git.git
18 years agoMerge branch 'jc/name' into next
Junio C Hamano [Tue, 21 Mar 2006 08:15:21 +0000 (00:15 -0800)]
Merge branch 'jc/name' into next

* jc/name:
  core.warnambiguousrefs: warns when "name" is used and both "name" branch and tag exists.
  contrib/git-svn: allow rebuild to work on non-linear remote heads
  http-push: don't assume char is signed
  http-push: add support for deleting remote branches
  Be verbose when !initial commit
  Fix multi-paragraph list items in OPTIONS section
  http-fetch: nicer warning for a server with unreliable 404 status

18 years agoMerge branch 'jc/clone' into next
Junio C Hamano [Tue, 21 Mar 2006 08:15:15 +0000 (00:15 -0800)]
Merge branch 'jc/clone' into next

* jc/clone:
  revamp git-clone (take #2).

18 years agorevamp git-clone (take #2).
Junio C Hamano [Tue, 21 Mar 2006 08:14:13 +0000 (00:14 -0800)]
revamp git-clone (take #2).

This builds on top of the previous one.

 * --use-separate-remote uses .git/refs/remotes/$origin/
   directory to keep track of the upstream branches.

 * The $origin above defaults to "origin" as usual, but the
   existing "-o $origin" option can be used to override it.

I am not yet convinced if we should make "$origin" the synonym to
"refs/remotes/$origin/$name" where $name is the primary branch
name of $origin upstream, nor if so how we should decide which
upstream branch is the primary one, but that is more or less
orthogonal to what the clone does here.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocore.warnambiguousrefs: warns when "name" is used and both "name" branch and tag...
Junio C Hamano [Tue, 21 Mar 2006 02:45:47 +0000 (18:45 -0800)]
core.warnambiguousrefs: warns when "name" is used and both "name" branch and tag exists.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocontrib/git-svn: allow rebuild to work on non-linear remote heads
Eric Wong [Tue, 21 Mar 2006 04:51:16 +0000 (20:51 -0800)]
contrib/git-svn: allow rebuild to work on non-linear remote heads

Because committing back to an SVN repository from different
machines can result in different lineages, two different
repositories running git-svn can result in different commit
SHA1s (but of the same tree).  Sometimes trees that are tracked
independently are merged together (usually via children),
resulting in non-unique git-svn-id: lines in rev-list.

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp-push: don't assume char is signed
Nick Hengeveld [Mon, 20 Mar 2006 19:50:51 +0000 (11:50 -0800)]
http-push: don't assume char is signed

Declare remote_dir_exists[] as signed char to be sure that values of -1
are valid.

Signed-off-by: Nick Hengeveld <nickh@reactrix.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp-push: add support for deleting remote branches
Nick Hengeveld [Mon, 20 Mar 2006 18:31:06 +0000 (10:31 -0800)]
http-push: add support for deleting remote branches

Processes new command-line arguments -d and -D to remove a remote branch
if the following conditions are met:
- one branch name is present on the command line
- the specified branch name matches exactly one remote branch name
- the remote HEAD is a symref
- the specified branch is not the remote HEAD
- the remote HEAD resolves to an object that exists locally (-d only)
- the specified branch resolves to an object that exists locally (-d only)
- the specified branch is an ancestor of the remote HEAD (-d only)

Signed-off-by: Nick Hengeveld <nickh@reactrix.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoBe verbose when !initial commit
Yasushi SHOJI [Mon, 20 Mar 2006 13:11:12 +0000 (22:11 +0900)]
Be verbose when !initial commit

verbose option in git-commit.sh lead us to run git-diff-index, which
needs a commit-ish we are making diff against.  When we are commiting
the fist set, we obviously don't have any commit-ish in the repo.  So
we just skip the git-diff-index run.

It might be possible to produce diff against empty but do we need
that?

Signed-off-by: Yasushi SHOJI <yashi@atmark-techno.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoFix multi-paragraph list items in OPTIONS section
Francis Daly [Mon, 20 Mar 2006 10:41:18 +0000 (10:41 +0000)]
Fix multi-paragraph list items in OPTIONS section

This patch makes the html docs right, makes the asciidoc docs a bit odd
but consistent with what is there already, and makes the manpages look
OK using docbook-xsl 1.68, but miss a paragraph separator when using 1.69.

For the manpages, current is like

       -A <author_file>
              Read a file with lines on the form

              username = User's Full Name <email@addr.es>

              and use "User's Full Name <email@addr.es>" as the GIT

With this patch, docbook-xsl v1.68 looks like

       -A <author_file>
              Read a file with lines on the form

                      username = User's Full Name <email@addr.es>

              and use "User's Full Name <email@addr.es>" as the GIT author and

while docbook-xsl v1.69 becomes

       -A <author_file>
              Read a file with lines on the form

                        username = User's Full Name <email@addr.es>
              and use "User's Full Name <email@addr.es>" as the GIT author and

The extra indentation is to keep the v1.69 manpage looking sane.

