Code

git.git
18 years agopacked objects: minor cleanup
Junio C Hamano [Wed, 15 Feb 2006 20:47:43 +0000 (12:47 -0800)]
packed objects: minor cleanup

The delta depth is unsigned.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agols-files --error-unmatch pathspec error reporting fix.
Junio C Hamano [Wed, 15 Feb 2006 09:10:13 +0000 (01:10 -0800)]
ls-files --error-unmatch pathspec error reporting fix.

Earlier patch mistakenly used prefix_len when it meant
prefix_offset.  The latter is to strip the leading directories
when run from a subdirectory.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'kh/svn'
Junio C Hamano [Wed, 15 Feb 2006 01:51:50 +0000 (17:51 -0800)]
Merge branch 'kh/svn'

* kh/svn:
  git-svnimport: -r adds svn revision number to commit messages

18 years agoMerge branch 'jc/commit'
Junio C Hamano [Wed, 15 Feb 2006 01:51:02 +0000 (17:51 -0800)]
Merge branch 'jc/commit'

* jc/commit:
  commit: detect misspelled pathspec while making a partial commit.
  combine-diff: diff-files fix (#2)
  combine-diff: diff-files fix.

18 years agoMerge branch 'jc/rebase'
Junio C Hamano [Wed, 15 Feb 2006 01:49:00 +0000 (17:49 -0800)]
Merge branch 'jc/rebase'

* jc/rebase:
  rebase: allow a hook to refuse rebasing.

18 years agoMerge branch 'ra/email'
Junio C Hamano [Wed, 15 Feb 2006 01:46:41 +0000 (17:46 -0800)]
Merge branch 'ra/email'

* ra/email:
  send-email: Add --cc
  send-email: Add some options for controlling how addresses are automatically added to the cc: list.

18 years agocommit: detect misspelled pathspec while making a partial commit.
Junio C Hamano [Tue, 14 Feb 2006 20:40:20 +0000 (12:40 -0800)]
commit: detect misspelled pathspec while making a partial commit.

When you say "git commit Documentaiton" to make partial commit
for the files only in that directory, we did not detect that as
a misspelled pathname and attempted to commit index without
change.  If nothing matched, there is no harm done, but if the
index gets modified otherwise by having another valid pathspec
or after an explicit update-index, a user will not notice
without paying attention to the "git status" preview.

This introduces --error-unmatch option to ls-files, and uses it
to detect this common user error.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-svnimport: -r adds svn revision number to commit messages
Karl Hasselström [Tue, 14 Feb 2006 02:43:34 +0000 (03:43 +0100)]
git-svnimport: -r adds svn revision number to commit messages

New -r flag for prepending the corresponding Subversion revision
number to each commit message.

Signed-off-by: Karl Hasselström <kha@treskal.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocombine-diff: diff-files fix (#2)
Junio C Hamano [Tue, 14 Feb 2006 09:11:42 +0000 (01:11 -0800)]
combine-diff: diff-files fix (#2)

The raw format "git-diff-files -c" to show unmerged state forgot
to initialize the status fields from parents, causing NUL
characters to be emitted.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge some proposed fixes
Junio C Hamano [Tue, 14 Feb 2006 07:34:58 +0000 (23:34 -0800)]
Merge some proposed fixes

Conflicts:

Documentation/git-commit.txt - taking the post 1.2.0 semantics.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'pb/bisect'
Junio C Hamano [Tue, 14 Feb 2006 07:26:53 +0000 (23:26 -0800)]
Merge branch 'pb/bisect'

* pb/bisect:
  Properly git-bisect reset after bisecting from non-master head

18 years agocombine-diff: diff-files fix.
Junio C Hamano [Tue, 14 Feb 2006 07:07:04 +0000 (23:07 -0800)]
combine-diff: diff-files fix.

When showing a conflicted merge from index stages and working
tree file, we did not fetch the mode from the working tree,
and mistook that as a deleted file.  Also if the manual
resolution (or automated resolution by git rerere) ended up
taking either parent's version, we did not show _anything_ for
that path.  Either was quite bad and confusing.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agos/SHELL/SHELL_PATH/ in Makefile
Fredrik Kuivinen [Mon, 13 Feb 2006 23:15:14 +0000 (00:15 +0100)]
s/SHELL/SHELL_PATH/ in Makefile

With the current Makefile we don't use the shell chosen by the
platform specific defines when we invoke GIT-VERSION-GEN.

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agobisect: remove BISECT_NAMES after done.
Junio C Hamano [Tue, 14 Feb 2006 05:25:38 +0000 (21:25 -0800)]
bisect: remove BISECT_NAMES after done.

I noticed that we forgot to clean this file and kept it that
way, while trying to help with Andrew's bisect problem.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDocumentation: git-ls-files asciidocco.
Junio C Hamano [Tue, 14 Feb 2006 05:52:10 +0000 (21:52 -0800)]
Documentation: git-ls-files asciidocco.

Noticed by Jon Nelson.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agosend-email: Add --cc
Ryan Anderson [Mon, 13 Feb 2006 08:05:15 +0000 (03:05 -0500)]
send-email: Add --cc

Since Junio used this in an example, and I've personally tried to use it, I
suppose the option should actually exist.

Signed-off-by: Ryan Anderson <ryan@michonline.com>
18 years agoDocumentation: git-commit in 1.2.X series defaults to --include.
Junio C Hamano [Mon, 13 Feb 2006 08:26:14 +0000 (00:26 -0800)]
Documentation: git-commit in 1.2.X series defaults to --include.

The documentation was mistakenly describing the --only semantics to
be default.  The 1.2.0 release and its maintenance series 1.2.X will
keep the traditional --include semantics as the default.  Clarify the
situation.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agosend-email: Add some options for controlling how addresses are automatically added...
Ryan Anderson [Mon, 13 Feb 2006 07:57:09 +0000 (02:57 -0500)]
send-email: Add some options for controlling how addresses are automatically added to the cc: list.

