Code

git.git
19 years ago[PATCH] git-daemon --inetd
lars.doelle@on-line.de [Thu, 8 Sep 2005 01:50:01 +0000 (03:50 +0200)]
[PATCH] git-daemon --inetd

git-daemon using inetd. does not work properly. inetd routes stderr onto the
network line just like stdout, which was apparently not expected to be so.

As the result of this, the stream is closed by the receiver, because some
"Packing %d objects\n" originating from pack_objects is first reported over
the line instead of the expected pack_header, and so the SIGNATURE test
fails.  Here is a workaround.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit --version tells which version of git you have.
Junio C Hamano [Thu, 8 Sep 2005 04:26:52 +0000 (21:26 -0700)]
git --version tells which version of git you have.

Originally from Martin Atukunda <matlads@dsmagic.com> but adjusted for
post-rename code.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Do not create bogus branch from flag to git branch
Amos Waterland [Thu, 8 Sep 2005 02:13:26 +0000 (21:13 -0500)]
[PATCH] Do not create bogus branch from flag to git branch

If you run `git branch --help', you will unexpectedly have created a new
branch named "--help".  This simple patch adds logic and a usage
statement to catch this and similar problems, and adds a testcase for it.

Signed-off-by: Amos Waterland <apw@rossby.metr.ou.edu>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Portability fix for Solaris 10/x86
Patrick Mauritz [Mon, 5 Sep 2005 23:24:03 +0000 (01:24 +0200)]
[PATCH] Portability fix for Solaris 10/x86

 * getdomainname unavailable there.
 * needs -lsocket for linkage.
 * needs __EXTENSIONS__ at the beginning of convert-objects.c

[JC: I've done this slightly differently from what Patrick originally
sent to the list and dropped the bit that deals with installations
that has curl header and library at non-default location.  I am
resisting the slipperly slope called autoconf.]

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoBig tool rename.
Junio C Hamano [Thu, 8 Sep 2005 00:26:23 +0000 (17:26 -0700)]
Big tool rename.

As promised, this is the "big tool rename" patch.  The primary differences
since 0.99.6 are:

  (1) git-*-script are no more.  The commands installed do not
      have any such suffix so users do not have to remember if
      something is implemented as a shell script or not.

  (2) Many command names with 'cache' in them are renamed with
      'index' if that is what they mean.

There are backward compatibility symblic links so that you and
Porcelains can keep using the old names, but the backward
compatibility support  is expected to be removed in the near
future.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoGIT 0.99.6 v0.99.6
Junio C Hamano [Wed, 7 Sep 2005 22:59:24 +0000 (15:59 -0700)]
GIT 0.99.6

19 years ago[PATCH] Simplify git script
David_Kågedal [Wed, 7 Sep 2005 07:29:20 +0000 (09:29 +0200)]
[PATCH] Simplify git script

The code for listing the available subcommands was unnecessarily
complex.

Signed-off-by: David Kågedal <davidk@lysator.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoDocumentation updates.
Junio C Hamano [Wed, 7 Sep 2005 21:28:21 +0000 (14:28 -0700)]
Documentation updates.

More commands are documented now; thanks Raymond.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Docs for git-build-rev-cache.
A Large Angry SCM [Wed, 7 Sep 2005 21:20:41 +0000 (17:20 -0400)]
[PATCH] Docs for git-build-rev-cache.

Signed-off-by: A Large Angry SCM <gitzilla@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Docs for git-show-rev-cache.
A Large Angry SCM [Wed, 7 Sep 2005 21:19:48 +0000 (17:19 -0400)]
[PATCH] Docs for git-show-rev-cache.

Signed-off-by: A Large Angry SCM <gitzilla@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Docs for git-reset-script.
A Large Angry SCM [Wed, 7 Sep 2005 21:18:51 +0000 (17:18 -0400)]
[PATCH] Docs for git-reset-script.

Signed-off-by: A Large Angry SCM <gitzilla@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Docs for git-checkout-script.
A Large Angry SCM [Wed, 7 Sep 2005 21:17:18 +0000 (17:17 -0400)]
[PATCH] Docs for git-checkout-script.

Signed-off-by: A Large Angry SCM <gitzilla@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoDocumentation updates.
Junio C Hamano [Wed, 7 Sep 2005 21:08:38 +0000 (14:08 -0700)]
Documentation updates.

parse-remote and rev-parse gets full documentation.  Add skeleton for
archimport.  Link them from the main git(7) page.  Also move git-daemon
and git-request-pull out of 'undocumented' section.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFlatten tools/ directory to make build procedure simpler.
Junio C Hamano [Wed, 7 Sep 2005 19:22:56 +0000 (12:22 -0700)]
Flatten tools/ directory to make build procedure simpler.

