Code

git.git
16 years agobuiltin-fetch.c (store_updated_refs): Honor update_local_ref() return value
Dmitry V. Levin [Wed, 28 May 2008 15:29:36 +0000 (19:29 +0400)]
builtin-fetch.c (store_updated_refs): Honor update_local_ref() return value

Sync with builtin-fetch--tool.c where append_fetch_head()
honors update_local_ref() return value.

This fixes non fast forward fetch exit status,
http://bugzilla.altlinux.org/show_bug.cgi?id=15037

Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agofix sha1_pack_index_name()
Junio C Hamano [Wed, 28 May 2008 16:47:43 +0000 (09:47 -0700)]
fix sha1_pack_index_name()

An earlier commit 633f43e (Remove redundant code, eliminate one static
variable, 2008-05-24) had a thinko (perhaps an eyeno) that broke
sha1_pack_index_name() function.  One symptom of this was that the http
walker is now completely broken.

This should fix it.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoGit::cat_blob: allow using an empty blob to fix git-svn breakage
Junio C Hamano [Wed, 28 May 2008 06:33:22 +0000 (23:33 -0700)]
Git::cat_blob: allow using an empty blob to fix git-svn breakage

Recent "git-svn optimization" series introduced Git::cat_blob() subroutine
whose interface was broken in that it returned the size of the blob but
signalled an error by returning 0.  You can never use an empty blob with
such an interface.

This fixes the interface to return a negative value to signal an error.

Reported by Björn Steinbrink.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agot5100: Avoid filename "nul"
Junio C Hamano [Wed, 28 May 2008 06:12:29 +0000 (23:12 -0700)]
t5100: Avoid filename "nul"

There are broken filesystems that cannot have a file whose name is "nul"
anywhere on it.  Rename the test file to make ourselves more portable.

Noticed by Mark Levedahl.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge 1.5.5.3 in
Junio C Hamano [Wed, 28 May 2008 05:22:47 +0000 (22:22 -0700)]
Merge 1.5.5.3 in

16 years agoGIT 1.5.5.3 v1.5.5.3
Junio C Hamano [Wed, 28 May 2008 05:20:53 +0000 (22:20 -0700)]
GIT 1.5.5.3

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agocommit --interactive: properly update the index before commiting
Gerrit Pape [Tue, 27 May 2008 08:59:16 +0000 (08:59 +0000)]
commit --interactive: properly update the index before commiting

When adding files through git commit --interactive, and 'quit'
afterwards, the message in the editor of the commit message indicates
that many (maybe all) files are deleted from the tree.  Dismissing that
and running git commit afterwards does the right thing.  This commit
fixes git commit --interactive to properly update the index before
commiting.

Reported by Jiří Paleček through
 http://bugs.debian.org/480429

Signed-off-by: Gerrit Pape <pape@smarden.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogitweb: only display "next" links in logs if there is a next page
Lea Wiemann [Tue, 27 May 2008 23:25:42 +0000 (01:25 +0200)]
gitweb: only display "next" links in logs if there is a next page

There was a bug in the implementation of the "next" links in
format_paging_nav (for log and shortlog), which caused the next links
to always be displayed, even if there is no next page.  This fixes it.

Signed-off-by: Lea Wiemann <LeWiemann@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoRevert "filter-branch: subdirectory filter needs --full-history"
Johannes Sixt [Mon, 26 May 2008 19:09:18 +0000 (21:09 +0200)]
Revert "filter-branch: subdirectory filter needs --full-history"

This reverts commit cfabd6eee1745cfec58cfcb794ce8847e43b888a. I had
implemented it without understanding what --full-history does. Consider
this history:

    C--M--N
   /  /  /
  A--B  /
   \   /
    D-/

where B and C modify a path, X, in the same way so that the result is
identical, and D does not modify it at all. With the path limiter X and
without --full-history this is simplified to

   A--B

i.e. only one of the paths via B or C is chosen. I had assumed that
--full-history would keep both paths like this

    C--M
   /  /
  A--B

removing the path via D; but in fact it keeps the entire history.

Currently, git does not have the capability to simplify to this
intermediary case. However, the other extreme to keep the entire history
is not wanted either in usual cases. I think we can expect that histories
like the above are rare, and in the usual cases we want a simplified
history. So let's remove --full-history again.

(Concerning t7003, subsequent tests depend on what the test case sets up,
so we can't just back out the entire test case.)

Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation/git-bundle.txt: fix synopsis
Gerrit Pape [Tue, 27 May 2008 08:51:22 +0000 (08:51 +0000)]
Documentation/git-bundle.txt: fix synopsis

The <git-rev-list args> are mandatory to git bundle create, not
optional.  The usage output of git bundle is already right on this.

Signed-off-by: Gerrit Pape <pape@smarden.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoclone: make sure we support the transport type
Jeff King [Tue, 27 May 2008 14:28:43 +0000 (10:28 -0400)]
clone: make sure we support the transport type

If we use an unsupported transport (e.g., http when curl
support is not compiled in), transport_get reports an error
to the user, but we still get a transport object. We need to
manually check and abort the clone process at that point, or
we end up with a segfault.

Noticed by Thomas Rast.

Signed-off-by: Jeff King <peff@peff.net>
Acked-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agot5700-clone-reference: Quote $U
Johannes Sixt [Tue, 27 May 2008 07:36:22 +0000 (09:36 +0200)]
t5700-clone-reference: Quote $U

The new "trash directory" bites again.

Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoAdd some tests for git update-ref -d
Karl Hasselström [Sun, 25 May 2008 16:14:29 +0000 (18:14 +0200)]
Add some tests for git update-ref -d

Signed-off-by: Karl Hasselström <kha@treskal.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'jc/diff-no-no-index'
Junio C Hamano [Tue, 27 May 2008 05:38:19 +0000 (22:38 -0700)]
Merge branch 'jc/diff-no-no-index'

* jc/diff-no-no-index:
  git diff --no-index: default to page like other diff frontends
  git-diff: allow  --no-index semantics a bit more
  "git diff": do not ignore index without --no-index
  diff-files: do not play --no-index games
  tests: do not use implicit "git diff --no-index"

16 years agogit diff --no-index: default to page like other diff frontends
Junio C Hamano [Tue, 27 May 2008 05:35:07 +0000 (22:35 -0700)]
git diff --no-index: default to page like other diff frontends

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-diff: allow --no-index semantics a bit more
Junio C Hamano [Tue, 27 May 2008 04:54:23 +0000 (21:54 -0700)]
git-diff: allow  --no-index semantics a bit more

Even when inside a git work tree, if two paths are given and at least one
is clearly outside the work tree, it cannot be a request to diff a tracked
path anyway; allow such an invocation to use --no-index semantics.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint'
Junio C Hamano [Tue, 27 May 2008 02:49:01 +0000 (19:49 -0700)]
Merge branch 'maint'

* maint:
  Documentation: fix graph in git-rev-parse.txt
  show-branch --current: do not barf on detached HEAD

16 years agoMerge branch 'gp/bisect-fix' into maint
Junio C Hamano [Tue, 27 May 2008 02:15:03 +0000 (19:15 -0700)]
Merge branch 'gp/bisect-fix' into maint

* gp/bisect-fix:
  bisect: print an error message when "git rev-list --bisect-vars" fails
  git-bisect.sh: don't accidentally override existing branch "bisect"

16 years agoMerge git://repo.or.cz/git-gui
Junio C Hamano [Tue, 27 May 2008 02:10:43 +0000 (19:10 -0700)]
Merge git://repo.or.cz/git-gui

* git://repo.or.cz/git-gui:
  git-gui: Vertically align textboxes with labels
  git-gui: Handle workdir detection when CYGWIN=nowinsymlinks
  git-gui: Add a --trace command line option

16 years agoDocumentation: fix graph in git-rev-parse.txt
Michele Ballabio [Mon, 26 May 2008 13:24:02 +0000 (15:24 +0200)]
Documentation: fix graph in git-rev-parse.txt

Preformatted html and man pages show a mangled graph, caused by a
backslash.

Commit f1ec6b22a8c1ab1cca0f1875f85aea5d2434e5a6 fixed this same issue,
but it seems that new versions of the Asciidoc toolchain changed their
behaviour.

Signed-off-by: Michele Ballabio <barra_cuda@katamail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-gui: Vertically align textboxes with labels gitgui-0.10.2
Twiinz [Mon, 19 May 2008 04:01:58 +0000 (00:01 -0400)]
git-gui: Vertically align textboxes with labels

In git-gui after clicking either on 'Create New Repository' or
'Open Existing Repository' the form elements aren't centered like
they are pretty much everywhere else in the app. At least when ran
on a mac, haven't checked on other platforms.