18 years agohttp-fetch: nicer warning for a server with unreliable 404 status
Junio C Hamano [Mon, 20 Mar 2006 22:07:59 +0000 (14:07 -0800)]
http-fetch: nicer warning for a server with unreliable 404 status

When a repository otherwise properly prepared is served by a
dumb HTTP server that sends "No such page" output with 200
status for human consumption to a request for a page that does
not exist, the users will get an alarming "File X corrupt" error
message.  Hint that they might be dealing with such a server at
the end and suggest running fsck-objects to check if the result
is OK (the pack-fallback code does the right thing in this case
so unless a loose object file was actually corrupt the result
should check OK).

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/merge' into next
Junio C Hamano [Mon, 20 Mar 2006 08:51:07 +0000 (00:51 -0800)]
Merge branch 'jc/merge' into next

* jc/merge:
  git-merge knows some strategies want to skip trivial merges
  generate-cmdlist: style cleanups.
  Add missing semicolon to sed command.
  unpack_delta_entry(): reduce memory footprint.
  git.el: Added a function to diff against the other heads in a merge.
  git.el: Get the default user name and email from the repository config.
  git.el: More robust handling of subprocess errors when returning strings.

18 years agoMerge branch 'jc/clone' into next
Junio C Hamano [Mon, 20 Mar 2006 08:50:58 +0000 (00:50 -0800)]
Merge branch 'jc/clone' into next

* jc/clone:
  revamp git-clone.

18 years agogit-merge knows some strategies want to skip trivial merges
Junio C Hamano [Sat, 18 Mar 2006 22:50:53 +0000 (14:50 -0800)]
git-merge knows some strategies want to skip trivial merges

Most notably "ours".  Also this makes sure we do not record
duplicated parents on the parent list of the resulting commit.

This is based on Mark Wooding's work, but does not change the UI
nor introduce new flags.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agorevamp git-clone.
Junio C Hamano [Mon, 20 Mar 2006 08:21:10 +0000 (00:21 -0800)]
revamp git-clone.

This does two things.

 * A new flag --reference can be used to name a local repository
   that is to be used as an alternate.  This is in response to
   an inquiry by James Cloos in the message on the list
   <m3r74ykue7.fsf@lugabout.cloos.reno.nv.us>.

 * A new flag --use-separate-remote stops contaminating local
   branch namespace by upstream branch names.  The upstream
   branch heads are copied in .git/refs/remotes/ instead of
   .git/refs/heads/ and .git/remotes/origin file is set up to
   reflect this as well.  It requires to have fetch/pull update
   to understand .git/refs/remotes by Eric Wong to further
   update the repository cloned this way.

For the former change, git-fetch-pack is taught a new flag --all
to fetch from all the remote heads.  Nobody uses the git-clone-pack
with this change, so we could deprecate the command, but removal
of the command will be left to a separate round.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogenerate-cmdlist: style cleanups.
Junio C Hamano [Mon, 20 Mar 2006 07:54:45 +0000 (23:54 -0800)]
generate-cmdlist: style cleanups.

Instead of giving multiple commands concatenated with semicolon
to sed, write them on separate lines.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoAdd missing semicolon to sed command.
Shawn Pearce [Mon, 20 Mar 2006 00:18:08 +0000 (19:18 -0500)]
Add missing semicolon to sed command.

generate-cmdlist.sh is giving errors messages from sed on Mac OS
10.4 due to a missing semicolon.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agounpack_delta_entry(): reduce memory footprint.
Junio C Hamano [Sun, 19 Mar 2006 21:43:42 +0000 (13:43 -0800)]
unpack_delta_entry(): reduce memory footprint.

Currently we unpack the delta data from the pack and then unpack
the base object to apply that delta data to it.  When getting an
object that is deeply deltified, we can reduce memory footprint
by unpacking the base object first and then unpacking the delta
data, because we will need to keep at most one delta data in
memory that way.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit.el: Added a function to diff against the other heads in a merge.
Alexandre Julliard [Sun, 19 Mar 2006 09:06:10 +0000 (10:06 +0100)]
git.el: Added a function to diff against the other heads in a merge.

git-diff-file-merge-head generates a diff against the first merge
head, or with a prefix argument against the nth head. Bound to `d h'
by default.

Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit.el: Get the default user name and email from the repository config.
Alexandre Julliard [Sun, 19 Mar 2006 09:05:48 +0000 (10:05 +0100)]
git.el: Get the default user name and email from the repository config.

If user name or email are not set explicitly, get them from the
user.name and user.email configuration values before falling back to
the Emacs defaults.

Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit.el: More robust handling of subprocess errors when returning strings.
Alexandre Julliard [Sun, 19 Mar 2006 09:05:22 +0000 (10:05 +0100)]
git.el: More robust handling of subprocess errors when returning strings.