Signed-off-by: Ryan Anderson <ryan@michonline.com>
18 years agorebase: allow a hook to refuse rebasing.
Junio C Hamano [Mon, 13 Feb 2006 07:17:04 +0000 (23:17 -0800)]
rebase: allow a hook to refuse rebasing.

This lets a hook to interfere a rebase and help prevent certain
branches from being rebased by mistake.  A sample hook to show
how to prevent a topic branch that has already been merged into
publish branch.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-commit: Now --only semantics is the default.
Junio C Hamano [Mon, 13 Feb 2006 07:55:07 +0000 (23:55 -0800)]
git-commit: Now --only semantics is the default.

This changes the "git commit paths..." to default to --only
semantics from traditional --include semantics, as agreed on the
list.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoGIT 1.2.0 v1.2.0
Junio C Hamano [Sun, 12 Feb 2006 21:14:53 +0000 (13:14 -0800)]
GIT 1.2.0

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoFix "test: unexpected operator" on bsd
Junio C Hamano [Sun, 12 Feb 2006 21:13:12 +0000 (13:13 -0800)]
Fix "test: unexpected operator" on bsd

This fixes the same issue as a previous fix by Alex Riesen does.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoProperly git-bisect reset after bisecting from non-master head
Petr Baudis [Sun, 12 Feb 2006 16:06:14 +0000 (17:06 +0100)]
Properly git-bisect reset after bisecting from non-master head

git-bisect reset without an argument would return to master even
if the bisecting started at a non-master branch. This patch makes
it save the original branch name to .git/head-name and restore it
afterwards.

This is also compatible with Cogito and cg-seek, so cg-status will
show that we are seeked on the bisect branch and cg-reset will
properly restore the original branch.

git-bisect start will refuse to work if it is not on a bisect but
.git/head-name exists; this is to protect against conflicts with
other seeking tools.

Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-commit: show dirtiness including index.
Junio C Hamano [Sun, 12 Feb 2006 21:05:53 +0000 (13:05 -0800)]
git-commit: show dirtiness including index.

Earlier, when we switched a branch we used diff-files to show
paths that are dirty in the working tree.  But we allow switching
branches with updated index ("read-tree -m -u $old $new" works that
way), and only showing paths that have differences in the working
tree but not paths that are different in index was confusing.

This shows both as modified from the top commit of the branch we
just have switched to.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMake pack-objects chattier.
Junio C Hamano [Sun, 12 Feb 2006 21:01:54 +0000 (13:01 -0800)]
Make pack-objects chattier.

You could give -q to squelch it, but currently no tool does it.
This would make 'git clone host:repo here' over ssh not silent
again.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoavoid echo -e, there are systems where it does not work
Alex Riesen [Sun, 12 Feb 2006 18:05:34 +0000 (19:05 +0100)]
avoid echo -e, there are systems where it does not work

FreeBSD 4.11 being one example: the built-in echo doesn't have -e,
and the installed /bin/echo does not do "-e" as well.
"printf" works, laking just "\e" and "\xAB'.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agofix "test: 2: unexpected operator" on bsd
Alex Riesen [Sun, 12 Feb 2006 18:03:16 +0000 (19:03 +0100)]
fix "test: 2: unexpected operator" on bsd

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoFix object re-hashing
Linus Torvalds [Sun, 12 Feb 2006 19:24:50 +0000 (11:24 -0800)]
Fix object re-hashing

The hashed object lookup had a subtle bug in re-hashing: it did

for (i = 0; i < count; i++)
if (objs[i]) {
.. rehash ..

where "count" was the old hash couny. Oon the face of it is obvious, since
it clearly re-hashes all the old objects.

However, it's wrong.

If the last old hash entry before re-hashing was in use (or became in use
by the re-hashing), then when re-hashing could have inserted an object
into the hash entries with idx >= count due to overflow. When we then
rehash the last old entry, that old entry might become empty, which means
that the overflow entries should be re-hashed again.

In other words, the loop has to be fixed to either traverse the whole
array, rather than just the old count.

(There's room for a slight optimization: instead of counting all the way
up, we can break when we see the first empty slot that is above the old
"count". At that point we know we don't have any collissions that we might
have to fix up any more. This patch only does the trivial fix)

[jc: with trivial fix on trivial fix]

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohashtable-based objects: minimum fixups.
Junio C Hamano [Sun, 12 Feb 2006 02:51:19 +0000 (18:51 -0800)]
hashtable-based objects: minimum fixups.

Calling hashtable_index from find_object before objs is created
would result in division by zero failure.  Avoid it.

Also the given object name may not be aligned suitably for
unsigned int; avoid dereferencing casted pointer.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoUse a hashtable for objects instead of a sorted list
Johannes Schindelin [Sun, 12 Feb 2006 01:57:57 +0000 (02:57 +0100)]
Use a hashtable for objects instead of a sorted list

In a simple test, this brings down the CPU time from 47 sec to 22 sec.

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoAdd howto about separating topics.
kent@lysator.liu.se [Sun, 12 Feb 2006 12:00:52 +0000 (13:00 +0100)]
Add howto about separating topics.

This howto consists of a footnote from an email by JC to the git
mailing list (<7vfyms0x4p.fsf@assigned-by-dhcp.cox.net>).