Using grid instead of pack seems to fix this.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
16 years agoshow-branch --current: do not barf on detached HEAD
Junio C Hamano [Mon, 26 May 2008 22:09:51 +0000 (15:09 -0700)]
show-branch --current: do not barf on detached HEAD

The code assumed that there always is the current branch, but the result
from resolve_ref() on detached HEAD does not even start with "refs/heads/".

Originally noticed and fixed by Stephan Beyer.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMakefile: fix dependency on wt-status.h
Junio C Hamano [Mon, 26 May 2008 21:56:19 +0000 (14:56 -0700)]
Makefile: fix dependency on wt-status.h

Noticed by Hannes, reported by Dscho.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agohg-to-git: add --verbose option
Johannes Schindelin [Mon, 26 May 2008 13:20:54 +0000 (14:20 +0100)]
hg-to-git: add --verbose option

This patch adds an option to make hg-to-git quiet by default.  Note:
it only suppresses those messages that would be printed when everything
was up-to-date.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Acked-by: Stelian Pop <stelian@popies.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation/git.txt: link to 1.5.5.2 documentation.
Junio C Hamano [Mon, 26 May 2008 04:14:09 +0000 (21:14 -0700)]
Documentation/git.txt: link to 1.5.5.2 documentation.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint'
Junio C Hamano [Mon, 26 May 2008 05:43:05 +0000 (22:43 -0700)]
Merge branch 'maint'

* maint:
  builtin-fast-export: Only output a single parent per line
  Release Notes for 1.5.5.2

Conflicts:

RelNotes

16 years agoMerge branch 'jk/maint-send-email-compose' into maint
Junio C Hamano [Mon, 26 May 2008 05:34:23 +0000 (22:34 -0700)]
Merge branch 'jk/maint-send-email-compose' into maint

* jk/maint-send-email-compose:
  send-email: rfc2047-quote subject lines with non-ascii characters
  send-email: specify content-type of --compose body

16 years agoMerge branch 'hb/maint-send-email-quote-recipients' into maint
Junio C Hamano [Mon, 26 May 2008 05:34:20 +0000 (22:34 -0700)]
Merge branch 'hb/maint-send-email-quote-recipients' into maint

* hb/maint-send-email-quote-recipients:
  Fix recipient santitization

16 years agoMerge branch 'maint-1.5.4' into maint
Junio C Hamano [Mon, 26 May 2008 05:34:05 +0000 (22:34 -0700)]
Merge branch 'maint-1.5.4' into maint

* maint-1.5.4:
  builtin-fast-export: Only output a single parent per line

16 years agobuiltin-fast-export: Only output a single parent per line
Pieter de Bie [Sat, 24 May 2008 23:21:53 +0000 (01:21 +0200)]
builtin-fast-export: Only output a single parent per line

According to the git-fast-import man-page, you can only put a single
committish per merge: line, like this:

  merge :10
  merge :11

However, git-fast-export puts all parents on a single line, like this:

  merge :10 :11

This changes fast-export to output a single parent per line.  Otherwise
neither git-fast-import nor bzr-fast-import can read its output.

[jc: fix-up to remove excess LF in the output that makes fast-import barf]

Signed-off-by: Pieter de Bie <pdebie@ai.rug.nl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoRelease Notes for 1.5.5.2
Junio C Hamano [Mon, 26 May 2008 04:11:24 +0000 (21:11 -0700)]
Release Notes for 1.5.5.2

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoIgnore no-op changes in paranoid update hook
Shawn O. Pearce [Mon, 26 May 2008 02:18:05 +0000 (22:18 -0400)]
Ignore no-op changes in paranoid update hook

If the hook gets invoked with identical old and new ids there
is no change taking place.  We probably should not have been
called, but instead of failing silently allow the no-op.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDon't load missing ACL files in paranoid update hook
Shawn O. Pearce [Mon, 26 May 2008 02:18:01 +0000 (22:18 -0400)]
Don't load missing ACL files in paranoid update hook

If a user or group ACL file does not exist in the current tip
revision of the acl repository we will get an error from cat-file
when we ask for that blob as it cannot be resolved.  A quick look
at the history by rev-list can tell us if there is a path there
or not.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDon't diff empty tree on branch creation in paranoid update hook
Shawn O. Pearce [Mon, 26 May 2008 02:17:57 +0000 (22:17 -0400)]
Don't diff empty tree on branch creation in paranoid update hook