Make sure that functions that call a git process and return a string
always return nil when the subprocess failed.

Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'master' into next
Junio C Hamano [Sat, 18 Mar 2006 22:58:20 +0000 (14:58 -0800)]
Merge branch 'master' into next

* master:
  Makefile: Add TAGS and tags targets
  ls-files: Don't require exclude files to end with a newline.

18 years agoMakefile: Add TAGS and tags targets
Fredrik Kuivinen [Sat, 18 Mar 2006 10:07:12 +0000 (11:07 +0100)]
Makefile: Add TAGS and tags targets

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agols-files: Don't require exclude files to end with a newline.
Alexandre Julliard [Sat, 18 Mar 2006 10:27:45 +0000 (11:27 +0100)]
ls-files: Don't require exclude files to end with a newline.

Without this patch, the last line of an exclude file is silently
ignored if it doesn't end with a newline.

Signed-off-by: Alexandre Julliard <julliard@winehq.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'master' into next
Junio C Hamano [Sat, 18 Mar 2006 10:08:10 +0000 (02:08 -0800)]
Merge branch 'master' into next

* master:
  git-pull: run repo-config with dash form.

18 years agogit-pull: run repo-config with dash form.
Junio C Hamano [Sat, 18 Mar 2006 10:07:59 +0000 (02:07 -0800)]
git-pull: run repo-config with dash form.

... as discussed on the list for consistency.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/cvsimport' into next
Junio C Hamano [Sat, 18 Mar 2006 10:05:22 +0000 (02:05 -0800)]
Merge branch 'jc/cvsimport' into next

* jc/cvsimport:
  cvsimport: fix reading from rev-parse

18 years agocvsimport: fix reading from rev-parse
Junio C Hamano [Sat, 18 Mar 2006 10:05:02 +0000 (02:05 -0800)]
cvsimport: fix reading from rev-parse

The updated code reads the tip of the current branch before and
after the import runs, but forgot to chomp what we read from the
command.  The read-tree command did not them with the trailing
LF.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/empty'
Junio C Hamano [Sat, 18 Mar 2006 08:43:47 +0000 (00:43 -0800)]
Merge branch 'jc/empty'

* jc/empty:
  revision traversal: --remove-empty fix (take #2).
  revision traversal: --remove-empty fix.

Conflicts:

revision.c (adjust for the updates by Fredrik)

18 years agoMerge branch 'master' into next
Junio C Hamano [Sat, 18 Mar 2006 04:43:15 +0000 (20:43 -0800)]
Merge branch 'master' into next

* master:
  3% tighter packs for free
  Rewrite synopsis to clarify the two primary uses of git-checkout.
  Fix minor typo.
  Reference git-commit-tree for env vars.
  Clarify git-rebase example commands.
  Document the default source of template files.
  Call out the two different uses of git-branch and fix a typo.
  Add git-show reference

18 years ago3% tighter packs for free
Nicolas Pitre [Sat, 18 Mar 2006 03:45:07 +0000 (22:45 -0500)]
3% tighter packs for free

This patch makes for 3.4% smaller pack with the git repository, and
a bit more than 3% smaller pack with the kernel repository.

And so with _no_ measurable CPU difference.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoRewrite synopsis to clarify the two primary uses of git-checkout.
Jon Loeliger [Sat, 18 Mar 2006 00:26:01 +0000 (18:26 -0600)]
Rewrite synopsis to clarify the two primary uses of git-checkout.

Fix a few typo/grammar problems.

Signed-off-by: Jon Loeliger <jdl@jdl.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoFix minor typo.
Jon Loeliger [Sat, 18 Mar 2006 00:25:46 +0000 (18:25 -0600)]
Fix minor typo.

Signed-off-by: Jon Loeliger <jdl@jdl.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoReference git-commit-tree for env vars.
Jon Loeliger [Sat, 18 Mar 2006 00:25:09 +0000 (18:25 -0600)]
Reference git-commit-tree for env vars.

Signed-off-by: Jon Loeliger <jdl@jdl.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoClarify git-rebase example commands.
Jon Loeliger [Sat, 18 Mar 2006 00:25:30 +0000 (18:25 -0600)]
Clarify git-rebase example commands.

Signed-off-by: Jon Loeliger <jdl@jdl.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDocument the default source of template files.
Jon Loeliger [Sat, 18 Mar 2006 00:24:49 +0000 (18:24 -0600)]
Document the default source of template files.

Also explain a bit more about how the template option works.

Signed-off-by: Jon Loeliger <jdl@jdl.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoCall out the two different uses of git-branch and fix a typo.
Jon Loeliger [Sat, 18 Mar 2006 00:24:37 +0000 (18:24 -0600)]
Call out the two different uses of git-branch and fix a typo.