Also make platform specific part more isolated.  Currently we only
have Darwin defined, but I've taken a look at SunOS specific patch
(which I dropped on the floor for now) as well.  Doing things this way
would make adding it easier.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agomailinfo: barf and exist upon nested multipart.
Junio C Hamano [Tue, 6 Sep 2005 23:46:34 +0000 (16:46 -0700)]
mailinfo: barf and exist upon nested multipart.

At least we can detect what we do not handle.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge branch 'master' of .
Junio C Hamano [Tue, 6 Sep 2005 20:51:33 +0000 (13:51 -0700)]
Merge branch 'master' of .

19 years ago[PATCH] Update documentation for git-get-tar-commit-id
Rene Scharfe [Tue, 6 Sep 2005 19:21:16 +0000 (21:21 +0200)]
[PATCH] Update documentation for git-get-tar-commit-id

... and add a copyright notice.

[jc: also move its entry in git.txt from undocumented section.]

Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] git-cvsimport-script: handling of tags
H. Peter Anvin [Tue, 6 Sep 2005 17:36:01 +0000 (10:36 -0700)]
[PATCH] git-cvsimport-script: handling of tags

This patch changes git-cvsimport-script so that it creates tag objects
instead of refs to commits, and adds an option, -u, to convert
underscores in branch and tag names to dots (since CVS doesn't allow
dots in branches and tags.)

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFormat fix for asciidoc documentation titles.
Junio C Hamano [Tue, 6 Sep 2005 16:06:32 +0000 (09:06 -0700)]
Format fix for asciidoc documentation titles.
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge branch 'master' of .
Junio C Hamano [Mon, 5 Sep 2005 08:42:11 +0000 (01:42 -0700)]
Merge branch 'master' of .

19 years agoRetire git-clone-dumb-http.
Junio C Hamano [Mon, 5 Sep 2005 07:47:39 +0000 (00:47 -0700)]
Retire git-clone-dumb-http.

... and fold it into git-clone-script.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Update documentation of --compose to git-send-email-script.txt
Ryan Anderson [Mon, 5 Sep 2005 05:13:07 +0000 (01:13 -0400)]
[PATCH] Update documentation of --compose to git-send-email-script.txt

Signed-off-by: Ryan Anderson <ryan@michonline.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Make git-send-email-script ignore some unnecessary options when operating...
Ryan Anderson [Mon, 5 Sep 2005 05:13:07 +0000 (01:13 -0400)]
[PATCH] Make git-send-email-script ignore some unnecessary options when operating in batch mode.

Add a "--compose" option that uses $EDITOR to edit an "introductory" email to the patch series.

Signed-off-by: Ryan Anderson <ryan@michonline.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoInstall archimport-script.
Junio C Hamano [Mon, 5 Sep 2005 06:30:08 +0000 (23:30 -0700)]
Install archimport-script.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge branch 'master' of .
Junio C Hamano [Sun, 4 Sep 2005 22:47:28 +0000 (15:47 -0700)]
Merge branch 'master' of .

19 years agoAdd copy/rename check for git-apply.
Junio C Hamano [Sun, 4 Sep 2005 22:40:18 +0000 (15:40 -0700)]
Add copy/rename check for git-apply.

The new test pattern is taken from HPA's klibc and klibc-kbuild trees,
which has many interesting renames (the commits
001eef5a19219e5b0601068a3d13874b88c0653e and
0037d1bc0deaf7daec3778496656cb04b4e4b9d0).  The test pattern exposes
problems in the apply.c changes currently in the proposed updates
branch.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoAdd a link from update-server-info documentation to repository layout.
Junio C Hamano [Sun, 4 Sep 2005 18:25:56 +0000 (11:25 -0700)]
Add a link from update-server-info documentation to repository layout.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Make git-apply understand incomplete lines in non-C locales
Fredrik Kuivinen [Sun, 4 Sep 2005 17:29:02 +0000 (19:29 +0200)]
[PATCH] Make git-apply understand incomplete lines in non-C locales

The message "\ No newline at end of file" used by diff(1) to mark
an incomplete line is locale dependent. We can't assume more than
that it begins with "\ ".