Listing all files in a branch during branch creation is silly;
the user's file-level ACLs probably don't mean anything at this
point.  We now treat the base case of 0{40} as an empty diff,
as this happens only when the user is creating the branch and
there are file level ACLs that diff against the old value of
the branch.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agolog --pretty: do not accept bogus "--prettyshort"
Junio C Hamano [Mon, 26 May 2008 03:19:30 +0000 (20:19 -0700)]
log --pretty: do not accept bogus "--prettyshort"

... nor bogus "format.pretty = '=short'".  Both are syntax errors.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agolog --graph: do not accept log --graphbogus
Junio C Hamano [Mon, 26 May 2008 03:19:25 +0000 (20:19 -0700)]
log --graph: do not accept log --graphbogus

An obvious fix to the argument parser.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoUpdate draft release notes for 1.5.6
Junio C Hamano [Mon, 26 May 2008 01:28:37 +0000 (18:28 -0700)]
Update draft release notes for 1.5.6

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'js/config-cb' v1.5.6-rc0
Junio C Hamano [Sun, 25 May 2008 21:25:02 +0000 (14:25 -0700)]
Merge branch 'js/config-cb'

* js/config-cb:
  Provide git_config with a callback-data parameter

Conflicts:

builtin-add.c
builtin-cat-file.c

16 years agoMerge branch 'as/graph'
Junio C Hamano [Sun, 25 May 2008 21:05:11 +0000 (14:05 -0700)]
Merge branch 'as/graph'

* as/graph:
  get_revision(): honor the topo_order flag for boundary commits
  Fix output of "git log --graph --boundary"
  log --graph --left-right: show left/right information in place of '*'
  graph API: don't print branch lines for uninteresting merge parents
  graph API: fix graph mis-alignment after uninteresting commits

16 years agoMerge branch 'js/mailinfo'
Junio C Hamano [Sun, 25 May 2008 21:05:09 +0000 (14:05 -0700)]
Merge branch 'js/mailinfo'

* js/mailinfo:
  mailsplit: minor clean-up in read_line_with_nul()
  mailinfo: apply the same fix not to lose NULs in BASE64 and QP codepaths
  mailsplit and mailinfo: gracefully handle NUL characters

16 years agoMerge branch 'jc/add-n-u'
Junio C Hamano [Sun, 25 May 2008 21:03:50 +0000 (14:03 -0700)]
Merge branch 'jc/add-n-u'

* jc/add-n-u:
  Make git add -n and git -u -n output consistent
  "git-add -n -u" should not add but just report

Conflicts:

builtin-add.c
builtin-mv.c
cache.h
read-cache.c

16 years agoMerge branch 'ar/t6031'
Junio C Hamano [Sun, 25 May 2008 20:49:16 +0000 (13:49 -0700)]
Merge branch 'ar/t6031'

* ar/t6031:
  Fix t6031 on filesystems without working exec bit

16 years agoMerge branch 'db/clone-in-c'
Junio C Hamano [Sun, 25 May 2008 20:38:44 +0000 (13:38 -0700)]
Merge branch 'db/clone-in-c'

* db/clone-in-c:
  Add test for cloning with "--reference" repo being a subset of source repo
  Add a test for another combination of --reference
  Test that --reference actually suppresses fetching referenced objects
  clone: fall back to copying if hardlinking fails
  builtin-clone.c: Need to closedir() in copy_or_link_directory()
  builtin-clone: fix initial checkout
  Build in clone
  Provide API access to init_db()
  Add a function to set a non-default work tree
  Allow for having for_each_ref() list extra refs
  Have a constant extern refspec for "--tags"
  Add a library function to add an alternate to the alternates file
  Add a lockfile function to append to a file
  Mark the list of refs to fetch as const

Conflicts:

cache.h
t/t5700-clone-reference.sh

16 years agoMerge branch 'jc/apply-whitespace'
Junio C Hamano [Sun, 25 May 2008 20:38:31 +0000 (13:38 -0700)]
Merge branch 'jc/apply-whitespace'

* jc/apply-whitespace:
  builtin-apply: do not declare patch is creation when we do not know it
  builtin-apply: accept patch to an empty file
  builtin-apply: typofix

16 years agoMerge branch 'jc/unpack-trees-reword'
Junio C Hamano [Sun, 25 May 2008 20:38:27 +0000 (13:38 -0700)]
Merge branch 'jc/unpack-trees-reword'

* jc/unpack-trees-reword:
  unpack-trees: allow Porcelain to give different error messages