Signed-off-by: Jon Loeliger <jdl@jdl.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoAdd git-show reference
Jon Loeliger [Sat, 18 Mar 2006 00:21:39 +0000 (18:21 -0600)]
Add git-show reference

Signed-off-by: Jon Loeliger <jdl@jdl.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/cvsimport' into next
Junio C Hamano [Fri, 17 Mar 2006 22:11:17 +0000 (14:11 -0800)]
Merge branch 'jc/cvsimport' into next

* jc/cvsimport:
  cvsimport: honor -i and non -i upon subsequent imports

18 years agoMerge branch 'jc/fetch' into next
Junio C Hamano [Fri, 17 Mar 2006 22:11:10 +0000 (14:11 -0800)]
Merge branch 'jc/fetch' into next

* jc/fetch:
  fetch: exit non-zero when fast-forward check fails.

18 years agoMerge branch 'ew/abbrev' into next
Junio C Hamano [Fri, 17 Mar 2006 22:11:00 +0000 (14:11 -0800)]
Merge branch 'ew/abbrev' into next

* ew/abbrev:
  ls-files: add --abbrev[=<n>] option
  ls-tree: add --abbrev[=<n>] option
  blame: Fix git-blame <directory>
  blame: Nicer output

18 years agofetch: exit non-zero when fast-forward check fails.
Junio C Hamano [Fri, 17 Mar 2006 06:04:18 +0000 (22:04 -0800)]
fetch: exit non-zero when fast-forward check fails.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agols-files: add --abbrev[=<n>] option
Eric Wong [Tue, 7 Mar 2006 19:59:17 +0000 (11:59 -0800)]
ls-files: add --abbrev[=<n>] option

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agols-tree: add --abbrev[=<n>] option
Eric Wong [Tue, 7 Mar 2006 13:52:02 +0000 (05:52 -0800)]
ls-tree: add --abbrev[=<n>] option

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agocvsimport: honor -i and non -i upon subsequent imports
Junio C Hamano [Fri, 17 Mar 2006 22:08:39 +0000 (14:08 -0800)]
cvsimport: honor -i and non -i upon subsequent imports

Documentation says -i is "import only", so without it,
subsequent import should update the current branch and working
tree files in a sensible way.

"A sensible way" defined by this commit is "act as if it is a
git pull from foreign repository which happens to be CVS not
git".  So:

 - If importing into the current branch (note that cvsimport
   requires the tracking branch is pristine -- you checked out
   the tracking branch but it is your responsibility not to make
   your own commits there), fast forward the branch head and
   match the index and working tree using two-way merge, just
   like "git pull" does.

 - If importing into a separate tracking branch, update that
   branch head, and merge it into your current branch, again,
   just like "git pull" does.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoblame: Fix git-blame <directory>
Fredrik Kuivinen [Fri, 17 Mar 2006 21:49:31 +0000 (22:49 +0100)]
blame: Fix git-blame <directory>

Before this patch git-blame <directory> gave non-sensible output. (It
assigned blame to some random file in <directory>) Abort with an error
message instead.

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoblame: Nicer output
Fredrik Kuivinen [Fri, 17 Mar 2006 21:49:28 +0000 (22:49 +0100)]
blame: Nicer output

As pointed out by Junio, it may be dangerous to cut off people's names
after 15 bytes. If the name is encoded in an encoding which uses more
than one byte per code point we may end up with outputting garbage.
Instead of trying to do something smart, just output the entire name.
We don't gain much screen space by chopping it off anyway.

Furthermore, only output the file name if we actually found any
renames.

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'master' into next
Junio C Hamano [Thu, 16 Mar 2006 00:15:23 +0000 (16:15 -0800)]
Merge branch 'master' into next

* master:
  fix imap-send for OSX
  Let merge set the default strategy.

18 years agofix imap-send for OSX
Randal L. Schwartz [Wed, 15 Mar 2006 23:35:06 +0000 (15:35 -0800)]
fix imap-send for OSX

This patch works... I've been using it to stay current.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoLet merge set the default strategy.
Mark Hollomon [Wed, 15 Mar 2006 22:51:41 +0000 (22:51 +0000)]
Let merge set the default strategy.

If the user does not set a merge strategy for git-pull,
let git-merge calculate a default strategy.

[jc: with minor stylistic tweaks]

Signed-off-by: Mark Hollomon <markhollomon@comcast.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'lt/diff' into next
Junio C Hamano [Wed, 15 Mar 2006 21:19:52 +0000 (13:19 -0800)]
Merge branch 'lt/diff' into next

* lt/diff:
  diffcore-delta: 64-byte-or-EOL ultrafast replacement (hash fix).

18 years agodiffcore-delta: 64-byte-or-EOL ultrafast replacement (hash fix).
Linus Torvalds [Wed, 15 Mar 2006 08:37:57 +0000 (00:37 -0800)]
diffcore-delta: 64-byte-or-EOL ultrafast replacement (hash fix).

