempted to update an
annotated tag.  Now we ignore this condition if --tag-name-filter is
given, so that we can later rewrite the tag.  If no such option was
provided, we warn the user that he might want to run with
"--tag-name-filter cat" to achieve the intended effect.

Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation: filter-branch: document how to filter all refs
Thomas Rast [Thu, 7 Aug 2008 14:16:03 +0000 (16:16 +0200)]
Documentation: filter-branch: document how to filter all refs

Document the '--' option that can be used to pass rev-list options
(not just arguments), and give an example usage of '-- --all'.  Remove
reference to "the new branch name"; filter-branch takes arbitrary
arguments to rev-list since dfd05e3.

Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMakefile: add a target which will abort compilation with ancient shells
Brandon Casey [Thu, 7 Aug 2008 19:06:26 +0000 (14:06 -0500)]
Makefile: add a target which will abort compilation with ancient shells

This adds a make target which can be used to try to execute certain shell
constructs which are required for compiling and running git.

This patch provides a test for the $() notation for command substition
which is used in the Makefile and extensively in the git scripts.

The make target is named in such a way as to be a hint to the user that
SHELL_PATH should be set to an appropriate shell. If the shell command
fails, the user should receive a message similar to the following:

make: *** [please_set_SHELL_PATH_to_a_more_modern_shell] Error 2

Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobash: remove redundant check for 'git stash apply' options
SZEDER Gábor [Wed, 6 Aug 2008 15:45:23 +0000 (17:45 +0200)]
bash: remove redundant check for 'git stash apply' options

It will never trigger anyway because of the first check, and even if it
would, it would not offer the command line option.

Signed-off-by: SZEDER Gábor <szeder@ira.uka.de>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoGIT-VERSION-GEN: mark the version 'dirty' only if there are modified files
Junio C Hamano [Fri, 8 Aug 2008 20:31:27 +0000 (13:31 -0700)]
GIT-VERSION-GEN: mark the version 'dirty' only if there are modified files

We used to mark the version string with '-dirty' if the cache was not up
to date, but the only thing we want to know is if the binaries are built
from modified source.  Refresh the cache to avoid false dirtyness.

Christian Jaeger noticed this issue while building under fakeroot
environment (without -u) that lies about the file ownership data.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMakefile: set SHELL to value of SHELL_PATH
Brandon Casey [Thu, 7 Aug 2008 19:03:42 +0000 (14:03 -0500)]
Makefile: set SHELL to value of SHELL_PATH

Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation: user-manual: "git commit -a" doesn't motivate .gitignore
Jonathan Nieder [Wed, 6 Aug 2008 21:22:00 +0000 (16:22 -0500)]
Documentation: user-manual: "git commit -a" doesn't motivate .gitignore

"git commit -a" ignores untracked files and follows all tracked
files, regardless of whether they are listed in .gitignore.  So
don't use it to motivate gitignore.

Signed-off-by: Jonathan Nieder <jrnieder@uchicago.edu>
Acked-by: J. Bruce Fields <bfields@fieldses.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-svn: add ability to specify --commit-url for dcommit
Eric Wong [Thu, 7 Aug 2008 09:06:16 +0000 (02:06 -0700)]
git-svn: add ability to specify --commit-url for dcommit

This allows one to use public svn:// URLs for fetch and
svn+ssh:// URLs for committing (without using the complicated
rewriteRoot option, reimporting or git-filter-branch).

Using this can also help avoid unnecessary server
authentication/encryption overhead on busy SVN servers.

Along with the new --revision option, this can also be allowed
to override the branch detection in dcommit, too.  This is
potentially dangerous and not recommended!  (And also purposely
undocumented, but the loaded gun is there in case somebody
wants to make it safe).

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation: commit-tree: remove 16 parents restriction
Thomas Rast [Fri, 8 Aug 2008 07:52:55 +0000 (09:52 +0200)]
Documentation: commit-tree: remove 16 parents restriction

ef98c5ca lifted the 16 parents restriction in builtin-commit-tree.c,
but forgot to update the documentation.

Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoasciidoc markup fixes
Junio C Hamano [Thu, 7 Aug 2008 23:05:25 +0000 (16:05 -0700)]
asciidoc markup fixes

I see quite a few pages on k.org site, e.g.

    http://www.kernel.org/pub/software/scm/git/docs/git-rerere.html
    (scroll down to find "After this test merge")

are misformatted to lose teletype text '+' that is followed by a comma,
and turns the following paragraph all typeset in teletype.

This patch seems to fix the issue at the site (meaning, with the
particular vintage of asciidoc and docbook toolchain), without breaking
things with the version I have at my primary development machine, but
wider testing is very much appreciated.

After this patch,

    git grep '`+`,' -- Documentation

should report noting.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoclone --mirror: avoid storing repeated tags
Johannes Schindelin [Fri, 8 Aug 2008 02:29:35 +0000 (04:29 +0200)]
clone --mirror: avoid storing repeated tags

With --mirror, clone asks for refs/* already, so it does not need to
ask for ref/tags/*, too.

Noticed by Cesar Eduardo Barros.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'lt/config-fsync' into maint
Junio C Hamano [Thu, 7 Aug 2008 01:56:23 +0000 (18:56 -0700)]
Merge branch 'lt/config-fsync' into maint

* lt/config-fsync:
  Add config option to enable 'fsync()' of object files
  Split up default "i18n" and "branch" config parsing into helper routines
  Split up default "user" config parsing into helper routine
  Split up default "core" config parsing into helper routine

16 years agoMerge branch 'jc/reflog-expire' into maint
Junio C Hamano [Thu, 7 Aug 2008 01:52:50 +0000 (18:52 -0700)]
Merge branch 'jc/reflog-expire' into maint

* jc/reflog-expire:
  Make default expiration period of reflog used for stash infinite
  Per-ref reflog expiry configuration

16 years agoMerge branch 'ag/rewrite_one' into maint
Junio C Hamano [Thu, 7 Aug 2008 01:48:32 +0000 (18:48 -0700)]
Merge branch 'ag/rewrite_one' into maint

* ag/rewrite_one:
  Fix quadratic performance in rewrite_one.

16 years agoFail properly when cloning from invalid HTTP URL
Petr Baudis [Thu, 7 Aug 2008 00:06:30 +0000 (02:06 +0200)]
Fail properly when cloning from invalid HTTP URL

Currently, when cloning from invalid HTTP URL, git clone will possibly
return curl error, then a confusing message about remote HEAD and then
return success and leave an empty repository behind, confusing either
the end-user or the automated service calling it (think repo.or.cz).

This patch changes the error() calls in get_refs_via_curl() to die()s,
akin to the other get_refs_*() functions.

Cc: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoGIT 1.6.0-rc2 v1.6.0-rc2
Junio C Hamano [Wed, 6 Aug 2008 19:44:55 +0000 (12:44 -0700)]
GIT 1.6.0-rc2

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoperl/Makefile: handle paths with spaces in the NO_PERL_MAKEMAKER section
Brandon Casey [Tue, 5 Aug 2008 23:23:08 +0000 (18:23 -0500)]
perl/Makefile: handle paths with spaces in the NO_PERL_MAKEMAKER section

Use double quotes to protect against paths which may contain spaces.

Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoSync with 1.5.6.5
Junio C Hamano [Wed, 6 Aug 2008 20:50:42 +0000 (13:50 -0700)]
Sync with 1.5.6.5

16 years agoGIT 1.5.6.5 v1.5.6.5
Junio C Hamano [Wed, 6 Aug 2008 19:04:06 +0000 (12:04 -0700)]
GIT 1.5.6.5

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoFiles given on the command line are relative to $cwd
Junio C Hamano [Wed, 6 Aug 2008 18:43:47 +0000 (11:43 -0700)]
Files given on the command line are relative to $cwd

When running "git commit -F file" and "git tag -F file" from a
subdirectory, we should take it as relative to the directory we started
from, not relative to the top-level directory.