16 years agoMerge branch 'ar/batch-cat'
Junio C Hamano [Sun, 25 May 2008 20:38:06 +0000 (13:38 -0700)]
Merge branch 'ar/batch-cat'

* ar/batch-cat:
  change quoting in test t1006-cat-file.sh
  builtin-cat-file.c: use parse_options()
  git-svn: Speed up fetch
  Git.pm: Add hash_and_insert_object and cat_blob
  Git.pm: Add command_bidi_pipe and command_close_bidi_pipe
  git-hash-object: Add --stdin-paths option
  Add more tests for git hash-object
  Move git-hash-object tests from t5303 to t1007
  git-cat-file: Add --batch option
  git-cat-file: Add --batch-check option
  git-cat-file: Make option parsing a little more flexible
  git-cat-file: Small refactor of cmd_cat_file
  Add tests for git cat-file

16 years agoMerge branch 'cc/bisect'
Junio C Hamano [Sun, 25 May 2008 20:38:03 +0000 (13:38 -0700)]
Merge branch 'cc/bisect'

* cc/bisect:
  bisect: use a detached HEAD to bisect
  bisect: trap critical errors in "bisect_start"
  bisect: fix left over "BISECT_START" file when starting with junk rev
  bisect: add test cases to check that "git bisect start" is atomic

16 years agoMerge branch 'ap/svn'
Junio C Hamano [Sun, 25 May 2008 20:37:25 +0000 (13:37 -0700)]
Merge branch 'ap/svn'

* ap/svn:
  git-svn: add test for --add-author-from and --use-log-author
  git-svn: add documentation for --add-author-from option.
  git-svn: Add --add-author-from option.
  git-svn: add documentation for --use-log-author option.

16 years agoMerge branch 'js/cvsexportcommit'
Junio C Hamano [Sun, 25 May 2008 20:37:20 +0000 (13:37 -0700)]
Merge branch 'js/cvsexportcommit'

* js/cvsexportcommit:
  cvsexportcommit: introduce -W for shared working trees (between Git and CVS)
  cvsexportcommit: chomp only removes trailing whitespace

Conflicts:

git-cvsexportcommit.perl

16 years agoMerge branch 'js/ignore-submodule'
Junio C Hamano [Sun, 25 May 2008 20:37:08 +0000 (13:37 -0700)]
Merge branch 'js/ignore-submodule'

* js/ignore-submodule:
  Ignore dirty submodule states during rebase and stash
  Teach update-index about --ignore-submodules
  diff options: Introduce --ignore-submodules

16 years agoMerge branch 'mo/cvsserver'
Junio C Hamano [Sun, 25 May 2008 20:37:04 +0000 (13:37 -0700)]
Merge branch 'mo/cvsserver'

* mo/cvsserver:
  Documentation: Fix skipped section level
  git-cvsserver: add ability to guess -kb from contents
  implement gitcvs.usecrlfattr
  git-cvsserver: add mechanism for managing working tree and current directory

16 years agomailsplit: minor clean-up in read_line_with_nul()
Junio C Hamano [Sun, 25 May 2008 08:23:55 +0000 (01:23 -0700)]
mailsplit: minor clean-up in read_line_with_nul()

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agomailinfo: apply the same fix not to lose NULs in BASE64 and QP codepaths
Junio C Hamano [Sun, 25 May 2008 08:16:05 +0000 (01:16 -0700)]
mailinfo: apply the same fix not to lose NULs in BASE64 and QP codepaths

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agomailsplit and mailinfo: gracefully handle NUL characters
Johannes Schindelin [Fri, 16 May 2008 13:03:30 +0000 (14:03 +0100)]
mailsplit and mailinfo: gracefully handle NUL characters

The function fgets() has a big problem with NUL characters: it reads
them, but nobody will know if the NUL comes from the file stream, or
was appended at the end of the line.

So implement a custom read_line_with_nul() function.

Noticed by Tommy Thorn.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agohttp-push: remove remote locks on exit signals
Clemens Buchacher [Sun, 25 May 2008 18:27:44 +0000 (20:27 +0200)]
http-push: remove remote locks on exit signals

If locks are not cleaned up the repository is inaccessible for 10 minutes.

Signed-off-by: Clemens Buchacher <drizzd@aon.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoReset the signal being handled
Clemens Buchacher [Sun, 25 May 2008 18:26:50 +0000 (20:26 +0200)]
Reset the signal being handled

This did not cause any problems, because remove_lock_file_on_signal is
only registered for SIGINT.