The rotating 64-bit number was not really rotating, and worse
yet ulong was longer than 64-bit on 64-bit architectures X-<.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'master' into next
Junio C Hamano [Wed, 15 Mar 2006 17:12:18 +0000 (09:12 -0800)]
Merge branch 'master' into next

* master:
  Fix broken slot reuse when fetching alternates

18 years agoFix broken slot reuse when fetching alternates
Nick Hengeveld [Wed, 15 Mar 2006 16:59:52 +0000 (08:59 -0800)]
Fix broken slot reuse when fetching alternates

When fetching alternates, http-fetch may reuse the slot to fetch non-http
alternates if http-alternates does not exist.  When doing so, it now needs
to update the slot's finished status so run_active_slot waits for the
non-http alternates request to finish.

Signed-off-by: Nick Hengeveld <nickh@reactrix.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'lt/diff' into next
Junio C Hamano [Wed, 15 Mar 2006 08:38:47 +0000 (00:38 -0800)]
Merge branch 'lt/diff' into next

* lt/diff:
  diffcore-delta: 64-byte-or-EOL ultrafast replacement.

18 years agodiffcore-delta: 64-byte-or-EOL ultrafast replacement.
Linus Torvalds [Wed, 15 Mar 2006 08:37:57 +0000 (00:37 -0800)]
diffcore-delta: 64-byte-or-EOL ultrafast replacement.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/pack'
Junio C Hamano [Mon, 13 Mar 2006 08:04:10 +0000 (00:04 -0800)]
Merge branch 'jc/pack'

* jc/pack:
  pack-objects: simplify "thin" pack.
  verify-pack -v: show delta-chain histogram.

18 years agoMerge branch 'jc/fsck'
Junio C Hamano [Mon, 13 Mar 2006 08:04:05 +0000 (00:04 -0800)]
Merge branch 'jc/fsck'

* jc/fsck:
  fsck-objects: Remove --standalone

18 years agoMerge branch 'nh/http'
Junio C Hamano [Mon, 13 Mar 2006 08:01:57 +0000 (00:01 -0800)]
Merge branch 'nh/http'

* nh/http:
  http-push: cleanup
  http-push: support for updating remote info/refs
  http-push: improve remote lock management
  http-push: refactor remote file/directory processing
  HTTP slot reuse fixes
  http-push: fix revision walk

18 years agoMerge branch 'fk/blame'
Junio C Hamano [Mon, 13 Mar 2006 08:01:52 +0000 (00:01 -0800)]
Merge branch 'fk/blame'

* fk/blame:
  blame: Rename detection (take 2)
  rev-lib: Make it easy to do rename tracking (take 2)
  Make it possible to not clobber object.util in sort_in_topological_order (take 2)

18 years agoMerge branch 'jc/diff' into next
Junio C Hamano [Mon, 13 Mar 2006 07:45:42 +0000 (23:45 -0800)]
Merge branch 'jc/diff' into next

* jc/diff:
  Fix up diffcore-rename scoring

18 years agoFix up diffcore-rename scoring
Linus Torvalds [Mon, 13 Mar 2006 06:26:34 +0000 (22:26 -0800)]
Fix up diffcore-rename scoring

The "score" calculation for diffcore-rename was totally broken.

It scaled "score" as

score = src_copied * MAX_SCORE / dst->size;

which means that you got a 100% similarity score even if src and dest were
different, if just every byte of dst was copied from src, even if source
was much larger than dst (eg we had copied 85% of the bytes, but _deleted_
the remaining 15%).

That's clearly bogus. We should do the score calculation relative not to
the destination size, but to the max size of the two.

This seems to fix it.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/diff' into next
Junio C Hamano [Mon, 13 Mar 2006 04:42:35 +0000 (20:42 -0800)]
Merge branch 'jc/diff' into next

* jc/diff:
  diffcore-delta: tweak hashbase value.

18 years agodiffcore-delta: tweak hashbase value.
Junio C Hamano [Mon, 13 Mar 2006 04:32:06 +0000 (20:32 -0800)]
diffcore-delta: tweak hashbase value.

This tweaks the maximum hashvalue we use to hash the string into
without making the maximum size of the hashtable can grow from
the current limit.  With this, the renames detected becomes a
bit more precise without incurring additional paging cost.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/diff' into next
Junio C Hamano [Mon, 13 Mar 2006 01:27:32 +0000 (17:27 -0800)]
Merge branch 'jc/diff' into next

* jc/diff:
  diffcore-delta: make the hash a bit denser.

18 years agoMerge branch 'jc/empty' into next
Junio C Hamano [Mon, 13 Mar 2006 01:27:23 +0000 (17:27 -0800)]
Merge branch 'jc/empty' into next

* jc/empty:
  revision traversal: --remove-empty fix (take #2).

18 years agodiffcore-delta: make the hash a bit denser.
Junio C Hamano [Mon, 13 Mar 2006 00:39:51 +0000 (16:39 -0800)]
diffcore-delta: make the hash a bit denser.