This adds a helper function "parse_options_fix_filename()" to make it more
convenient to fix this class of issues.  Ideally, parse_options() should
support a new type of option, "OPT_FILENAME", to do this uniformly, but
this patch is meant to go to 'maint' to fix it minimally.

One thing to note is that value for "commit template file" that comes from
the command line is taken as relative to $cwd just like other parameters,
but when it comes from the configuration varilable 'commit.template', it
is taken as relative to the working tree root as before.  I think this
difference actually is sensible (not that I particularly think
commit.template itself is sensible).

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge git://repo.or.cz/git-gui
Junio C Hamano [Wed, 6 Aug 2008 20:32:18 +0000 (13:32 -0700)]
Merge git://repo.or.cz/git-gui

* git://repo.or.cz/git-gui:
  git-gui: Adapt discovery of oguilib to execdir 'libexec/git-core'
  git-gui: add a part about format strings in po/README
  git-gui: update po/it.po
  git-gui: update Japanese translation
  git-gui: Update swedish translation.
  git-gui: Update git-gui.pot for 0.11 nearing release
  git-gui: Update German translation

16 years agoMerge git://git.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Wed, 6 Aug 2008 20:32:12 +0000 (13:32 -0700)]
Merge git://git.kernel.org/pub/scm/gitk/gitk

* git://git.kernel.org/pub/scm/gitk/gitk:
  gitk: Update swedish translation.
  gitk: Updated German translation
  gitk: Fallback to selecting the head commit upon load
  gitk: Fixed automatic row selection during load
  gitk: Fixed broken exception handling in diff
  gitk: On Windows, use a Cygwin-specific flag for kill
  gitk: Arrange to kill diff-files & diff-index on quit
  gitk: Kill back-end processes on window close

16 years agoMerge branch 'rs/archive-parse-options'
Junio C Hamano [Wed, 6 Aug 2008 20:31:38 +0000 (13:31 -0700)]
Merge branch 'rs/archive-parse-options'

* rs/archive-parse-options:
  archive: allow --exec and --remote without equal sign

16 years agofix diff-tree --stdin documentation
Junio C Hamano [Wed, 6 Aug 2008 05:32:28 +0000 (22:32 -0700)]
fix diff-tree --stdin documentation

Long time ago, the feature of "diff-tree --stdin" to take a commit and its
parents on one line was broken, and did not support the common:

    git rev-list --parents $commits... -- $paths... |
    git diff-tree --stdin -v -p

usage pattern by Porcelains properly.  For diff-tree to talk sensibly
about commits, it needs to see commits, not just trees; the code was fixed
to take list of commits on the standard input in 1.2.0.

However we left the documentation stale for a long time, until Karl
Hasselström finally noticed it very recently.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoOptimize sha1_object_info for loose objects, not concurrent repacks
Steven Grimm [Tue, 5 Aug 2008 20:08:41 +0000 (13:08 -0700)]
Optimize sha1_object_info for loose objects, not concurrent repacks

When dealing with a repository with lots of loose objects, sha1_object_info
would rescan the packs directory every time an unpacked object was referenced
before finally giving up and looking for the loose object. This caused a lot
of extra unnecessary system calls during git pack-objects; the code was
rereading the entire pack directory once for each loose object file.

This patch looks for a loose object before falling back to rescanning the
pack directory, rather than the other way around.

Signed-off-by: Steven Grimm <koreth@midwinter.com>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoTeach fsck and prune that tmp_obj_ file names may not be 14 bytes long
Brandon Casey [Tue, 5 Aug 2008 18:01:50 +0000 (13:01 -0500)]
Teach fsck and prune that tmp_obj_ file names may not be 14 bytes long

As Shawn pointed out, not all temporary file creation routines can
ensure that the generated temporary file is of a certain length.
e.g. Java's createTempFile(prefix, suffix). So just depend on the
prefix 'tmp_obj_' for detection.

Update prune, and fix the "fix" introduced by a08c53a1 :)

Signed-off-by: Brandon "appendixless" Casey <casey@nrlssc.navy.mil>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>