For example, given two files, "foo" and "bar", with appropriate
contents, 'diff -u foo bar' will produce the following output on
my system:

    --- foo 2005-09-04 18:59:38.000000000 +0200
    +++ bar 2005-09-04 18:59:16.000000000 +0200
    @@ -1 +1 @@
    -foobar
    +foo
    \ Ingen nyrad vid filslut

[jc: the check for the marker still uses the line length being no less
than 12 bytes for a sanity check, but I think it is safe to assume
that in other locales. I haven't checked the .po files from diff, tho'.]

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-applymbox: fix '-c'.
Junio C Hamano [Sun, 4 Sep 2005 17:37:07 +0000 (10:37 -0700)]
git-applymbox: fix '-c'.

Earlier round b50abe8843006e9856c633b5abeb4eab53b46629 broke it
by carelessly rewriting the main loop.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] archimport: avoid committing on an Arch tag
Martin Langhoff [Sun, 4 Sep 2005 10:55:29 +0000 (22:55 +1200)]
[PATCH] archimport: avoid committing on an Arch tag

Arch tags are full commits (without any changed files) as well. Trust Arch
to have put an unchanged tree in place (which seems to do reliably), and
just add a tag & new branch. Speeds up Arch imports significantly, and leaves
history in a much saner state.

Signed-off-by: Martin Langhoff <martin@catalyst.net.nz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] archimport autodetects import status, supports incremental imports
Martin Langhoff [Sun, 4 Sep 2005 10:55:06 +0000 (22:55 +1200)]
[PATCH] archimport autodetects import status, supports incremental imports

If there is no GIT directory, archimport will assume it is an initial import.

It now also supports incremental imports,  skipping "seen" commits. You can
now run it repeatedly to pull new commits from the Arch repository.

Signed-off-by: Martin Langhoff <martin@catalyst.net.nz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] NUL terminate the object data in patch_delta()
Sergey Vlasov [Sun, 4 Sep 2005 09:43:16 +0000 (13:43 +0400)]
[PATCH] NUL terminate the object data in patch_delta()

At least pretty_print_commit() expects to get NUL-terminated commit data to
work properly.  unpack_sha1_rest(), which reads objects from separate files,
and unpack_non_delta_entry(), which reads non-delta-compressed objects from
pack files, already add the NUL byte after the object data, but patch_delta()
did not do it, which caused problems with, e.g., git-rev-list --pretty when
there are delta-compressed commit objects.

Signed-off-by: Sergey Vlasov <vsu@altlinux.ru>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Possible cleanups for local-pull.c
Peter Hagervall [Fri, 2 Sep 2005 12:17:10 +0000 (14:17 +0200)]
[PATCH] Possible cleanups for local-pull.c

Hi. This patch contains the following possible cleanups:

 * Make some needlessly global functions in local-pull.c static
 * Change 'char *' to 'const char *' where appropriate

Signed-off-by: Peter Hagervall <hager@cs.umu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Doc: replace read-cache with git-read-tree.
Paolo 'Blaisorblade' Giarrusso [Fri, 2 Sep 2005 09:56:52 +0000 (11:56 +0200)]
[PATCH] Doc: replace read-cache with git-read-tree.

Replace references to "read-cache" with references to git-read-tree in the
documentation. I chose that because reference say "see read-cache about
stages", and stages are explained in git-read-tree.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoDocument hooks.
Junio C Hamano [Sat, 3 Sep 2005 04:19:26 +0000 (21:19 -0700)]
Document hooks.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge branch 'master' of .
Junio C Hamano [Fri, 2 Sep 2005 22:32:55 +0000 (15:32 -0700)]
Merge branch 'master' of .

19 years agoMention post-update when we first talk about publishing a repository.
Junio C Hamano [Fri, 2 Sep 2005 18:46:43 +0000 (11:46 -0700)]
Mention post-update when we first talk about publishing a repository.

There is more detailed instruction for `project lead` later in
the tutorial to talk about the same, but at this point in the
flow of tutorial, the first time reader has no way of knowing it.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoscripts: equality test '==' is not portable.
Junio C Hamano [Fri, 2 Sep 2005 17:53:15 +0000 (10:53 -0700)]
scripts: equality test '==' is not portable.

On NetBSD 3 we trigger an error:

    [: ==: unexpected operator

Double-equal is accepted by bash built-in '[' and bash(1) suggests
using '=' for strict POSIX compliance (test(1) from coreutils does not
mention '==').  Eradicate their uses everywhere.

[jc: Somebody with a pseudonym kindly sent a message to let
 me know about the problem privately; I do not have access to a NetBSD
 box.]

