git.git
18 years agodiff-files: show diffs with stage0 and unmerged stage at the same time.
Junio C Hamano [Wed, 30 Nov 2005 10:16:36 +0000 (02:16 -0800)]
diff-files: show diffs with stage0 and unmerged stage at the same time.

After thinking about it more, I realized that much of the change
I did on top of Linus' version does not make much sense.  This
commit reverts it so that it by default shows diffs with stage0
paths or stage2 paths with working tree; the unmerged stage to
use can be overridden with -1/-2/-3 option (-2 is the default so
essentially is a no-op).

When the index file is unmerged, we are by definition in the
middle of a conflicting merge, and we should show the diff with
stage 2 by default.  More importantly, paths without conflicts
are updated in the working tree and collapsed to stage0 in the
index, so showing diff with stage0 at the same time does not
hurt.  In normal cases, stage0 entries should be in sync with
the working tree files and does not clutter the output.  It even
helps the user to realize that the working tree has local
changes unrelated to the merge and remember to be careful not to
do a "git-commit -a" after resolving the conflicts.

When there is no unmerged entries, giving diff_unmerged_stage a
default value of 2 does not cause any harm, because it would not
be used anyway.  So in all, always showing diff between stage0
paths and unmerged entries from a stage (defaulting to 2) is the
right thing to do, as Linus originally did.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agomerge-one-file: leave unmerged index entries upon automerge failure.
Linus Torvalds [Wed, 30 Nov 2005 05:06:10 +0000 (21:06 -0800)]
merge-one-file: leave unmerged index entries upon automerge failure.

When automerge fails, we used to collapse the path to stage0
from "our" branch, to help "diff-files" users to view the
half-merged state against the current HEAD.  Now diff-files has
been taught how to compare with unmerged stage2,leaving them
unmerged is a better thing to do, especially this prevents the
unresolved conflicts to be committed by mistake.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agodiff-files: -1/-2/-3 to diff against unmerged stage.
Linus Torvalds [Wed, 30 Nov 2005 05:06:10 +0000 (21:06 -0800)]
diff-files: -1/-2/-3 to diff against unmerged stage.

While resolving conflicted merge, it was not easy to compare the
working tree file with unmerged index entries.  This commit
introduces new options -1/-2/-3 (with synonyms --base, --ours,
and --theirs) to compare working tree files with specified
stages.

When none of these options are given, the command defaults to -2
if the index file is unmerged, otherwise it acts as before.

[jc: majorly butchered from the version Linus originally posted.]

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-diff: do not turn off -p/-M with any diff options.
Junio C Hamano [Wed, 30 Nov 2005 05:59:04 +0000 (21:59 -0800)]
git-diff: do not turn off -p/-M with any diff options.

When the user gives a diff option (e.g. --cached) to "git diff",
we turned off the built-in default option -p, which is usually
not what user wants to see.

This commit makes lack of --name-status, --name-only nor -r to
add -p, and lack of -B*, -C* nor -M* to add -M to the flags
given to the underlying diff.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge http://www.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Tue, 29 Nov 2005 23:47:32 +0000 (15:47 -0800)]
Merge http://www.kernel.org/pub/scm/gitk/gitk

19 years ago[PATCH] Add tests for git-mv in subdirectories
Alex Riesen [Tue, 29 Nov 2005 22:10:24 +0000 (23:10 +0100)]
[PATCH] Add tests for git-mv in subdirectories

Junio C Hamano, Sat, Nov 26, 2005 03:45:52 +0100:

> I haven't seriously used git-mv myself, so
> somebody needs to test it, and if it actually works and Ack on
> it, please.

It actually works in subdirs.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Make git-mv work in subdirectories, too
Johannes Schindelin [Sat, 26 Nov 2005 02:45:52 +0000 (18:45 -0800)]
[PATCH] Make git-mv work in subdirectories, too

Turns out, all git programs git-mv uses are capable of operating in
a subdirectory just fine. So don't complain about it.

[jc: I think that sounds sane.  You need to grab the exit status from
 `git-rev-parse --git-dir`, which I added.  Alex Riesen says this
 worked fine.]

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogitk: Add a preferences dialog with some basic stuff
Paul Mackerras [Tue, 29 Nov 2005 22:28:16 +0000 (09:28 +1100)]
gitk: Add a preferences dialog with some basic stuff

There is a lot more that could be put in, such as a selector for
the font family etc., but this is a start.

Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agoSVN import: Use one log call
Matthias Urlichs [Tue, 29 Nov 2005 07:13:04 +0000 (08:13 +0100)]
SVN import: Use one log call

One "svn log" (or its equivalent) per revision adds delay and server load.
Instead, open two SVN connections -- one for the log, and one for the files.

Positive side effect: Only those log entries which actually contain data
are committed => no more empty commits.

Also, change the "-l" option to set the maximum revision to be pulled,
not the number of revisions.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix typos and minor format issues.
jdl@freescale.com [Tue, 29 Nov 2005 14:59:56 +0000 (08:59 -0600)]
Fix typos and minor format issues.