Signed-off-by: Kent Engstrom <kent@lysator.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'pb/repo'
Junio C Hamano [Sun, 12 Feb 2006 13:02:30 +0000 (05:02 -0800)]
Merge branch 'pb/repo'

* pb/repo:
  Add support for explicit type specifiers when calling git-repo-config

18 years agoMerge branch 'jc/fixdiff'
Junio C Hamano [Sun, 12 Feb 2006 13:02:25 +0000 (05:02 -0800)]
Merge branch 'jc/fixdiff'

* jc/fixdiff:
  diff-tree: do not default to -c

18 years agoAvoid using "git-var -l" until it gets fixed.
Junio C Hamano [Sat, 11 Feb 2006 20:39:11 +0000 (12:39 -0800)]
Avoid using "git-var -l" until it gets fixed.

This is to be nicer to people with unusable GECOS field.

"git-var -l" is currently broken in that when used by a user who
does not have a usable GECOS field and has not corrected it by
exporting GIT_COMMITTER_NAME environment variable it dies when
it tries to output GIT_COMMITTER_IDENT (same thing for AUTHOR).

"git-pull" used "git-var -l" only because it needed to get a
configuration variable before "git-repo-config --get" was
introduced.  Use the latter tool designed exactly for this
purpose.

"git-sh-setup" used "git-var GIT_AUTHOR_IDENT" without actually
wanting to use its value.  The only purpose was to cause the
command to check and barf if the repository format version
recorded in the $GIT_DIR/config file is too new for us to deal
with correctly.  Instead, use "repo-config --get" on a random
property and see if it die()s, and check if the exit status is
128 (comes from die -- missing variable is reported with exit
status 1, so we can tell that case apart).

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoAdd support for explicit type specifiers when calling git-repo-config
Petr Baudis [Sun, 12 Feb 2006 03:14:48 +0000 (04:14 +0100)]
Add support for explicit type specifiers when calling git-repo-config

Currently, git-repo-config will just return the raw value of option
as specified in the config file; this makes things difficult for scripts
calling it, especially if the value is supposed to be boolean.

This patch makes it possible to ask git-repo-config to check if the option
is of the given type (int or bool) and write out the value in its
canonical form. If you do not pass --int or --bool, the behaviour stays
unchanged and the raw value is emitted.

This also incidentally fixes the segfault when option with no value is
encountered.

[jc: tweaked the option parsing a bit to make it easier to see
 that the patch does not change anything but the type stuff in
 the diff output.  Also changed to avoid "foo ? : bar" construct. ]

Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agodiff-tree: do not default to -c
Junio C Hamano [Sun, 12 Feb 2006 00:43:30 +0000 (16:43 -0800)]
diff-tree: do not default to -c

Marco says it breaks qgit.  This makes the flags a bit more
orthogonal.

  $ git-diff-tree -r --abbrev ca18

    No output from this command because you asked to skip merge by
    not having -m there.

  $ git-diff-tree -r -m --abbrev ca18
  ca182053c7710a286d72102f4576cf32e0dafcfb
  :100644 100644 538d21d... 59042d1... M Makefile
  :100644 100644 410b758... 6c47c3a... M entry.c
  ca182053c7710a286d72102f4576cf32e0dafcfb
  :100644 100644 30479b4... 59042d1... M Makefile

    The same "independent sets of diff" as before without -c.

  $ git-diff-tree -r -m -c --abbrev ca18
  ca182053c7710a286d72102f4576cf32e0dafcfb
  ::100644 100644 100644 538d21d... 30479b4... 59042d1... MM Makefile

    Combined.

  $ git-diff-tree -r -c --abbrev ca18
  ca182053c7710a286d72102f4576cf32e0dafcfb
  ::100644 100644 100644 538d21d... 30479b4... 59042d1... MM Makefile

    Asking for combined without -m does not make sense, so -c
    implies -m.

We need to supply -c as default to whatchanged, which is a
one-liner.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agot5500: adjust to change in pack-object reporting behaviour.
Junio C Hamano [Sun, 12 Feb 2006 07:08:23 +0000 (23:08 -0800)]
t5500: adjust to change in pack-object reporting behaviour.

Now pack-object is not as chatty when its stderr is not connected
to a terminal, so the test needs to be adjusted for that.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoOnly call git-rerere if $GIT_DIR/rr-cache exists.
Junio C Hamano [Sun, 12 Feb 2006 02:55:43 +0000 (18:55 -0800)]
Only call git-rerere if $GIT_DIR/rr-cache exists.

Johannes noticed that git-rerere depends on Digest.pm, and if
one does not use the command, one can live without it.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoUse a relative path for SVN importing
Christian Biesinger [Sat, 11 Feb 2006 15:44:11 +0000 (16:44 +0100)]
Use a relative path for SVN importing

The absolute path (with the leading slash) breaks SVN importing,
because it then looks for /trunk/... instead of /svn/trunk/...
(in my case, the repository URL was https://servername/svn/)

Signed-off-by: Christian Biesinger <cbiesinger@web.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agofetch-clone progress: finishing touches.
Junio C Hamano [Sun, 12 Feb 2006 01:54:18 +0000 (17:54 -0800)]
fetch-clone progress: finishing touches.

This makes fetch-pack also report the progress of packing part.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoFix fetch-clone in the presense of signals
Linus Torvalds [Sat, 11 Feb 2006 18:41:22 +0000 (10:41 -0800)]
Fix fetch-clone in the presense of signals

We shouldn't fail a fetch just because a signal might have interrupted
the read.

Normally, we don't install any signal handlers, so EINTR really shouldn't
happen. That said, really old versions of Linux will interrupt an
interruptible system call even for signals that turn out to be ignored
(SIGWINCH is the classic example - resizing your xterm would cause it).
The same might well be true elsewhere too.

Also, since receive_keep_pack() doesn't control the caller, it can't know
that no signal handlers exist.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMake "git clone" pack-fetching download statistics better
Linus Torvalds [Sat, 11 Feb 2006 18:43:56 +0000 (10:43 -0800)]
Make "git clone" pack-fetching download statistics better

Average it out over a few events to make the numbers stable, and fix the
silly usec->binary-ms conversion.