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-checkout-script: Remove unnecessary variable.
Junio C Hamano [Fri, 2 Sep 2005 17:39:57 +0000 (10:39 -0700)]
git-checkout-script: Remove unnecessary variable.

There was unused variable $i that counted the number of arguments
being processed.  Remove it.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix automerge message.
Junio C Hamano [Fri, 2 Sep 2005 09:34:13 +0000 (02:34 -0700)]
Fix automerge message.

The rewrite done while adding the shorthand support made the remote
refname recorded in the commit message too long for human consumption,
while losing information by using the shorthand not the real URL to
name the remote repository there.  They were both bad changes done
without enough thinking.

Pointed out by Linus.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Explain what went wrong on update-cache of new file
Amos Waterland [Thu, 1 Sep 2005 14:13:50 +0000 (09:13 -0500)]
[PATCH] Explain what went wrong on update-cache of new file

If somebody tries to run `git update-cache foo', where foo is a new
file, git dies with a rather cryptic error message:

 fatal: Unable to add foo to database

This trivial patch makes git explain what probably went wrong.  It is
not a perfect diagnosis of all error paths, but for 90% of the cases it
should provide the user with the clue they need.

[jc: I ended up wording slightly differently, and fixed another
 confusing error message I noticed while reviewing the code.]

Signed-off-by: Amos Waterland <apw@rossby.metr.ou.edu>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoAdd repository-layout document.
Junio C Hamano [Thu, 1 Sep 2005 23:56:13 +0000 (16:56 -0700)]
Add repository-layout document.

... and link to it from both the main index and the tutorial.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Thu, 1 Sep 2005 08:02:39 +0000 (01:02 -0700)]
Merge refs/heads/master from .

19 years agoUse 'git status' now it can handle initial commit.
Junio C Hamano [Thu, 1 Sep 2005 00:15:25 +0000 (17:15 -0700)]
Use 'git status' now it can handle initial commit.

Update 'git commit' to use the updated `git status`.  Also earlier
the `-s` flag was ignored for the initial commit.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix git-status when HEAD is invalid.
Junio C Hamano [Thu, 1 Sep 2005 00:13:48 +0000 (17:13 -0700)]
Fix git-status when HEAD is invalid.

It tried to do git-diff-cache against HEAD, of course.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix git-log-script when HEAD is invalid.
Junio C Hamano [Thu, 1 Sep 2005 00:11:39 +0000 (17:11 -0700)]
Fix git-log-script when HEAD is invalid.

It used 'die' without including git-sh-setup-script; since everything
it uses are subdirectory-aware, instead of including the script to
force it to be run from the top, use echo & exit.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoAdd Pine 4.63 help from Daniel.
Junio C Hamano [Wed, 31 Aug 2005 18:48:41 +0000 (11:48 -0700)]
Add Pine 4.63 help from Daniel.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Wed, 31 Aug 2005 06:11:12 +0000 (23:11 -0700)]
Merge refs/heads/master from .

19 years agoUpdate tutorial.
Junio C Hamano [Wed, 31 Aug 2005 06:08:06 +0000 (23:08 -0700)]
Update tutorial.

Finally I bit the bullet and did a full sweep of this document.
The changes are mostly clarifications, adjusting old terminology
to the glossary compatible one, and asciidoc formatting.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoshow-branch: make it work in a subdirectory.
Junio C Hamano [Tue, 30 Aug 2005 23:59:37 +0000 (16:59 -0700)]
show-branch: make it work in a subdirectory.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Initial import of git-archimport-script
Martin Langhoff [Tue, 30 Aug 2005 09:56:52 +0000 (21:56 +1200)]
[PATCH] Initial import of git-archimport-script

Imports a project history from one or more Arch repositories, following
the branching and tagging across repositories. Note you should import
separate projects to separate GIT repositories.

Supported
 - Imports, tags and simple commits.
 - File renames
 - Arch tags
 - Binary files
 - Large trees
 - Multiple repositories
 - Branches

TODO:
 - Allow re-running the import on an already-imported project
 - Follow merges using Arch data
 - Audit shell escaping of filenames
 - Better support for file metadata
 - Better/safer creation of temp directories

Unsupported:
 - Arch 'configuration'

[jc: my arch/tla is very rusty and after Tom announced he is stepping
 down as the maintainer I have very little motivation to relearn it,
 so I would appreciate if people discuss any bugs or enhancements
 directly with Martin.  Of course I can help with the git end of the
 issues.]

Signed-off-by: Martin Langhoff <martin@catalyst.net.nz>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Tue, 30 Aug 2005 21:33:27 +0000 (14:33 -0700)]
Merge refs/heads/master from .