Signed-off-by: Jon Loeliger <jdl@freescale.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agohttp-push cleanup
Nick Hengeveld [Tue, 29 Nov 2005 17:33:36 +0000 (09:33 -0800)]
http-push cleanup

The malloc patch from Jan Andres fixed the problem that was causing a
segfault when freeing the lock token, and Johannes Schindelin found
and fixed a problem when no URL is specified on the command line.

Signed-off-by: Nick Hengeveld <nickh@reactrix.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-clone --shared should imply --local
Pavel Roskin [Tue, 29 Nov 2005 06:20:49 +0000 (01:20 -0500)]
git-clone --shared should imply --local

The "--shared" option to git-clone is silently ignored if "--local" is
not specified.  The manual doesn't mention such dependency.  Make
"--shared" imply "--local".

Signed-off-by: Pavel Roskin <proski@gnu.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agowrite_name_quoted(): make one of the path a counted string.
Junio C Hamano [Tue, 29 Nov 2005 06:55:25 +0000 (22:55 -0800)]
write_name_quoted(): make one of the path a counted string.

This is to prepare for ls-tree updates.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoDocumentation: Describe merge operation a bit better.
Junio C Hamano [Tue, 29 Nov 2005 06:54:30 +0000 (22:54 -0800)]
Documentation: Describe merge operation a bit better.

In git-merge documentation, add a section to describe what happens to
the index and working tree during merge, and what their cleanliness
requirements are before the merge.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix typo in http-push.c
Jan Andres [Tue, 29 Nov 2005 00:51:54 +0000 (01:51 +0100)]
Fix typo in http-push.c

Typo resulted in accessing past the beginning of a string causing segfaults.

[jc: signoffs?]

19 years agoname-rev: fix parent counting.
Junio C Hamano [Tue, 29 Nov 2005 04:51:44 +0000 (20:51 -0800)]
name-rev: fix parent counting.

Noticed by linux@horizon.com.  The first merge parent (typically
"our branch") is ^1, not ^0, and the first other branch is ^2.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agorebase: one safety net, one bugfix and one optimization.
Junio C Hamano [Mon, 28 Nov 2005 21:00:31 +0000 (13:00 -0800)]
rebase: one safety net, one bugfix and one optimization.

When a .dotest from a previously failed rebase or patch
application exists, rebase got confused and tried to apply
mixture of what was already there and what is being rebased.
Check the existence of the directory and barf.

It failed with an mysterious "fatal: cannot read mbox" message
if the branch being rebased is fully in sync with the base.
Also if the branch is a proper descendant of the base, there is
no need to run rebase logic.  Prevent these from happening by
checking where the merge-base is.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge http://www.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Mon, 28 Nov 2005 11:11:11 +0000 (03:11 -0800)]
Merge http://www.kernel.org/pub/scm/gitk/gitk

19 years agogit-mv: quote $src in regexp properly.
Junio C Hamano [Mon, 28 Nov 2005 10:54:05 +0000 (02:54 -0800)]
git-mv: quote $src in regexp properly.

Noticed and fixed by Matthias Urlichs and Josef Weidendorfer.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] gitk: Use i18n.commitencoding configuration item.
Junio C Hamano [Mon, 28 Nov 2005 00:12:51 +0000 (16:12 -0800)]
[PATCH] gitk: Use i18n.commitencoding configuration item.

Hardcoding "utf-8" in the script breaks projects that use local
encoding, so allow setting i18n.commitEncoding.

Signed-off-by: Junio C Hamano <junkio@cox.net>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agoconfig.c: constness tightening to avoid compilation warning.
Junio C Hamano [Mon, 28 Nov 2005 09:46:15 +0000 (01:46 -0800)]
config.c: constness tightening to avoid compilation warning.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogitk: Fix diff this->selected and selected->this functions
Paul Mackerras [Mon, 28 Nov 2005 09:41:56 +0000 (20:41 +1100)]
gitk: Fix diff this->selected and selected->this functions

The change in 8b7e5d76e836396a097bb6f61cf930ea872a7bd3, which makes
a couple of git-diff-tree calls supply only one id rather than two,
fixes the display when showing what a single commit did with dense
revlists, but broke the diff this->selected and diff selected->this
right-click menu functions.

Yann Dirson pointed this out and had a patch that fixed the diff
menu functions by passing a "singlecommit" flag around.  This fixes
it a bit differently, by making the ids and diffids variables be
either a single id, in the case of showing what a commit did, or
{oldid newid}, in the case of the diff menu functions.  That way
we can just pass $ids to git-diff-tree as is.  Most of the changes
in fact are just reversing the order of ids in $ids and $diffids,
because they used to be {child parent}, but git-diff-tree requires
old id before new id.

Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agomailinfo: Do not use -u=<encoding>; say --encoding=<encoding>
Junio C Hamano [Mon, 28 Nov 2005 09:29:52 +0000 (01:29 -0800)]
mailinfo: Do not use -u=<encoding>; say --encoding=<encoding>

Specifying the value for a single letter, single dash option
parameter with equal sign looked funny, and more importantly
calling the flag to override encoding from utf-8 to something
else "-u" (obviously abbreviated from "utf-8") did not make any
sense.  So spell it out.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agomailinfo: Use i18n.commitencoding
Junio C Hamano [Mon, 28 Nov 2005 00:29:38 +0000 (16:29 -0800)]
mailinfo: Use i18n.commitencoding

This uses i18n.commitencoding configuration item to pick up the
default commit encoding for the repository when converting form
e-mail encoding to commit encoding (the default is utf8).