To reduce wasted memory, wait until the hash fills up more
densely before we rehash.  This reduces the working set size a
bit further.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agorevision traversal: --remove-empty fix (take #2).
Junio C Hamano [Sun, 12 Mar 2006 21:39:31 +0000 (13:39 -0800)]
revision traversal: --remove-empty fix (take #2).

Marco Costalba reports that --remove-empty omits the commit that
created paths we are interested in.  try_to_simplify_commit()
logic was dropping a parent we introduced those paths against,
which I think is not what we meant.  Instead, this makes such
parent parentless.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/empty' into next
Junio C Hamano [Sun, 12 Mar 2006 21:43:36 +0000 (13:43 -0800)]
Merge branch 'jc/empty' into next

* jc/empty:
  revision traversal: --remove-empty fix.
  annotate-tests: override VISUAL when running tests.

18 years agorevision traversal: --remove-empty fix.
Junio C Hamano [Sun, 12 Mar 2006 21:39:31 +0000 (13:39 -0800)]
revision traversal: --remove-empty fix.

Marco Costalba reports that --remove-empty omits the commit that
created paths we are interested in.  try_to_simplify_commit()
logic was dropping a parent we introduced those paths against,
which I think is not what we meant.  Instead, this marks such
parent uninteresting.  The traversal does not go beyond that
parent as advertised, but we still say that the current commit
changed things from that parent.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoannotate-tests: override VISUAL when running tests.
Mark Wooding [Sun, 12 Mar 2006 14:00:30 +0000 (14:00 +0000)]
annotate-tests: override VISUAL when running tests.

The tests hang for me waiting for Emacs with its output directed
somewhere strage, because I hedged my bets and set both EDITOR and
VISUAL to run Emacs.

Signed-off-by: Mark Wooding <mdw@distorted.org.uk>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'master' into next
Junio C Hamano [Sun, 12 Mar 2006 11:29:11 +0000 (03:29 -0800)]
Merge branch 'master' into next

* master:
  imap-send: Add missing #include for macosx
  git-diff: -p disables rename detection.
  imap-send: cleanup execl() call to use NULL sentinel instead of 0
  annotate.perl triggers rpm bug

18 years agoMerge branch 'jc/diff' into next
Junio C Hamano [Sun, 12 Mar 2006 11:29:09 +0000 (03:29 -0800)]
Merge branch 'jc/diff' into next

* jc/diff:
  diffcore-rename: somewhat optimized.

18 years agoimap-send: Add missing #include for macosx
Johannes Schindelin [Sun, 12 Mar 2006 10:55:53 +0000 (11:55 +0100)]
imap-send: Add missing #include for macosx

There is a compile error without that.

Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agodiffcore-rename: somewhat optimized.
Junio C Hamano [Sun, 12 Mar 2006 11:22:10 +0000 (03:22 -0800)]
diffcore-rename: somewhat optimized.

This changes diffcore-rename to reuse statistics information
gathered during similarity estimation, and updates the hashtable
implementation used to keep track of the statistics to be
denser.  This seems to give better performance.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agogit-diff: -p disables rename detection.
Junio C Hamano [Sun, 12 Mar 2006 01:44:10 +0000 (17:44 -0800)]
git-diff: -p disables rename detection.

18 years agoimap-send: cleanup execl() call to use NULL sentinel instead of 0
Marco Roeland [Sat, 11 Mar 2006 08:55:50 +0000 (09:55 +0100)]
imap-send: cleanup execl() call to use NULL sentinel instead of 0

Some versions of gcc check that calls to the exec() family have the proper
sentinel for variadic calls. This should be (char *) NULL according to the
man page. Although for all other purposes the 0 is equivalent, gcc
nevertheless does emit a warning for 0 and not for NULL. This also makes the
usage consistent throughout git.

The whitespace in function calls throughout imap-send.c has its own style,
so I left it that way.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoannotate.perl triggers rpm bug
sean [Sat, 11 Mar 2006 07:39:28 +0000 (02:39 -0500)]
annotate.perl triggers rpm bug

RPM, at least on Fedora boxes, automatically creates a
dependency for any perl "use" lines, and one of the help text
lines unfortunately begins like this:

    -S, --rev-file revs-file
            use revs from revs-file instead of calling git-rev-list

RPM gets confused and creates a false dependecy for the
nonexistent perl package "revs".  Obviously this creates a
problem when someone goes to install the git-core rpm.