19 years agocat-file: make it work in a subdirectory.
Junio C Hamano [Tue, 30 Aug 2005 16:07:50 +0000 (09:07 -0700)]
cat-file: make it work in a subdirectory.

Signed-off-by: Junio C Hamano <junkio@cox.net>
(cherry picked from 73244994ee4abd6e436e8a1d597dd917271d77a9 commit)

19 years agoDocumentaion updates.
Junio C Hamano [Tue, 30 Aug 2005 20:51:01 +0000 (13:51 -0700)]
Documentaion updates.

Mostly making the formatted html prettier.

Signed-off-by: Junio C Hamano <junkio@cox.net>
(cherry picked from 7adf1f15ebe074d4767df941817a6cf86d8e2533 commit)

19 years ago'git bisect visualize'
Junio C Hamano [Tue, 30 Aug 2005 19:45:41 +0000 (12:45 -0700)]
'git bisect visualize'

Linus says:

    I'm testing bisection to find a bug that causes my G5 to no longer boot,
    and during the process have found this command line very nice:

    gitk bisect/bad --not $(cd .git/refs ; ls bisect/good-*)

    it basically shows the state of bisection with the known bad commit as the
    top, and cutting off all the good commits - so what you see are the
    potential buggy commits.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Fix bisection terminating condition
Linus Torvalds [Tue, 30 Aug 2005 18:04:39 +0000 (11:04 -0700)]
[PATCH] Fix bisection terminating condition

When testing bisection and using gitk to visualize the result, it was
obvious that the termination condition was broken.

We know what the bad entry is only when the bisection ends up telling us
to test the known-bad entry again.

Also, add a safety net: if somebody marks as good something that includes
the known-bad point, we now notice and complain, instead of writing an
empty revision to the new bisection branch.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Tue, 30 Aug 2005 08:10:10 +0000 (01:10 -0700)]
Merge refs/heads/master from .

19 years agoparse-remote: trivial fix to allow refs/{heads,tags}/ spelled easier.
Junio C Hamano [Tue, 30 Aug 2005 08:08:24 +0000 (01:08 -0700)]
parse-remote: trivial fix to allow refs/{heads,tags}/ spelled easier.

Earlier we always prefixed refs/heads to the token given to "git fetch"
(and "git pull") as refspec.  This was a mistake.  Allow them to be
spelled like "master:refs/tags/paulus" to mean "I want to fetch the
master there and store it as my local "paulus" tag.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from paulus
Junio C Hamano [Tue, 30 Aug 2005 07:44:44 +0000 (00:44 -0700)]
Merge refs/heads/master from paulus

19 years agoMake sure howto/*.html is built as well.
Junio C Hamano [Tue, 30 Aug 2005 06:09:22 +0000 (23:09 -0700)]
Make sure howto/*.html is built as well.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Make "git resolve" less scary
Linus Torvalds [Tue, 30 Aug 2005 05:36:16 +0000 (22:36 -0700)]
[PATCH] Make "git resolve" less scary

When we resolve a merge between two branches, and it removes a file in the
current branch, we notify the person doing the resolve with a big nice
notice like

Removing xyzzy

which is all well and good.

HOWEVER, we also do this when the file was actually removed in the current
branch, and we're merging with another branch that didn't have it removed
(or, indeed, if the other branch _did_ have it removed, but the common
parent was far enough back that the file still existed in there).

And that just doesn't make sense. In that case we're not removing
anything: the file didn't exist in the branch we're merging into in the
first place. So the message just makes people nervous, and makes no sense.

This has been around forever, but I never bothered to do anything about
it.

Until now.

The trivial fix is to only talk about removing files if the file existed
in the branch we're merging into, but will not exist in the result.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Tue, 30 Aug 2005 05:38:45 +0000 (22:38 -0700)]
Merge refs/heads/master from .

19 years agoAllow asciidoc formatted documentation in howto/
Junio C Hamano [Tue, 30 Aug 2005 05:38:12 +0000 (22:38 -0700)]
Allow asciidoc formatted documentation in howto/

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Tue, 30 Aug 2005 04:51:26 +0000 (21:51 -0700)]
Merge refs/heads/master from .

19 years agoAdd [HOWTO] revert/branch/rebase.
Junio C Hamano [Tue, 30 Aug 2005 04:50:49 +0000 (21:50 -0700)]
Add [HOWTO] revert/branch/rebase.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit commit: Allow editing on all occasions.
Junio C Hamano [Tue, 30 Aug 2005 04:25:19 +0000 (21:25 -0700)]
git commit: Allow editing on all occasions.