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agomailinfo: allow -u to fall back on latin1 to utf8 conversion.
Junio C Hamano [Mon, 28 Nov 2005 00:22:16 +0000 (16:22 -0800)]
mailinfo: allow -u to fall back on latin1 to utf8 conversion.

When the message body does not identify what encoding it is in,
-u assumes it is in latin-1 and converts it to utf8, which is
the recommended encoding for git commit log messages.

With -u=<encoding>, the conversion is made into the specified
one, instead of utf8, to allow project-local policies.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoIntroduce i18n.commitencoding.
Junio C Hamano [Mon, 28 Nov 2005 00:09:40 +0000 (16:09 -0800)]
Introduce i18n.commitencoding.

This is to hold what the project-local rule as to the
charset/encoding for the commit log message is.  Lack of it
defaults to utf-8.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix gitk this->selected diffs
Yann Dirson [Sun, 27 Nov 2005 22:29:30 +0000 (23:29 +0100)]
Fix gitk this->selected diffs

The change made in 8b7e5d76e836396a097bb6f61cf930ea872a7bd3 to
accomodate dense revlists in single-commit diffs has broken computing
of diffs between arbitrary trees, which does need to consider two
commit ids.

This patch changes the two git-diff-tree calls to get the necessary
two ids in this case.  It does so by propagating a "singlecommit" flag
through all functions involved via an additional argument.

Signed-off-by: Yann Dirson <ydirson@altern.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge http://www.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Sun, 27 Nov 2005 22:42:15 +0000 (14:42 -0800)]
Merge http://www.kernel.org/pub/scm/gitk/gitk

19 years agogit-mv: follow -k request even on failing renames
Josef Weidendorfer [Sun, 27 Nov 2005 21:11:33 +0000 (22:11 +0100)]
git-mv: follow -k request even on failing renames

-k requests to keep running on an error condition.
Previously, git-mv stopped on failing renames even with -k.

There are some error conditions which are not checked in the
first phase of git-mv, eg. 'permission denied'. Still, option
-k should work.

Signed-off-by: Josef Weidendorfer <Josef.Weidendorfer@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoSmall fixes in Documentation/git-mv.txt
Josef Weidendorfer [Sun, 27 Nov 2005 21:08:33 +0000 (22:08 +0100)]
Small fixes in Documentation/git-mv.txt

The two synopsis lines have to be prefixed with a space
so that asciidoc inserts a line break inbetween for the
manual page.

Signed-off-by: Josef Weidendorfer <Josef.Weidendorfer@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-mv: fully detect 'directory moved into itself'
Josef Weidendorfer [Sun, 27 Nov 2005 21:06:42 +0000 (22:06 +0100)]
git-mv: fully detect 'directory moved into itself'

This gives a better error message when trying to move a directory
into some subdirectory of itself; ie. no real bug fix: renaming
already failed before, but with a strange "invalid argument".

Signed-off-by: Josef Weidendorfer <Josef.Weidendorfer@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-mv: keep git index consistent with file system on failed rename
Josef Weidendorfer [Sun, 27 Nov 2005 21:04:14 +0000 (22:04 +0100)]
git-mv: keep git index consistent with file system on failed rename

When doing multiple renames, and a rename in the middle fails,
git-mv did not store the successful renames in the git index;
this is fixed by delaying the error message on a failed rename
to after the git updating.

Signed-off-by: Josef Weidendorfer <Josef.Weidendorfer@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-mv: shrink usage, no usage on error
Josef Weidendorfer [Sun, 27 Nov 2005 20:58:52 +0000 (21:58 +0100)]
git-mv: shrink usage, no usage on error

Small fixes to be consistent with other git scripts:
- usage message is only about options and arguments
- on error, exit(1) without the usage message

Additionally, "beautifies" output with -n a little bit

Signed-off-by: Josef Weidendorfer <Josef.Weidendorfer@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agosetup_git_directory(): check repository format version.
Junio C Hamano [Sat, 26 Nov 2005 00:08:48 +0000 (16:08 -0800)]
setup_git_directory(): check repository format version.

After figuring out the GIT_DIR location, make sure the
repository is of the right vintage, by calling
check_repository_format(). .

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoinit-db: check template and repository format.
Junio C Hamano [Sat, 26 Nov 2005 00:03:56 +0000 (16:03 -0800)]
init-db: check template and repository format.

This makes init-db repository version aware.

It checks if an existing config file says the repository being
reinitialized is of a wrong version and aborts before doing
further harm.

When copying the templates, it makes sure the they are of the
right repository format version.  Otherwise the templates are
ignored with an warning message.

It copies the templates before creating the HEAD, and if the
config file is copied from the template directory, reads it,
primarily to pick up the value of core.symrefsonly.

It changes the way the result of the filemode reliability test
is written to the configuration file using git_config_set().
The test is done even if the config file was copied from the
templates.

And finally, our own repository format version is written to the
config file.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoCheck repository format version in enter_repo().
Junio C Hamano [Fri, 25 Nov 2005 18:48:26 +0000 (10:48 -0800)]
Check repository format version in enter_repo().