Yeah, yeah, it's arguably eye-candy to keep the user calm, but let's do
that right.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMake "git clone" less of a deathly quiet experience
Linus Torvalds [Sat, 11 Feb 2006 04:31:09 +0000 (20:31 -0800)]
Make "git clone" less of a deathly quiet experience

It used to be that "git-unpack-objects" would give nice percentages, but
now that we don't unpack the initial clone pack any more, it doesn't. And
I'd love to do that nice percentage view in the pack objects downloader
too, but the thing doesn't even read the pack header, much less know how
much it's going to get, so I was lazy and didn't.

Instead, it at least prints out how much data it's gotten, and what the
packing speed is. Which makes the user realize that it's actually doing
something useful instead of sitting there silently (and if the recipient
knows how large the final result is, he can at least make a guess about
when it migt be done).

So with this patch, I get something like this on my DSL line:

[torvalds@g5 ~]$ time git clone master.kernel.org:/pub/scm/linux/kernel/git/torvalds/linux-2.6 clone-test
Packing 188543 objects
  48.398MB  (154 kB/s)

where even the speed approximation seems to be roughtly correct (even
though my algorithm is a truly stupid one, and only really gives "speed in
the last half second or so").

Anyway, _something_ like this is definitely needed. It could certainly be
better (if it showed the same kind of thing that git-unpack-objects did,
that would be much nicer, but would require parsing the object stream as
it comes in). But this is  big step forward, I think.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDefine GIT_(AUTHOR|COMMITTER)_(NAME|EMAIL) to known values.
Junio C Hamano [Sat, 11 Feb 2006 03:11:23 +0000 (19:11 -0800)]
Define GIT_(AUTHOR|COMMITTER)_(NAME|EMAIL) to known values.

Without these, running tests with an account with empty gecos
field would fail.

We might want to loosen error from "git-var -l" (but not
"git-var GIT_AUTHOR_NAME") later, but that is more or less an
independent issue.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'lt/diff-tree'
Junio C Hamano [Sat, 11 Feb 2006 02:47:41 +0000 (18:47 -0800)]
Merge branch 'lt/diff-tree'

* lt/diff-tree:
  combine-diff: Record diff status a bit more faithfully
  find_unique_abbrev() simplification.
  combine-diff: move formatting logic to show_combined_diff()
  combined-diff: use diffcore before intersecting paths.
  diff-tree -c raw output

18 years agogit-commit -v: have patch at the end.
Junio C Hamano [Sat, 11 Feb 2006 02:38:24 +0000 (18:38 -0800)]
git-commit -v: have patch at the end.

It was pointed out that otherwise more important summary
information prefixed with '#' would become prone to be missed.

Also instead of chopping at the first '^---$' line, stop at the
first 'diff --git a/' line.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agorev-list: default to abbreviate merge parent names under --pretty.
Junio C Hamano [Fri, 10 Feb 2006 19:56:42 +0000 (11:56 -0800)]
rev-list: default to abbreviate merge parent names under --pretty.

When we prettyprint commit log messages, merge parent names were
often very long and there was no way to abbreviate it.

This changes them to be abbreviated by default, and non-default
abbreviations can be specified with --no-abbrev or --abbrev=<n>
options.

Note that this affects only the prettyprinted parent names.  The
output from --show-parents is meant for machine consumption and
is not affected by this flag.

18 years agodelta micro optimization
Nicolas Pitre [Fri, 10 Feb 2006 18:42:05 +0000 (13:42 -0500)]
delta micro optimization

My kernel work habit made me look at the generated assembly for the
delta code, and one obvious albeit small improvement is this patch.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocount-delta.c: comment fixes
Nicolas Pitre [Fri, 10 Feb 2006 15:20:40 +0000 (10:20 -0500)]
count-delta.c: comment fixes

There was a stale comment that explains why the old code could
undercount when delta data copied things around inside detination
buffer.  We do not use that kind of delta, so the comment does
not apply.

18 years agoMerge branch 'jc/empty-commit'
Junio C Hamano [Fri, 10 Feb 2006 15:14:55 +0000 (07:14 -0800)]
Merge branch 'jc/empty-commit'

* jc/empty-commit:
  t6000: fix a careless test library add-on.
  Do not allow empty name or email.

18 years agocombine-diff: Record diff status a bit more faithfully
Junio C Hamano [Fri, 10 Feb 2006 10:30:52 +0000 (02:30 -0800)]
combine-diff: Record diff status a bit more faithfully

This shows "new file mode XXXX" and "deleted file mode XXXX"
lines like two-way diff-patch output does, by checking the
status from each parent.

The diff-raw output for combined diff is made a bit uglier by
showing diff status letters with each parent.  While most of the
case you would see "MM" in the output, an Evil Merge that
touches a path that was added by inheriting from one parent is
possible and it would be shown like these:

    $ git-diff-tree --abbrev -c HEAD
    2d7ca89675eb8888b0b88a91102f096d4471f09f
    ::000000 000000 100644 0000000... 0000000... 31dd686... AA b
    ::000000 100644 100644 0000000... 6c884ae... c6d4fa8... AM d
    ::100644 100644 100644 4f7cbe7... f8c295c... 19d5d80... RR e

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agofind_unique_abbrev() simplification.
Junio C Hamano [Fri, 10 Feb 2006 09:51:12 +0000 (01:51 -0800)]
find_unique_abbrev() simplification.