Sometimes it may be handy to be able to edit messages that come
from somewhere other than an existing commit.

This makes 'git commit -F <file> -e' to start editor with the initial
log message contents taken from <file>.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoDo not verify reverted/cherry-picked/rebased patches.
Junio C Hamano [Tue, 30 Aug 2005 04:19:04 +0000 (21:19 -0700)]
Do not verify reverted/cherry-picked/rebased patches.

The original committer may have used validation criteria that is less
stricter than yours.  You do not want to lose the changes even if they
are done in substandard way from your 'commit -v' verifier's point of
view.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoFix compilation warnings.
Junio C Hamano [Tue, 30 Aug 2005 04:17:21 +0000 (21:17 -0700)]
Fix compilation warnings.

... found by compiling them with gcc 2.95.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Update Thunderbird specific hints.
A Large Angry SCM [Tue, 30 Aug 2005 02:34:07 +0000 (22:34 -0400)]
[PATCH] Update Thunderbird specific hints.

Setting the wraplength to zero keeps the bird from trimming WS.

Signed-off-by: <gitzilla@gmail.com>
(cherry picked from 1d535d525d6a0ddddc3755065d721278bc5f0aff commit)

19 years ago[PATCH] Documentation for git-request-pull-script.
A Large Angry SCM [Tue, 30 Aug 2005 02:33:14 +0000 (22:33 -0400)]
[PATCH] Documentation for git-request-pull-script.

Copy & paste source comments into documentation.

Signed-off-by: <gitzilla@gmail.com>
(cherry picked from 10dce95490cb02f66b3a08984ccdee0a92b48236 commit)

19 years ago[PATCH] Documentation for git-clone-dumb-http.
A Large Angry SCM [Tue, 30 Aug 2005 02:32:04 +0000 (22:32 -0400)]
[PATCH] Documentation for git-clone-dumb-http.

Copy & paste source comments into documentation.

Signed-off-by: <gitzilla@gmail.com>
(cherry picked from f79528e5d05a64d68b8e09a18521950775e99ec1 commit)

19 years ago[PATCH] Documentation for git-daemon.
A Large Angry SCM [Tue, 30 Aug 2005 02:31:22 +0000 (22:31 -0400)]
[PATCH] Documentation for git-daemon.

Copy & paste source comments into documentation.

Signed-off-by: <gitzilla@gmail.com>
(cherry picked from 5d0a4efeff62cfa363437f91308453b5b9fd8cf5 commit)

19 years agoMerge refs/heads/master from .
Junio C Hamano [Tue, 30 Aug 2005 02:09:48 +0000 (19:09 -0700)]
Merge refs/heads/master from .

19 years agoDraw selected graph line thicker and make arrowheads active.
Paul Mackerras [Tue, 30 Aug 2005 00:57:11 +0000 (10:57 +1000)]
Draw selected graph line thicker and make arrowheads active.

19 years agoAccumulated documentation updates.
Junio C Hamano [Tue, 30 Aug 2005 00:21:06 +0000 (17:21 -0700)]
Accumulated documentation updates.
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agogit-repack-script: Add option to repack all objects.
Junio C Hamano [Mon, 29 Aug 2005 17:29:53 +0000 (10:29 -0700)]
git-repack-script: Add option to repack all objects.

This originally came from Frank Sorenson, but with a bit of rework to
allow future enhancements without changing the external interface for
pack pruning part.

With the '-a' option, all objects in the current repository are packed
into a single pack.  When the '-d' option is given at the same time,
existing packs that were made redundant by this round of repacking are
deleted.

Since we currently have only two repacking strategies, one with '-a'
(everything into one) and the other without '-a' (incrementally pack
only the unpacked ones), the '-d' option is meaningful only when used
with '-a'; it removes the packs existed before we did the "everything
into one" repacking.  At least for now.

Signed-off-by: Junio C Hamano <junkio@cox.net>
Acked-by: Frank Sorenson <frank@tuxrocks.com>
(cherry picked from bfed505327e31221d8de796b3af880bad696b149 commit)

19 years agoUpdates: show-branch
Junio C Hamano [Tue, 30 Aug 2005 00:19:47 +0000 (17:19 -0700)]
Updates: show-branch

Earlier show-branch gave names only to commits reachable via first
parent ancestry chain.  Change the naming code to name everybody.