Since other help sentences all start with capital letter, make
this one match them by upcasing "Use".  As a side effect, RPM
stops getting confused.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'nh/http' into next
Junio C Hamano [Sat, 11 Mar 2006 07:02:23 +0000 (23:02 -0800)]
Merge branch 'nh/http' into next

* nh/http:
  http-push: cleanup
  http-push: support for updating remote info/refs
  http-push: improve remote lock management
  http-push: refactor remote file/directory processing
  HTTP slot reuse fixes
  http-push: fix revision walk

18 years agohttp-push: cleanup
Nick Hengeveld [Sat, 11 Mar 2006 04:18:33 +0000 (20:18 -0800)]
http-push: cleanup

More consistent usage string, condense push output, remove extra slashes
in URLs, fix unused variables, include HTTP method name in failure
messages.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp-push: support for updating remote info/refs
Nick Hengeveld [Sat, 11 Mar 2006 04:18:18 +0000 (20:18 -0800)]
http-push: support for updating remote info/refs

If info/refs exists on the remote, get a lock on info/refs, make sure that
there is a local copy of the object referenced in each remote ref (in case
someone else added a tag we don't have locally), do all the refspec updates,
and generate and send an updated info/refs file.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp-push: improve remote lock management
Nick Hengeveld [Sat, 11 Mar 2006 04:18:12 +0000 (20:18 -0800)]
http-push: improve remote lock management

Associate the remote locks with the remote repo, add a function to check
and refresh all current locks.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp-push: refactor remote file/directory processing
Nick Hengeveld [Sat, 11 Mar 2006 04:18:08 +0000 (20:18 -0800)]
http-push: refactor remote file/directory processing

Replace single-use functions with one that can get a list of remote
collections and pass file/directory information to user-defined functions
for processing.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoHTTP slot reuse fixes
Nick Hengeveld [Sat, 11 Mar 2006 04:18:01 +0000 (20:18 -0800)]
HTTP slot reuse fixes

Incorporate into http-push a fix related to accessing slot results after
the slot was reused, and fix a case in run_active_slot where a
finished slot wasn't detected if the slot was reused.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agohttp-push: fix revision walk
Nick Hengeveld [Sat, 11 Mar 2006 04:17:55 +0000 (20:17 -0800)]
http-push: fix revision walk

The revision walk was not including tags because setup_revisions zeroes out
the revs flags.  Pass --objects so it picks up all the necessary bits.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'ew/remote' into next
Junio C Hamano [Sat, 11 Mar 2006 06:33:06 +0000 (22:33 -0800)]
Merge branch 'ew/remote' into next

* ew/remote:
  fetch,parse-remote,fmt-merge-msg: refs/remotes/* support

18 years agoMerge branch 'fk/blame' into next
Junio C Hamano [Sat, 11 Mar 2006 06:32:59 +0000 (22:32 -0800)]
Merge branch 'fk/blame' into next

* fk/blame:
  blame: Rename detection (take 2)
  rev-lib: Make it easy to do rename tracking (take 2)
  Make it possible to not clobber object.util in sort_in_topological_order (take 2)
  Add git-imap-send, derived from isync 1.0.1.
  repack: prune loose objects when -d is given
  try_to_simplify_commit(): do not skip inspecting tree change at boundary.
  Fix t1200 test for breakage caused by removal of full-stop at the end of fast-forward message.
  Describe how to add extra mail header lines in mail generated by git-format-patch.
  Document the --attach flag.
  allow double click on current HEAD id after git-pull

18 years agofetch,parse-remote,fmt-merge-msg: refs/remotes/* support
Eric Wong [Fri, 10 Mar 2006 12:19:07 +0000 (04:19 -0800)]
fetch,parse-remote,fmt-merge-msg: refs/remotes/* support

We can now easily fetch and merge things from heads in the
refs/remotes/ hierarchy in remote repositories.

The refs/remotes/ hierarchy is likely to become the standard for
tracking foreign SCMs, as well as the location of Pull: targets
for tracking remote branches in newly cloned repositories.

Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoblame: Rename detection (take 2)
Fredrik Kuivinen [Fri, 10 Mar 2006 09:21:41 +0000 (10:21 +0100)]
blame: Rename detection (take 2)

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agorev-lib: Make it easy to do rename tracking (take 2)
Fredrik Kuivinen [Fri, 10 Mar 2006 09:21:39 +0000 (10:21 +0100)]
rev-lib: Make it easy to do rename tracking (take 2)

prune_fn in the rev_info structure is called in place of
try_to_simplify_commit. This makes it possible to do rename tracking
with a custom try_to_simplify_commit-like function.

This commit also introduces init_revisions which initialises the rev_info
structure with default values.

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMake it possible to not clobber object.util in sort_in_topological_order (take 2)
Fredrik Kuivinen [Fri, 10 Mar 2006 09:21:37 +0000 (10:21 +0100)]
Make it possible to not clobber object.util in sort_in_topological_order (take 2)

Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se>
Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoAdd git-imap-send, derived from isync 1.0.1.
Mike McCormack [Fri, 10 Mar 2006 05:32:50 +0000 (14:32 +0900)]
Add git-imap-send, derived from isync 1.0.1.

git-imap-send drops a patch series generated by git-format-patch into an
IMAP folder. This allows patch submitters to send patches through their
own mail program.

git-imap-send uses the following values from the GIT repository
configuration:

The target IMAP folder:

[imap]
         Folder = "INBOX.Drafts"

A command to open an ssh tunnel to the imap mail server.