Earlier it did not grok the 0{40} SHA1 very well, but what it
needed to do was to find the shortest 0{N} that is not used as a
valid object name to be consistent with the way names of valid
objects are abbreviated.  This makes some users simpler.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-status -v
Junio C Hamano [Fri, 10 Feb 2006 08:45:59 +0000 (00:45 -0800)]
git-status -v

This revamps the git-status command to take the same set of
parameters as git commit.  It gives a preview of what is being
committed with that command.  With -v flag, it shows the diff
output between the HEAD commit and the index that would be
committed if these flags were given to git-commit command.

git-commit also acquires -v flag (it used to mean "verify" but
that is the default anyway and there is --no-verify to turn it
off, so not much is lost), which uses the updated git-status -v
to seed the commit log buffer.  This is handy for writing a log
message while reviewing the changes one last time.

Now, git-commit and git-status are internally share the same
implementation.

Unlike previous git-commit change, this uses a temporary index
to prepare the index file that would become the real index file
after a successful commit, and moves it to the real index file
once the commit is actually made.  This makes it safer than the
previous scheme, which stashed away the original index file and
restored it after an aborted commit.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/ls-files-o'
Junio C Hamano [Fri, 10 Feb 2006 06:19:07 +0000 (22:19 -0800)]
Merge branch 'jc/ls-files-o'

* jc/ls-files-o:
  ls-files: honour per-directory ignore file from higher directories.

18 years agocount-delta.c: Match the delta data semantics change in version 3.
Junio C Hamano [Fri, 10 Feb 2006 01:15:59 +0000 (17:15 -0800)]
count-delta.c: Match the delta data semantics change in version 3.

This matches the count_delta() logic to the change previous
commit introduces to patch_delta().

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoremove delta-against-self bit
Nicolas Pitre [Thu, 9 Feb 2006 22:50:04 +0000 (17:50 -0500)]
remove delta-against-self bit

After experimenting with code to add the ability to encode a delta
against part of the deltified file, it turns out that resulting packs
are _bigger_ than when this ability is not used.  The raw delta output
might be smaller, but it doesn't compress as well using gzip with a
negative net saving on average.

Said bit would in fact be more useful to allow for encoding the copying
of chunks larger than 64KB providing more savings with large files.
This will correspond to packs version 3.

While the current code still produces packs version 2, it is made future
proof so pack versions 2 and 3 are accepted.  Any pack version 2 are
compatible with version 3 since the redefined bit was never used before.
When enough time has passed, code to use that bit to produce version 3
packs could be added.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agostat() for existence in safe_create_leading_directories()
Jason Riedy [Fri, 10 Feb 2006 01:56:13 +0000 (17:56 -0800)]
stat() for existence in safe_create_leading_directories()

Use stat() to explicitly check for existence rather than
relying on the non-portable EEXIST error in sha1_file.c's
safe_create_leading_directories().  There certainly are
optimizations possible, but then the code becomes almost
the same as that in coreutil's lib/mkdir-p.c.

Other uses of EEXIST seem ok.  Tested on Solaris 8, AIX 5.2L,
and a few Linux versions.  AIX has some unrelated (I think)
failures right now; I haven't tried many recent gits there.
Anyone have an old Ultrix box to break everything?  ;)

Also remove extraneous #includes.  Everything's already in
git-compat-util.h, included through cache.h.

Signed-off-by: Jason Riedy <ejr@cs.berkeley.edu>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocombine-diff: move formatting logic to show_combined_diff()
Junio C Hamano [Thu, 9 Feb 2006 23:23:06 +0000 (15:23 -0800)]
combine-diff: move formatting logic to show_combined_diff()

This way, diff-files can make use of it.  Also implement the
full suite of what diff_flush_raw() supports just for
consistency.  With this, 'diff-tree -c -r --name-status' would
show what is expected.

There is no way to get the historical output (useful for
debugging and low-level Plumbing work) anymore, so tentatively
it makes '-m' to mean "do not combine and show individual diffs
with parents".

diff-files matches diff-tree to produce raw output for -c.  For
textual combined diff, use -p -c.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocall git_config() after setup_git_directory()
Junio C Hamano [Thu, 9 Feb 2006 22:41:39 +0000 (14:41 -0800)]
call git_config() after setup_git_directory()

If you call setup_git_directory() to work from a subdirectory,
that should be run first before running git_config().  Otherwise
you would not read the configuration file from the correct place.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocombined-diff: use diffcore before intersecting paths.
Junio C Hamano [Thu, 9 Feb 2006 22:35:19 +0000 (14:35 -0800)]
combined-diff: use diffcore before intersecting paths.

This is needed to make "diff-tree -c -M" to work semi-sensibly.
Otherwise rename detection, pickaxe and friends would never be
invoked.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoAdd --diff-filter= documentation paragraph
Jon Loeliger [Thu, 9 Feb 2006 15:12:11 +0000 (09:12 -0600)]
Add --diff-filter= documentation paragraph

Signed-off-by: Jon Loeliger <jdl@jdl.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agodiff-tree -c raw output
Linus Torvalds [Thu, 9 Feb 2006 18:30:28 +0000 (10:30 -0800)]
diff-tree -c raw output

NOTE! This makes "-c" be the default, which effectively means that merges
are never ignored any more, and "-m" is a no-op. So it changes semantics.

I would also like to make "--cc" the default if you do patches, but didn't
actually do that.

The raw output format is not wonderfully pretty, but it's distinguishable
from a "normal patch" in that a normal patch with just one parent has just
one colon at the beginning, while a multi-parent raw diff has <n> colons
for <n> parents.

So now, in the kernel, when you do

git-diff-tree cce0cac125623f9b68f25dd1350f6d616220a8dd

(to see the manual ARM merge that had a conflict in arch/arm/Kconfig), you
get

cce0cac125623f9b68f25dd1350f6d616220a8dd
::100644 100644 100644 4a63a8e2e45247a11c068c6ed66c6e7aba29ddd9 77eee38762d69d3de95ae45dd9278df9b8225e2c 2f61726d2f4b636f6e66696700dbf71a59dad287       arch/arm/Kconfig

ie you see two colons (two parents), then three modes (parent modes
followed by result mode), then three sha1s (parent sha1s followed by
result sha1).