The original idea was to stop at the first merge point in the
topological order, and --more=<n> to show commits until we show <n>
more extra merge points.  However depending on the order of how we
discover the commits, it additionally showed parents of the <n>th
merge points, which was unnecessary.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Mon, 29 Aug 2005 20:07:55 +0000 (13:07 -0700)]
Merge refs/heads/master from .

19 years agomailinfo and applymbox updates
Junio C Hamano [Sun, 28 Aug 2005 19:33:16 +0000 (12:33 -0700)]
mailinfo and applymbox updates

This attempts to minimally cope with a subset of MIME "features" often
seen in patches sent to our mailing lists.  Namely:

 - People's name spelled in characters outside ASCII (both on From:
   header and the signed-off-by line).

 - Content-transfer-encoding using quoted-printable (both in
   multipart and non-multipart messages).

These MIME features are detected and decoded by "git mailinfo".

Optionally, with the '-u' flag, the output to .info and .msg is
transliterated from its original chaset to utf-8.  This is to
encourage people to use utf8 in their commit messages for
interoperability.

Applymbox accepts additional flag '-u' which is passed to mailinfo.

Signed-off-by: Junio C Hamano / 濱野 純 <junkio@cox.net>
19 years agoDocument "git cherry-pick" and "git revert"
Junio C Hamano [Sun, 28 Aug 2005 10:01:09 +0000 (03:01 -0700)]
Document "git cherry-pick" and "git revert"

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoRemove git-apply-patch-script.
Junio C Hamano [Sat, 27 Aug 2005 00:27:07 +0000 (17:27 -0700)]
Remove git-apply-patch-script.

Now the rebase is rewritten to use git cherry-pick, there is no user
for that ancient script.  I've checked Cogito and StGIT to make sure
they do not use it.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoRedo "revert" using three-way merge machinery.
Junio C Hamano [Sun, 28 Aug 2005 06:53:27 +0000 (23:53 -0700)]
Redo "revert" using three-way merge machinery.

The reverse patch application using "git apply" sometimes is too
rigid.  Since the user would get used to resolving conflicting merges
by hand during the normal merge experience, using the same machinery
would be more helpful rather than just giving up.

Cherry-picking and reverting are essentially the same operation.
You pick one commit, and apply the difference that commit introduces
to its own commit ancestry chain to the current tree.  Revert applies
the diff in reverse while cherry-pick applies it forward.  They share
the same logic, just different messages and merge direction.

Rewrite "git rebase" using "git cherry-pick".

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoRevert "Replace zero-length array decls with []."
Junio C Hamano [Mon, 29 Aug 2005 19:41:03 +0000 (12:41 -0700)]
Revert "Replace zero-length array decls with []."

This reverts 6c5f9baa3bc0d63e141e0afc23110205379905a4 commit, whose
change breaks gcc-2.95.

Not that I ignore portability to compilers that are properly C99, but
keeping compilation with GCC working is more important, at least for
now.  We would probably end up declaring with "name[1]" and teach the
allocator to subtract one if we really aimed for portability, but that
is left for later rounds.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Mon, 29 Aug 2005 06:03:23 +0000 (23:03 -0700)]
Merge refs/heads/master from .

19 years agoMerge refs/heads/portable from http://www.cs.berkeley.edu/~ejr/gits/git.git
Junio C Hamano [Mon, 29 Aug 2005 06:02:01 +0000 (23:02 -0700)]
Merge refs/heads/portable from cs.berkeley.edu/~ejr/gits/git.git

19 years ago[PATCH] tutorial note about git branch
Amos Waterland [Mon, 29 Aug 2005 04:05:36 +0000 (23:05 -0500)]
[PATCH] tutorial note about git branch

Explain that an asterisk will be displayed in front of the current
branch when you run `git branch' to see which are available.

Signed-off-by: Amos Waterland <apw@rossby.metr.ou.edu>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge refs/heads/master from .
Junio C Hamano [Mon, 29 Aug 2005 00:57:45 +0000 (17:57 -0700)]
Merge refs/heads/master from .

19 years agoUpdate SubmittingPatches.
Junio C Hamano [Mon, 29 Aug 2005 00:54:18 +0000 (17:54 -0700)]
Update SubmittingPatches.

- It does not matter how I read git list.  What matters is that
  I do not necessarily read everything on it.

- Talk a bit about how to use applymbox to check one's own
  patches.

- Talk a bit about PGP signed patches.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Add footnote about Thunderbird about trimming trailing WS.
A Large Angry SCM [Sun, 28 Aug 2005 21:17:15 +0000 (17:17 -0400)]
[PATCH] Add footnote about Thunderbird about trimming trailing WS.