After daemon, upload-pack and receive-pack find out where the
git directory is and chdir() there, make sure that repository is
in a format we understand, after putenv("GIT_DIR=.") so that it
knows to pick up the configuration file from there.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoRepository format version check.
Junio C Hamano [Fri, 25 Nov 2005 23:59:09 +0000 (15:59 -0800)]
Repository format version check.

This adds the repository format version code, first done by
Martin Atukunda.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoformat-patch: output filename reported to stdout verbatim.
Junio C Hamano [Sat, 26 Nov 2005 20:09:07 +0000 (12:09 -0800)]
format-patch: output filename reported to stdout verbatim.

Prepending asterisk to the output was just adding noise, and
making scripts like proposed git-send-mail by Andreas Ericsson
do unnecessary work.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoname-rev: fix off-by-one error in --stdin.
Junio C Hamano [Sat, 26 Nov 2005 07:36:58 +0000 (23:36 -0800)]
name-rev: fix off-by-one error in --stdin.

It dropped the last hexdigit in the object name.

[jc: Noticed and patch supplied by ALASCM, reworked to apply at
the right place by me]

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-mv is not able to handle big directories
Alexander Litvinov [Wed, 23 Nov 2005 10:19:41 +0000 (16:19 +0600)]
git-mv is not able to handle big directories

Use update-index --stdin to handle large number of files without
breaking exec() argument storage limit.

[jc: with minor cleanup from the version posted on the list]

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoshell.c: complain on insufficient arguments.
Junio C Hamano [Sat, 26 Nov 2005 04:57:02 +0000 (20:57 -0800)]
shell.c: complain on insufficient arguments.

Originally noticed by Tommi Virtanen, but done slightly differently.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-sh-setup: move the repository check to a core program.
Junio C Hamano [Fri, 25 Nov 2005 23:52:57 +0000 (15:52 -0800)]
git-sh-setup: move the repository check to a core program.

Any core commands that use setup_git_directory() now check if
given GIT_DIR is really a valid repository, so the same check in
git-sh-setup can use it without reimplementing it in shell.
This commit changes git-sh-setup to use git-var command for
that, although any other commands would do.

Note that we export GIT_DIR explicitly when calling git-var;
without it, the caller of this script would use GIT_DIR that we
return (which is to assume ./.git unless the caller has it
elsewhere) while git-var would go up to find a .git directory in
our parent directories, which would be checking a different
directory from what our callers will be using.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agosetup_git_directory: make sure GIT_DIR is a valid repository.
Junio C Hamano [Fri, 25 Nov 2005 23:43:41 +0000 (15:43 -0800)]
setup_git_directory: make sure GIT_DIR is a valid repository.

setup_git_directory() always trusted what the user told where
GIT_DIR was, and assumed that is a valid .git/ directory.  This
commit changes it to at least do the same level validation as
is_toplevel_directory() does -- has refs/, has objects/ unless
GIT_OBJECT_DIRECTORY is set, and has valid HEAD symlink or
symref.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-var: make it work in subdirectory.
Junio C Hamano [Thu, 24 Nov 2005 09:31:01 +0000 (01:31 -0800)]
git-var: make it work in subdirectory.

Use setup_git_directory() so that it can find its .git
directory.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-sh-setup: die if outside git repository.
Junio C Hamano [Thu, 24 Nov 2005 08:12:11 +0000 (00:12 -0800)]
git-sh-setup: die if outside git repository.

Now all the users of this script detect its exit status and die,
complaining that it is outside git repository.  So move the code
that dies from all callers to git-sh-setup script.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoparse-remote and ls-remote clean-up.
Junio C Hamano [Thu, 24 Nov 2005 07:46:13 +0000 (23:46 -0800)]
parse-remote and ls-remote clean-up.

There is no reason to use git-sh-setup from git-ls-remote.
git-parse-remote can help the caller to use .git/remotes
shortcut if it is run inside a git repository, but can still be
useful outside a git repositoryas long as the caller does not
use any shortcut.  Use "git-rev-parse --git-dir" to figure out
where the GIT_DIR is, instead of using git-sh-setup.

This also makes "git-ls-remote origin" to work from inside a
subdirectory of a git managed repository as a side effect.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoPrevent "git-commit -a path1 path2..."
Junio C Hamano [Fri, 25 Nov 2005 21:33:14 +0000 (13:33 -0800)]
Prevent "git-commit -a path1 path2..."

When you want to create a partial commit, giving -a by mistake
would ignore the given path.  Prevent it.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoconfig.c: avoid shadowing global.
Junio C Hamano [Fri, 25 Nov 2005 19:10:49 +0000 (11:10 -0800)]
config.c: avoid shadowing global.

This is purely cosmetic, but avoid shadowing "FILE *config_file"
global in git_config_set_multivar() function.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix a warning about unused value.
YOSHIFUJI Hideaki / 吉藤英明 [Fri, 25 Nov 2005 10:03:05 +0000 (19:03 +0900)]
Fix a warning about unused value.

Fix a warning:
  git.c:276: warning: value computed is not used

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agocvsimport: Don't let local settings hinder cvs user-migration.
Andreas Ericsson [Fri, 25 Nov 2005 01:37:09 +0000 (02:37 +0100)]
cvsimport: Don't let local settings hinder cvs user-migration.