[imap]
         Tunnel = "ssh -q user@imap.server.com /usr/bin/imapd ./Maildir
2> /dev/null"

[imap]
         Host = imap.server.com
         User = bob
         Password = pwd
         Port = 143

18 years agorepack: prune loose objects when -d is given
Junio C Hamano [Sat, 11 Mar 2006 06:05:47 +0000 (22:05 -0800)]
repack: prune loose objects when -d is given

[jc: the request originally came from Martin Atukunda, which was
 improved further by Alex Riesen]

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agotry_to_simplify_commit(): do not skip inspecting tree change at boundary.
Junio C Hamano [Sat, 11 Mar 2006 05:59:37 +0000 (21:59 -0800)]
try_to_simplify_commit(): do not skip inspecting tree change at boundary.

When git-rev-list (and git-log) collapsed ancestry chain to
commits that touch specified paths, we failed to inspect and
notice tree changes when we are about to hit uninteresting
parent.  This resulted in "git rev-list since.. -- file" to
always show the child commit after the lower bound, even if it
does not touch the file.  This commit fixes it.

Thanks for Catalin for reporting this.

See also:
461cf59f8924f174d7a0dcc3d77f576d93ed29a4

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoFix t1200 test for breakage caused by removal of full-stop at the end of fast-forward...
Junio C Hamano [Fri, 10 Mar 2006 06:51:21 +0000 (22:51 -0800)]
Fix t1200 test for breakage caused by removal of full-stop at the end of fast-forward message.

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoDescribe how to add extra mail header lines in mail generated by git-format-patch.
Mike McCormack [Fri, 10 Mar 2006 04:47:59 +0000 (13:47 +0900)]
Describe how to add extra mail header lines in mail generated by git-format-patch.

18 years agoDocument the --attach flag.
Mike McCormack [Fri, 10 Mar 2006 04:47:24 +0000 (13:47 +0900)]
Document the --attach flag.

18 years agoallow double click on current HEAD id after git-pull
Olaf Hering [Sat, 11 Feb 2006 11:26:30 +0000 (12:26 +0100)]
allow double click on current HEAD id after git-pull

Double click on to current HEAD commit id is not possible,
the dot has to go.

[jc: by popular requests.]

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agoMerge branch 'jc/fsck' into next
Junio C Hamano [Thu, 9 Mar 2006 21:10:50 +0000 (13:10 -0800)]
Merge branch 'jc/fsck' into next

* jc/fsck:
  fsck-objects: Remove --standalone
  refs.c::do_for_each_ref(): Finish error message lines with "\n"
  Nicer output from 'git'
  Use #define ARRAY_SIZE(x) (sizeof(x)/sizeof(x[0]))
  Remove trailing dot after short description
  Fix some inconsistencies in the docs
  contrib/git-svn: fix a harmless warning on rebuild (with old repos)
  contrib/git-svn: remove the --no-stop-on-copy flag
  contrib/git-svn: fix svn compat and fetch args
  Don't recurse into parents marked uninteresting.
  diff-delta: bound hash list length to avoid O(m*n) behavior
  test-delta needs zlib to compile
  git-fmt-merge-msg cleanup

18 years agofsck-objects: Remove --standalone
Junio C Hamano [Thu, 9 Mar 2006 09:44:19 +0000 (01:44 -0800)]
fsck-objects: Remove --standalone

The fsck-objects command (back then it was called fsck-cache)
used to complain if objects referred to by files in .git/refs/
or objects stored in files under .git/objects/??/ were not found
as stand-alone SHA1 files (i.e.  found in alternate object pools
or packed archives stored under .git/objects/pack).  Back then,
packs and alternates were new curiosity and having everything as
loose objects were the norm.

When we adjusted the behaviour of fsck-cache to consider objects
found in packs are OK, we introduced the --standalone flag as a
backward compatibility measure.

It still correctly checks if your repository is complete and
consists only of loose objects, so in that sense it is doing the
"right" thing, but checking that is pointless these days.  This
commit removes --standalone flag.

See also:

23676d407c63a6f67f8ce3ff192199bda03e6a03
8a498a05c3c6b2f53db669b24f36257ab213eb4c

Signed-off-by: Junio C Hamano <junkio@cox.net>
18 years agorefs.c::do_for_each_ref(): Finish error message lines with "\n"
Junio C Hamano [Thu, 9 Mar 2006 20:59:16 +0000 (12:59 -0800)]
refs.c::do_for_each_ref(): Finish error message lines with "\n"

We used fprintf() to show an error message without terminating
it with LF; use error() for that.

cf. c401cb48e77459a4ccad76888ad31bef252facc5

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