Which is pretty close to the normal raw diff output.

Cool/stupid exercise:

$ git-whatchanged | grep '^::' | cut -f2- | sort |
  uniq -c | sort -n | less -S

will show which files have needed the most file-level merge conflict
resolution. Useful? Probably not. But kind of interesting.

For the kernel, it's

     ....
     10 arch/ia64/Kconfig
     11 drivers/scsi/Kconfig
     12 drivers/net/Makefile
     17 include/linux/libata.h
     18 include/linux/pci_ids.h
     23 drivers/net/Kconfig
     24 drivers/scsi/libata-scsi.c
     28 drivers/scsi/libata-core.c
     43 MAINTAINERS

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agols-files: honour per-directory ignore file from higher directories.
Junio C Hamano [Thu, 9 Feb 2006 08:08:31 +0000 (00:08 -0800)]
ls-files: honour per-directory ignore file from higher directories.

When git-ls-files -o --exclude-per-directory=.gitignore is run
from a subdirectory, it did not read from .gitignore from its
parent directory.  Reading from them makes output from these two
commands consistent:

    $ git ls-files -o --exclude-per-directory=.gitignore Documentation
    $ cd Documentation &&
      git ls-files -o --exclude-per-directory=.gitignore

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agot6000: fix a careless test library add-on.
Junio C Hamano [Tue, 7 Feb 2006 23:35:46 +0000 (15:35 -0800)]
t6000: fix a careless test library add-on.

It tried to "restore" GIT_AUTHOR_EMAIL environment variable but
the variable started out as unset, so ended up setting it to an
empty string.  This is now caught as an error.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDo not allow empty name or email.
Junio C Hamano [Tue, 7 Feb 2006 21:19:10 +0000 (13:19 -0800)]
Do not allow empty name or email.

Instead of silently allowing to create a bogus commit that lacks
information by mistake, complain loudly and die.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years ago.gitignore git-rerere and config.mak
Andreas Ericsson [Tue, 7 Feb 2006 17:21:02 +0000 (18:21 +0100)]
.gitignore git-rerere and config.mak

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoFix "git diff a..b" breakage
Linus Torvalds [Tue, 7 Feb 2006 18:26:41 +0000 (10:26 -0800)]
Fix "git diff a..b" breakage

The "--cc" implies "-p", but without the recursive part.

Linus

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoBasic documentation for git-show
Petr Baudis [Tue, 7 Feb 2006 11:47:47 +0000 (12:47 +0100)]
Basic documentation for git-show

Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDocument git-diff-tree --always
Petr Baudis [Tue, 7 Feb 2006 11:47:44 +0000 (12:47 +0100)]
Document git-diff-tree --always

Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp-fetch: Abort requests for objects which arrived in packs
Mark Wooding [Tue, 7 Feb 2006 10:07:39 +0000 (10:07 +0000)]
http-fetch: Abort requests for objects which arrived in packs

In fetch_object, there's a call to release an object request if the
object mysteriously arrived, say in a pack.  Unfortunately, the fetch
attempt for this object might already be in progress, and we'll leak the
descriptor.  Instead, try to tidy away the request.

Signed-off-by: Mark Wooding <mdw@distorted.org.uk>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoformat-patch: Remove last vestiges of --mbox option
Andreas Ericsson [Tue, 7 Feb 2006 09:37:54 +0000 (09:37 +0000)]
format-patch: Remove last vestiges of --mbox option

Don't mention it in docs or --help output.
Remove mbox, date and author variables from git-format-patch.sh.

Use DESCRIPTION text from man-page to update LONG_USAGE output. It's
a bit silly to have two texts saying the same thing in different words,
and I'm too lazy to update both.

Signed-off-by: Andreas Ericsson <ae@op5.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge git://git.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Tue, 7 Feb 2006 09:19:49 +0000 (01:19 -0800)]
Merge git://git.kernel.org/pub/scm/gitk/gitk

* git://git.kernel.org/pub/scm/gitk/gitk:
  gitk: Use git-diff-tree --cc for showing the diffs for merges
  gitk: Add braces around if expressions

18 years agogit-commit: finishing touches.
Junio C Hamano [Mon, 6 Feb 2006 00:08:01 +0000 (16:08 -0800)]
git-commit: finishing touches.

Introduce --only flag to allow the new "partial commit"
semantics when paths are specified.  The default is still the
traditional --include semantics.  Once peoples' fingers and
scripts that want the traditional behaviour are updated to
explicitly say --include, we could change it to either default
to --only, or refuse to operate without either --only/--include
when paths are specified.

This also fixes a couple of bugs in the previous round.  Namely:

 - forgot to save/restore index in some cases.

 - forgot to use the temporary index to show status when '--only
   paths...' semantics was used.

 - --author did not take precedence when reusing an existing
   commit.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-commit: revamp the git-commit semantics.
Junio C Hamano [Sun, 5 Feb 2006 08:07:44 +0000 (00:07 -0800)]
git-commit: revamp the git-commit semantics.

 - "git commit" without _any_ parameter keeps the traditional
   behaviour.  It commits the current index.

   We commit the whole index even when this form is run from a
   subdirectory.

 - "git commit --include paths..." (or "git commit -i paths...")
   is equivalent to:

    git update-index --remove paths...
        git commit

 - "git commit paths..." acquires a new semantics.  This is an
   incompatible change that needs user training, which I am
   still a bit reluctant to swallow, but enough people seem to
   have complained that it is confusing to them.  It

   1. refuses to run if $GIT_DIR/MERGE_HEAD exists, and reminds
      trained git users that the traditional semantics now needs
      -i flag.