Avoid this by passing "--norc" to cvsps.

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoRename git-config-set to git-repo-config
Johannes Schindelin [Thu, 24 Nov 2005 10:36:01 +0000 (11:36 +0100)]
Rename git-config-set to git-repo-config

... and adjust all references.

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agodiff-tree: teach single-commit diff-tree to honour grafts
Junio C Hamano [Tue, 22 Nov 2005 05:49:06 +0000 (21:49 -0800)]
diff-tree: teach single-commit diff-tree to honour grafts

We used to read the commit objects by hand and ignored the grafts.
Rewrite it using lookup_commit() API, to make it grafts-aware.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoTeach update-index to read from ls-tree.
Junio C Hamano [Tue, 22 Nov 2005 05:46:57 +0000 (21:46 -0800)]
Teach update-index to read from ls-tree.

git-update-index --index-info can almost be usable to read from ls-tree
output to update the index (and not the working tree file) to HEAD commit,
but not quite.  It was designed to read from git-apply --index-info
output, and does not want " blob " in ls-tree output.  Accept that as well.

This lets us update "git-checkout <ent> <path>" that used to filter the
extra " blob " string out.  Noted by Luben.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-revert: make --edit default.
Junio C Hamano [Thu, 24 Nov 2005 00:56:55 +0000 (16:56 -0800)]
git-revert: make --edit default.

Revert always should explain why, so make --edit the default,
unless stdin is not a terminal.  If you really don't want to say
anything, you can say "git-revert --no-edit $commit", or if you
are really sick, you could also say "git-revert $commit </dev/null".
But please don't.

You can also say "git-cherry-pick --edit $commit".  Not editting
the commit log message is the default for cherry-pick.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoAllow editing of a revert-message
Linus Torvalds [Wed, 23 Nov 2005 17:57:54 +0000 (09:57 -0800)]
Allow editing of a revert-message

I think all commit operations should allow editing of the message (ie we
should do this for merges too), but that's _particularly_ true of doing a
"git revert".

We should always explain why we needed to revert something.

This patch adds a "-e" or "--edit" flag to "git revert", although I
actually suspect it should be on by default (and we should have a
"--no-edit" flag to disable it, probably together with an automatic
disable if stdin isn't a terminal).

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix git-mv for moving directories with 1 file
Josef Weidendorfer [Wed, 23 Nov 2005 11:04:23 +0000 (12:04 +0100)]
Fix git-mv for moving directories with 1 file

This is fixed by putting the file into @changedfiles/@addedfiles,
and not the directory this file is in.

Additionally, this fixes the behavior for attempting to overwrite
a file with a directory, and gives a message for all cases where
overwriting is not possible (file->dir,dir->file,dir->dir).

Thanks for Alexander Litvinov for noting this problem.

Signed-off-by: Josef Weidendorfer <Josef.Weidendorfer@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-merge: make recursive the default strategy
Junio C Hamano [Thu, 24 Nov 2005 00:23:11 +0000 (16:23 -0800)]
git-merge: make recursive the default strategy

git-pull invoked merge with recursive as the default strategy
for some time now; match it in the git-merge itself.  Also avoid
listing more than one strategy on default because we have only
one strategy that can resolve an octopus and we are already
counting heads here.  This reduces the need to stash away local
modifications.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agopack-redundant: type cleanups.
Junio C Hamano [Thu, 24 Nov 2005 00:08:36 +0000 (16:08 -0800)]
pack-redundant: type cleanups.

Binary representation of object names are unsigned char[20], not
signed.  Also verbose output had %lu format printing size_t
without (unsigned long) cast other places already had, so match
that.  Using format %zu was suggested but might not be supported
as widely.

Noted by Morten Welinder, fixed with input from H. Peter Anvin
and Hideaki Yoshifuji.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] gitk: UTF-8 support
Pavel Roskin [Wed, 23 Nov 2005 04:15:01 +0000 (23:15 -0500)]
[PATCH] gitk: UTF-8 support

Add gitencoding variable and set it to "utf-8".  Use it for converting
git-rev-list output.

Signed-off-by: Pavel Roskin <proski@gnu.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] gitk: put braces around exprs
Jeff Hobbs [Wed, 23 Nov 2005 01:39:53 +0000 (17:39 -0800)]
[PATCH] gitk: put braces around exprs

This braces all exprs.  It just seemed to be a few that were missed.

Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agoFixed git:// IPv4 address problem when compiled with -DNO_IPV6.
Paul Serice [Tue, 22 Nov 2005 13:54:23 +0000 (07:54 -0600)]
Fixed git:// IPv4 address problem when compiled with -DNO_IPV6.

Failure to dereference a pointer caused incorrect initialization of
the IPv4 address when calling connect() when compiled with -DNO_IPV6.

With this patch and yesterday's patch for git-daemon, it should now be
possible to use the native git protocol for both the client and server
on Cygwin.