Add footnote about Thunderbird about trimming trailing WS.

Signed-off-by: <gitzilla@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Add some documentation.
A Large Angry SCM [Sun, 28 Aug 2005 21:16:01 +0000 (17:16 -0400)]
[PATCH] Add some documentation.

Add some documentation.

Text taken from the the commit messages and the command sources.

Signed-off-by: <gitzilla@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] New git-apply test cases for scanning forwards and backwards.
Robert Fitzsimons [Sun, 28 Aug 2005 15:25:10 +0000 (15:25 +0000)]
[PATCH] New git-apply test cases for scanning forwards and backwards.

Added a new test case for the scanning forwards and backwards for the
correct location to apply a patch fragment.

Signed-off-by: Robert Fitzsimons <robfitz@273k.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] New git-apply test cases for patches with mulitple fragments.
Robert Fitzsimons [Sun, 28 Aug 2005 15:25:06 +0000 (15:25 +0000)]
[PATCH] New git-apply test cases for patches with mulitple fragments.

Added a test case for patches with multiple fragments.

Signed-off-by: Robert Fitzsimons <robfitz@273k.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Fix git patch header processing in git-apply.
Robert Fitzsimons [Sun, 28 Aug 2005 15:24:27 +0000 (15:24 +0000)]
[PATCH] Fix git patch header processing in git-apply.

Stop processing and return NULL if we encounter a '\n' character
before we have two matching names in the git header.

Signed-off-by: Robert Fitzsimons <robfitz@273k.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoTeach git-status-script about git-ls-files --others
Junio C Hamano [Fri, 26 Aug 2005 09:12:50 +0000 (02:12 -0700)]
Teach git-status-script about git-ls-files --others

When there is non-empty $GIT_DIR/info/exclude file, use it along
with .gitignore per-directory exclude pattern files (which was
a convention agreed on the list while ago and is compatible with
Cogito) to generate a list of ignored files as well.

Signed-off-by: Junio C Hamano <junkio@cox.net>
(cherry picked from d330948a5ff0df55c2f12627c0583b4e16f1ea4d commit)

19 years agoShow the branch name more prominently in "git status".
Junio C Hamano [Fri, 26 Aug 2005 23:53:26 +0000 (16:53 -0700)]
Show the branch name more prominently in "git status".

When not working on "master" branch, remind the user at the beginning
of the status message, not at the end.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoBarf nicely when "git push" is run without parameter.
Junio C Hamano [Fri, 26 Aug 2005 17:37:17 +0000 (10:37 -0700)]
Barf nicely when "git push" is run without parameter.

Saying "internal error" makes it look like my fault ;-).

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years ago[PATCH] Make .git directory validation code test HEAD
Linus Torvalds [Sat, 27 Aug 2005 20:54:42 +0000 (13:54 -0700)]
[PATCH] Make .git directory validation code test HEAD

Inspired by a report by Kalle Valo, this changes git-sh-setup-script and
the "setup_git_directory()" function to test that $GIT_DIR/HEAD is a
symlink, since a number of core git features depend on that these days.

We used to allow a regular file there, but git-fsck-cache has been
complaining about that for a while, and anything that uses branches
depends on the HEAD file being a symlink, so let's just encode that as a
fundamental requirement.

Before, a non-symlink HEAD file would appear to work, but have subtle bugs
like not having the HEAD show up as a valid reference (because it wasn't
under "refs"). Now, we will complain loudly, and the user can fix it up
trivially instead of getting strange behaviour.

This also removes the tests for "$GIT_DIR" and "$GIT_OBJECT_DIRECTORY"
being directories, since the other tests will implicitly test for that
anyway (ie the tests for HEAD, refs and 00 would fail).

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoUpdate SubmittingPatches to add MUA specific notes.
Junio C Hamano [Sat, 27 Aug 2005 06:53:07 +0000 (23:53 -0700)]
Update SubmittingPatches to add MUA specific notes.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoMerge master branch changes into release candidate branch.
Junio C Hamano [Sat, 27 Aug 2005 05:47:27 +0000 (22:47 -0700)]
Merge master branch changes into release candidate branch.

Signed-off-by: Junio C Hamano <junkio@cox.net>
19 years agoAccept -m and friends for initial commits and merge commits.
Junio C Hamano [Fri, 26 Aug 2005 01:57:35 +0000 (18:57 -0700)]
Accept -m and friends for initial commits and merge commits.

Yes it was irritating not to be able to do so.

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