   2. refuses to run if named paths... are different in HEAD and
      the index (ditto about reminding).  Added paths are OK.

   3. reads HEAD commit into a temporary index file.

   4. updates named paths... from the working tree in this
      temporary index.

   5. does the same updates of the paths... from the working
      tree to the real index.

   6. makes a commit using the temporary index that has the
      current HEAD as the parent, and updates the HEAD with this
      new commit.

 - "git commit --all" can run from a subdirectory, but it updates
   the index with all the modified files and does a whole tree
   commit.

 - In all cases, when the command decides not to create a new
   commit, the index is left as it was before the command is
   run.  This means that the two "git diff" in the following
   sequence:

       $ git diff
       $ git commit -a
       $ git diff

   would show the same diff if you abort the commit process by
   making the commit log message empty.

This commit also introduces much requested --author option.

$ git commit --author 'A U Thor <author@example.com>'

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-rerere: reuse recorded resolve.
Junio C Hamano [Sun, 29 Jan 2006 07:15:24 +0000 (23:15 -0800)]
git-rerere: reuse recorded resolve.

In a workflow that employs relatively long lived topic branches,
the developer sometimes needs to resolve the same conflict over
and over again until the topic branches are done (either merged
to the "release" branch, or sent out and accepted upstream).

This commit introduces a new command, "git rerere", to help this
process by recording the conflicted automerge results and
corresponding hand-resolve results on the initial manual merge,
and later by noticing the same conflicted automerge and applying
the previously recorded hand resolution using three-way merge.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agofmt-merge-msg: show summary of what is merged.
Junio C Hamano [Sat, 4 Feb 2006 03:20:16 +0000 (19:20 -0800)]
fmt-merge-msg: show summary of what is merged.

In addition to the branch names, populate the log message with
one-line description from actual commits that are being merged.

This was prompted by Len's 12-way octopus.  You need to have
'merge.summary' in the configuration file to enable it:

$ git repo-config merge.summary yes

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoread-tree --aggressive
Junio C Hamano [Sat, 4 Feb 2006 06:04:14 +0000 (22:04 -0800)]
read-tree --aggressive

A new flag --aggressive resolves what we traditionally resolved
with external git-merge-one-file inside index while read-tree
3-way merge works.

git-merge-octopus and git-merge-resolve use this flag before
running git-merge-index with git-merge-one-file.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years ago[PATCH] mailinfo: reset CTE after each multipart
Junio C Hamano [Tue, 7 Feb 2006 05:35:25 +0000 (21:35 -0800)]
[PATCH] mailinfo: reset CTE after each multipart

If the first part uses quoted-printable to protect iso8859-1
name in the commit log, and the second part was plain ascii text
patchfile without even Content-Transfer-Encoding subheader, we
incorrectly tried to decode the patch as quoted printable.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDocs: minor git-push copyediting
J. Bruce Fields [Mon, 6 Feb 2006 00:42:27 +0000 (19:42 -0500)]
Docs: minor git-push copyediting

Minor git-push copyediting

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDocs: move git url and remotes text to separate sections
J. Bruce Fields [Sun, 5 Feb 2006 23:29:49 +0000 (18:29 -0500)]
Docs: move git url and remotes text to separate sections

The sections on git urls and remotes files in the git-fetch,
git-pull, and git-push manpages seem long enough to be worth a
manpage section of their own.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDocs: split up pull-fetch-param.txt
J. Bruce Fields [Sun, 5 Feb 2006 22:43:47 +0000 (17:43 -0500)]
Docs: split up pull-fetch-param.txt

The push and pull man pages include a bunch of shared text from
pull-fetch-param.txt.  This simplifies maintenance somewhat, but
there's actually quite a bit of text that applies only to one or the
other.

So, separate out the push- and pull/fetch-specific text into
pull-fetch-param.txt and git-push.txt, then include the largest chunk
of common stuff (the description of protocols and url's) from
urls.txt.  That cuts some irrelevant stuff from the man pages without
making us duplicate too much.

Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocombine-diff: do not punt on removed or added files.
Junio C Hamano [Tue, 7 Feb 2006 02:54:08 +0000 (18:54 -0800)]
combine-diff: do not punt on removed or added files.

When we remove a file, the parents' contents are all removed so
it is not that interesting to show all of them, but the fact it
was removed when all parents had it *is* unusual.  When we add a
file, similarly the fact it was added when no parent wanted it
*is* unusual, and in addition the result matters, so show it.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogitk: Use git-diff-tree --cc for showing the diffs for merges
Paul Mackerras [Mon, 6 Feb 2006 22:13:52 +0000 (09:13 +1100)]
gitk: Use git-diff-tree --cc for showing the diffs for merges

This replaces a lot of code that used the result from several 2-way
diffs to generate a combined diff for a merge.  Now we just use
git-diff-tree --cc and colorize the output a bit, which is a lot
simpler, and has the enormous advantage that if the diff doesn't
show quite what someone thinks it should show, I can deflect the
blame to someone else. :)

Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years agogitk: Add braces around if expressions
Paul Mackerras [Mon, 6 Feb 2006 22:10:18 +0000 (09:10 +1100)]
gitk: Add braces around if expressions

Apparently this simplifies things for the parser/compiler and makes
it go slightly faster (since without the braces, it potentially has
to do two levels of substitutions rather than one).

Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years agocombine-diff: show mode changes as well.
Junio C Hamano [Mon, 6 Feb 2006 20:53:07 +0000 (12:53 -0800)]
combine-diff: show mode changes as well.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocombine-diff: do not send NULL to printf
Junio C Hamano [Mon, 6 Feb 2006 20:30:00 +0000 (12:30 -0800)]
combine-diff: do not send NULL to printf

When we run combined diff from working tree (diff-files --cc),
we sent NULL to printf that is returned by find_unique_abbrev().