Signed-off-by: Paul Serice <paul@serice.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoarguments cleanup and some formatting
Alex Riesen [Tue, 22 Nov 2005 14:59:22 +0000 (15:59 +0100)]
arguments cleanup and some formatting

Signed-off-by: Alex Riesen <ariesen@harmanbecker.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoremove unused variable
Alex Riesen [Tue, 22 Nov 2005 14:58:41 +0000 (15:58 +0100)]
remove unused variable

It is just assigned, nothing more.

Signed-off-by: Alex Riesen <ariesen@harmanbecker.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agospeedup allocation in pack-redundant.c
Alex Riesen [Tue, 22 Nov 2005 14:56:35 +0000 (15:56 +0100)]
speedup allocation in pack-redundant.c

Reuse discarded nodes of llists

Signed-off-by: Alex Riesen <ariesen@harmanbecker.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMake git-pack-redundant take a list of unimportant objs on stdin
Lukas Sandström [Fri, 18 Nov 2005 22:17:50 +0000 (23:17 +0100)]
Make git-pack-redundant take a list of unimportant objs on stdin

This lets us do "git-fsck-objects --full --unreachable | cut -d ' ' -f3 |
git-pack-redundant --all", which will keep git-pack-redundant from keeping
packs just because they contain unreachable objects.

Also add some more --verbose output.

Signed-off-by: Lukas Sandström <lukass@etek.chalmers.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge http://www.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Tue, 22 Nov 2005 09:55:15 +0000 (01:55 -0800)]
Merge http://www.kernel.org/pub/scm/gitk/gitk

19 years agogit-var: constness and globalness cleanup.
Junio C Hamano [Tue, 22 Nov 2005 07:44:35 +0000 (23:44 -0800)]
git-var: constness and globalness cleanup.

var.c::git_var read function did not have to return writable
strings; make it and the functions it points at return const char *
instead.

ident.c::get_ident() did not need to be global, so make it
static.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMove diff.renamelimit out of default configuration.
Junio C Hamano [Tue, 22 Nov 2005 06:52:37 +0000 (22:52 -0800)]
Move diff.renamelimit out of default configuration.

Otherwise we would end up linking all the unneeded stuff into git-daemon
only to link with git_default_config.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agomax-count in terms of intersection
Luben Tuikov [Fri, 18 Nov 2005 21:29:04 +0000 (16:29 -0500)]
max-count in terms of intersection

When a path designation is given, max-count counts the number
of commits therein (intersection), not globally.

This avoids the case where in case path has been inactive
for the last N commits, --max-count=N and path designation
at git-rev-list is given, would give no commits.

Signed-off-by: Luben Tuikov <ltuikov@yahoo.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoGIT: Fix compilation error in connect.c
YOSHIFUJI Hideaki / 吉藤英明 [Tue, 22 Nov 2005 03:18:23 +0000 (12:18 +0900)]
GIT: Fix compilation error in connect.c

Fix compilation error for gcc-2.95.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoIntroduce $(ALL_PROGRAMS) for 'all:' and 'install:' to operate on.
Andreas Ericsson [Mon, 21 Nov 2005 23:44:15 +0000 (00:44 +0100)]
Introduce $(ALL_PROGRAMS) for 'all:' and 'install:' to operate on.

Remove $(SIMPLE_PROGRAMS) from $(PROGRAMS) so buildrules don't have
to be overridden.

Put $(SCRIPTS) with the other target-macros so it doesn't get lonely.

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agorename/copy score parsing updates.
H. Peter Anvin [Mon, 21 Nov 2005 22:17:12 +0000 (14:17 -0800)]
rename/copy score parsing updates.

Better variant, which handles stuff like "4.5%" and rejects
"192.168.0.1".  Additionally, make sure numbers are unsigned (I'm making
them unsigned long just for the hell of it), to make sure that
artificial wraparound scenarios don't cause harm.

-hpa

[jc: with this, -M100 changes its meaning back to 10%.  People
wanting to say "pure renames only" should now say -M100% or
-M1.0; sounds a bit like an earthquake, but arguably things are
more consistent this way ;-)]

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-daemon not listening when compiled with -DNO_IPV6
Paul Serice [Mon, 21 Nov 2005 17:07:23 +0000 (11:07 -0600)]
git-daemon not listening when compiled with -DNO_IPV6

git-daemon was not listening when compiled with -DNO_IPV6.
socksetup() was not returning socket count when compiled with -DNO_IPV6.

Signed-off-by: Paul Serice <paul@serice.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-repack: Properly abort in corrupt repository
Linus Torvalds [Mon, 21 Nov 2005 20:38:31 +0000 (12:38 -0800)]
git-repack: Properly abort in corrupt repository

In a corrupt repository, git-repack produces a pack that does not
contain needed objects without complaining, and the result of this
combined with -d flag can be very painful -- e.g. a lossage of one
tree object can lead to lossage of blobs reachable only through that
tree.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoAllow hierarchical section names
Johannes Schindelin [Sun, 20 Nov 2005 20:22:19 +0000 (21:22 +0100)]
Allow hierarchical section names

A .git/config like follows becomes valid with this patch:

[remote.junio]
url = git://git.kernel.org/pub/scm/git/git.git
pull = master:junio todo:todo +pu:pu

[remote.ibook]
url = ibook:git/
pull = master:ibook
push = master:quetzal

(This patch only does the ini file thing, git-fetch and friends still
ignore these values).

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-config-set: Properly terminate strings with '\0'
Johannes Schindelin [Mon, 21 Nov 2005 10:18:20 +0000 (11:18 +0100)]
git-config-set: Properly terminate strings with '\0'

When a lowercase version of the key was generated, it was not
terminated. Strangely enough, it worked on Linux and macosx anyway.
Just cygwin barfed.

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-proxy updates.
Junio C Hamano [Sat, 19 Nov 2005 11:48:56 +0000 (03:48 -0800)]
git-proxy updates.

This builds on top of the git-proxy mechanism Paul Collins did,
and updates its configuration mechanism.

 * GIT_PROXY_COMMAND environment variable is used as the
   catch-all fallback, as in the original.  This has not
   changed.

 * Renames proxy configuration variables to core.gitproxy; this
   has become a multi-value variable per list discussion, most
   notably from suggestion by Linus.

[core]
;# matches www.kernel.org as well
gitproxy = netcatter for kernel.org
gitproxy = netscatter for sample.xz
gitproxy = none for mydomain.xz
gitproxy = netcatter-default

   The values are command names, followed by an optional " for "
   and domainname; the first tail-match of the domainname
   determines which proxy command is used.  An entry without "
   for " matches any domain and can be used as the default.

   The command name "none" is special -- it tells the mechanism
   not to use any proxy command and use the native git://
   connection.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoproxy-command support for git://
Paul Collins [Fri, 4 Nov 2005 14:57:16 +0000 (14:57 +0000)]
proxy-command support for git://

Here is an updated patch that first looks for GIT_PROXY_COMMAND
in the environment and then git.proxycommand in the repository's
configuration file.  I have left the calling convention the same
argv[1] is the host and argv[2] is the port.

I've taken the hostname parsing verbatim from git_tcp_connect(),
so it should now support an explicit port number and whatever
that business with the square brackets is.  (Should I move this
to a helper function?)

Regarding internal vs. external hosts, the proxy command can
simply run netcat locally to internal hosts, so perhaps that is
sufficient.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agodaemon: further tweaks.
Junio C Hamano [Mon, 21 Nov 2005 09:21:18 +0000 (01:21 -0800)]
daemon: further tweaks.

 - Do validation only on canonicalized paths
 - Run upload-pack with "." as repository argument

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-daemon support for user-relative paths.
Andreas Ericsson [Thu, 17 Nov 2005 19:37:14 +0000 (20:37 +0100)]
git-daemon support for user-relative paths.

Dropped a fair amount of reundant code in favour of the library code
in path.c

Added option --strict-paths with documentation, with backwards
compatibility for whitelist entries with symlinks.

Everything that worked earlier still works insofar as I have
remembered testing it.

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agotutorial: setting up a tree for subsystem maintainers
Junio C Hamano [Mon, 21 Nov 2005 21:42:55 +0000 (13:42 -0800)]
tutorial: setting up a tree for subsystem maintainers

The "copying over packs" step is to prevent the objects
available in upstream repository to get expanted in the
subsystem maintainer tree, and is still valid if the upstream
repository do not live on the same machine.  But if they are on
the same machine using objects/info/alternates is cleaner.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agorename detection with -M100 means "exact renames only".
Junio C Hamano [Mon, 21 Nov 2005 20:18:23 +0000 (12:18 -0800)]
rename detection with -M100 means "exact renames only".

When the user is interested in pure renames, there is no point
doing the similarity scores.  This changes the score argument
parsing to special case -M100 (otherwise, it is a precision
scaled value 0 <= v < 1 and would mean 0.1, not 1.0 --- if you
do mean 0.1, you can say -M1), and optimizes the diffcore_rename
transformation to only look at pure renames in that case.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoformat-patch: fix two-argument special case, and make it easier to pick single commits
Junio C Hamano [Mon, 21 Nov 2005 20:17:22 +0000 (12:17 -0800)]
format-patch: fix two-argument special case, and make it easier to pick single commits

Luben Tuikov noticed that sometimes being able to say
'git-format-patch <commit>' to format the change a single commit
introduces relative to its parent is handy.

This patch does not support that directly, but it makes sense to
interpret a single argument "rev" to mean "rev^1..rev".

With this, the backward compatibility syntaxes still apply:

 - "format-patch master" means "format-patch master..HEAD"
 - "format-patch origin master" means "format-patch origin..master"
 - "format-patch origin.." means "format-patch origin..HEAD"

But "format-patch a b c d e" formats the changes these five
commits introduce relative to their respective parents.  Earlier
it rejected these arguments not in "one..two" form.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix hooks/update template.
Junio C Hamano [Mon, 21 Nov 2005 18:59:31 +0000 (10:59 -0800)]
Fix hooks/update template.

Make the example address RFC2606 (aka BCP0032) compliant.  Also
fix a couple of shell script errors.

Noted and fixed by Matthew Wilcox and Andreas Ericsson.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMake sure heads/foo and tags/foo do not confuse things.
Junio C Hamano [Mon, 21 Nov 2005 08:43:12 +0000 (00:43 -0800)]
Make sure heads/foo and tags/foo do not confuse things.