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocore-tutorial: adjust to recent reality.
Junio C Hamano [Mon, 6 Feb 2006 20:27:33 +0000 (12:27 -0800)]
core-tutorial: adjust to recent reality.

We still talked about HEAD symlinks but these days we use
symrefs by default.

Also 'failed/prevented' message is now gone from the merge
output.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-diff: do not fall back on --cc when -[123], --ours etc. are given.
Junio C Hamano [Mon, 6 Feb 2006 19:17:56 +0000 (11:17 -0800)]
git-diff: do not fall back on --cc when -[123], --ours etc. are given.

These flags ask diff with a specific unmerged stage, so it
should fall back on -p instead.  Also when -c is given, we
should not do --cc.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/diff'
Junio C Hamano [Mon, 6 Feb 2006 07:55:45 +0000 (23:55 -0800)]
Merge branch 'jc/diff'

* jc/diff:
  git-diff-tree --stdin: show all parents.
  combine-diff: remove misguided --show-empty hack.

18 years agoMerge branches 'lt/show' and 'lt/revlist'
Junio C Hamano [Mon, 6 Feb 2006 07:55:09 +0000 (23:55 -0800)]
Merge branches 'lt/show' and 'lt/revlist'

* lt/show:
  git-show

* lt/revlist:
  rev-parse lstat() workaround cleanup.

18 years agoMerge branches 'jc/daemon' and 'mw/http'
Junio C Hamano [Mon, 6 Feb 2006 07:54:14 +0000 (23:54 -0800)]
Merge branches 'jc/daemon' and 'mw/http'

* jc/daemon:
  daemon: extend user-relative path notation.
  daemon: Set SO_REUSEADDR on listening sockets.
  daemon: do not forbid user relative paths unconditionally under --base-path

* mw/http:
  http-fetch: Tidy control flow in process_alternate_response
  http: Turn on verbose Curl messages if GIT_CURL_VERBOSE set in environment
  http-fetch: Fix message reporting rename of object file.
  http-fetch: Fix object list corruption in fill_active_slots().

18 years agogit-diff-tree --stdin: show all parents.
Junio C Hamano [Mon, 6 Feb 2006 07:00:41 +0000 (23:00 -0800)]
git-diff-tree --stdin: show all parents.

git-diff-tree --stdin ignored second and subsequent parents when
fed git-rev-list --parents output.  Update diff_tree_commit()
function to take a commit object, and pass a fabricated commit
object after grafting the fake parents from diff_tree_stdin().

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocombine-diff: remove misguided --show-empty hack.
Junio C Hamano [Mon, 6 Feb 2006 06:25:00 +0000 (22:25 -0800)]
combine-diff: remove misguided --show-empty hack.

Now --always flag is available in diff-tree, there is no reason
to have that hack in the diffcore side.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agorev-parse lstat() workaround cleanup.
Linus Torvalds [Mon, 6 Feb 2006 05:41:47 +0000 (21:41 -0800)]
rev-parse lstat() workaround cleanup.

Earlier we had a workaround to avoid misspelled revision name to
be taken as a filename when "--no-revs --no-flags" are in
effect.  This cleans up the logic.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-send-email: Fully implement --quiet and document it.
Ryan Anderson [Mon, 6 Feb 2006 01:13:52 +0000 (20:13 -0500)]
git-send-email: Fully implement --quiet and document it.

Also reorganizes the man page to list options alphabetically.

Signed-off-by: Ryan Anderson <ryan@michonline.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agodaemon: extend user-relative path notation.
Junio C Hamano [Sun, 5 Feb 2006 06:27:29 +0000 (22:27 -0800)]
daemon: extend user-relative path notation.

Earlier, we made --base-path to automatically forbid
user-relative paths, which was probably a mistake.  This
introduces --user-path (or --user-path=path) option to control
the use of user-relative paths independently.  The latter form
of the option can be used to restrict accesses to a part of each
user's home directory, similar to "public_html" some webservers
supports.

If we're invoked with --user-path=FOO option, then a URL of the
form git://~USER/PATH/... resolves to the path HOME/FOO/PATH/...,
where HOME is USER's home directory.

[jc: This is much reworked by me so bugs are mine, but the
 original patch was done by Mark Wooding.]

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agodaemon: Set SO_REUSEADDR on listening sockets.
Mark Wooding [Fri, 3 Feb 2006 20:27:04 +0000 (20:27 +0000)]
daemon: Set SO_REUSEADDR on listening sockets.

Without this, you can silently lose the ability to receive IPv4
connections if you stop and restart the daemon.

[jc: tweaked code organization a bit and made this controllable
 from a command line option.]

Signed-off-by: Mark Wooding <mdw@distorted.org.uk>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agodaemon: do not forbid user relative paths unconditionally under --base-path
Junio C Hamano [Sat, 4 Feb 2006 07:50:55 +0000 (23:50 -0800)]
daemon: do not forbid user relative paths unconditionally under --base-path

Using base-path to relocate the server public space does not
have anything to do with allowing or forbidding user relative
paths.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp-fetch: Tidy control flow in process_alternate_response
Mark Wooding [Wed, 1 Feb 2006 11:44:39 +0000 (11:44 +0000)]
http-fetch: Tidy control flow in process_alternate_response

It's a bit convoluted.  Tidy it up.

Signed-off-by: Mark Wooding <mdw@distorted.org.uk>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp: Turn on verbose Curl messages if GIT_CURL_VERBOSE set in environment
Mark Wooding [Wed, 1 Feb 2006 11:44:37 +0000 (11:44 +0000)]
http: Turn on verbose Curl messages if GIT_CURL_VERBOSE set in environment

Signed-off-by: Mark Wooding <mdw@distorted.org.uk>
Signed-off-by: Junio C Hamano <junkio@cox.net>