When both heads/foo and tags/foo exist, get_sha1_basic("foo")
picked up the tag without complaining, which is quite confusing.
Make sure we require unambiguous form, "heads/foo" or "tags/foo"
in such cases.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix "do not DWIM" patch to enter_repo"
Junio C Hamano [Mon, 21 Nov 2005 07:37:13 +0000 (23:37 -0800)]
Fix "do not DWIM" patch to enter_repo"

We wanted --strict to mean "do not DWIM", but the code required to
see absolute path.  daemon does its own path verification and chdirs
to the verified repository, so enter_repo() called from upload-pack
will always enter ".".  Requiring absolute path does not make any sense.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-reset.txt: Small fix + clarifications.
Andreas Ericsson [Sun, 20 Nov 2005 21:42:11 +0000 (22:42 +0100)]
git-reset.txt: Small fix + clarifications.

This basically translates the man-page from 'git-developerish' to plain
english, adding some almost-sample output from git-status so users can
recognize what will happen.

Also mention explicitly that --mixed updates the index, while --soft
doesn't. I understood the old text to mean "--mixed is exactly like
--soft, but verbose".

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoAdd Python version checks to the Makefile to automatically set WITH_OWN_SUBPROCESS_PY
Ryan Anderson [Mon, 21 Nov 2005 05:11:22 +0000 (00:11 -0500)]
Add Python version checks to the Makefile to automatically set WITH_OWN_SUBPROCESS_PY

Also rearrange some path settings in the Makefile in the process.

Signed-off-by: Ryan Anderson <ryan@michonline.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix sparse warnings
Timo Hirvonen [Mon, 21 Nov 2005 00:52:52 +0000 (02:52 +0200)]
Fix sparse warnings

Make some functions static and convert func() function prototypes to to
func(void).  Fix declaration after statement, missing declaration and
redundant declaration warnings.

Signed-off-by: Timo Hirvonen <tihirvon@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogitk: Disable fastdate stuff for now
Paul Mackerras [Sun, 20 Nov 2005 22:56:07 +0000 (09:56 +1100)]
gitk: Disable fastdate stuff for now

It has a fatal flaw in that it only handles timezones that are a
multiple of an hour.  It's really only needed with Tk8.5, where
the clock format command has been reimplemented in Tcl and is much
slower than in Tk8.4.

Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agoMerge http://www.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Sun, 20 Nov 2005 20:18:13 +0000 (12:18 -0800)]
Merge http://www.kernel.org/pub/scm/gitk/gitk

19 years agogit-config-set: support selecting values by non-matching regex
Johannes Schindelin [Sun, 20 Nov 2005 12:24:18 +0000 (13:24 +0100)]
git-config-set: support selecting values by non-matching regex

Extend the regex syntax of value_regex so that prepending an exclamation
mark means non-match:

[core]
quetzal = "Dodo" for Brainf*ck
quetzal = "T. Rex" for Malbolge
quetzal = "cat"

You can match the third line with

git-config-set --get quetzal '! for '

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agomerge-recursive: Replace 'except:'
Fredrik Kuivinen [Sun, 20 Nov 2005 12:14:37 +0000 (13:14 +0100)]
merge-recursive: Replace 'except:'

Plain except:s are evil as they will catch all kinds of exceptions
including NameError and AttrubiteError.

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agomerge-one-file: use rmdir -p
Junio C Hamano [Sun, 20 Nov 2005 18:40:31 +0000 (10:40 -0800)]
merge-one-file: use rmdir -p

The flag is universally available, even on VMS; use it.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogitk: Fix some bugs introduced by speedup changes
Paul Mackerras [Sun, 20 Nov 2005 12:08:22 +0000 (23:08 +1100)]
gitk: Fix some bugs introduced by speedup changes

Commits that weren't read from git-rev-list, i.e. the ones displayed
with an open circle, were displayed incorrectly: the headline was
null if there was only one line, and the commit comment was put all
on one line.  Also, the terminal commits weren't displayed when -r
was used.

Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agoDocumentation: add hooks/update example.
Junio C Hamano [Sun, 20 Nov 2005 07:50:48 +0000 (23:50 -0800)]
Documentation: add hooks/update example.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-config-set: add more options
Johannes Schindelin [Sun, 20 Nov 2005 05:52:22 +0000 (06:52 +0100)]
git-config-set: add more options

... namely

--replace-all, to replace any amount of matching lines, not just 0 or 1,
--get, to get the value of one key,
--get-all, the multivar version of --get, and
--unset-all, which deletes all matching lines from .git/config

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoDocumentation update for user-relative paths.
Andreas Ericsson [Thu, 17 Nov 2005 19:37:14 +0000 (20:37 +0100)]
Documentation update for user-relative paths.

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoClient side support for user-relative paths.
Andreas Ericsson [Thu, 17 Nov 2005 19:37:14 +0000 (20:37 +0100)]
Client side support for user-relative paths.

With this patch, the client side passes identical paths for these two:
ssh://host.xz/~junio/repo
host.xz:~junio/repo

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoServer-side support for user-relative paths.
Andreas Ericsson [Thu, 17 Nov 2005 19:37:14 +0000 (20:37 +0100)]
Server-side support for user-relative paths.

This patch basically just removes the redundant code from
{receive,upload}-pack.c in favour of the library code in path.c.

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>