Signed-off-by: Clemens Buchacher <drizzd@aon.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMake git-cvsimport remove ['s from tags, as bad_ref_char doesn't allow them.
Paul Oliver [Fri, 23 May 2008 18:29:22 +0000 (19:29 +0100)]
Make git-cvsimport remove ['s from tags, as bad_ref_char doesn't allow them.

Signed-off-by: Paul Oliver <puzza007@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoget_revision(): honor the topo_order flag for boundary commits
Adam Simpkins [Sat, 24 May 2008 23:02:05 +0000 (16:02 -0700)]
get_revision(): honor the topo_order flag for boundary commits

Now get_revision() sorts the boundary commits when topo_order is set.
Since sort_in_topological_order() takes a struct commit_list, it first
places the boundary commits into revs->commits.

Signed-off-by: Adam Simpkins <adam@adamsimpkins.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoFix output of "git log --graph --boundary"
Adam Simpkins [Sat, 24 May 2008 23:02:04 +0000 (16:02 -0700)]
Fix output of "git log --graph --boundary"

Previously the graphing API wasn't aware of the revs->boundary flag, and
it always assumed that commits marked UNINTERESTING would not be
displayed.  As a result, the boundary commits were printed at the end of
the log output, but they didn't have any branch lines connecting them to
their children in the graph.

There was also another bug in the get_revision() code that caused
graph_update() to be called twice on the first boundary commit.  This
caused the graph API to think that a commit had been skipped, and print
a "..." line in the output.

Signed-off-by: Adam Simpkins <adam@adamsimpkins.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agolog --graph --left-right: show left/right information in place of '*'
Adam Simpkins [Sun, 25 May 2008 07:07:21 +0000 (00:07 -0700)]
log --graph --left-right: show left/right information in place of '*'

With the --graph option, the graph already outputs 'o' instead of '*'
for boundary commits.  Make it emit '<' or '>' when --left-right is
specified.

(This change also disables the '^' prefix for UNINTERESTING commits.
The graph code currently doesn't print anything special for these
commits, since it assumes no UNINTERESTING, non-BOUNDARY commits are
displayed.  This is potentially a bug if UNINTERESTING non-BOUNDARY
commits can actually be displayed via some code path.)

[jc: squashed the left-right change from Dscho and Adam's fixup into one]

Signed-off-by: Adam Simpkins <adam@adamsimpkins.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agograph API: don't print branch lines for uninteresting merge parents
Adam Simpkins [Sat, 24 May 2008 02:24:11 +0000 (19:24 -0700)]
graph API: don't print branch lines for uninteresting merge parents

Previously, the graphing code printed lines coming out of a merge commit
for all of its parents, even if some of them were uninteresting.  Now it
only prints lines for interesting commits.

For example, for a merge commit where only the first parent is
interesting, the code now prints:

  *  merge commit
  *  interesting child

instead of:

  M  merge commit
  |\
  *  interesting child

Signed-off-by: Adam Simpkins <adam@adamsimpkins.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agograph API: fix graph mis-alignment after uninteresting commits
Adam Simpkins [Sat, 24 May 2008 02:24:10 +0000 (19:24 -0700)]
graph API: fix graph mis-alignment after uninteresting commits

The graphing code had a bug that caused it to output branch lines
incorrectly after ignoring an uninteresting commit.  When computing how
to match up the branch lines from the current commit to the next one, it
forgot to take into account that it needed to initially start with 2
empty spaces where the missing commit would have gone.

So, instead of drawing this,

  | * | <- Commit with uninteresting parent
  |  /
  * |

It used to incorrectly draw this:

  | * | <- Commit with uninteresting parent
  * |

Signed-off-by: Adam Simpkins <adam@adamsimpkins.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoDocumentation: convert tutorials to man pages
Christian Couder [Sat, 24 May 2008 18:56:44 +0000 (20:56 +0200)]
Documentation: convert tutorials to man pages

This patch renames the following documents and at the same time converts
them to the man page format:

cvs-migration.txt -> gitcvs-migration.txt
tutorial.txt      -> gittutorial.txt
tutorial-2.txt    -> gittutorial-2.txt

These new man pages are put in section 7, and other documents that reference
the above ones are change accordingly.

[jc: with help from Nanako to clean things up]

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoCodingGuidelines: Add a note to avoid assignments inside if()
Miklos Vajna [Thu, 22 May 2008 23:26:09 +0000 (01:26 +0200)]
CodingGuidelines: Add a note to avoid assignments inside if()

Signed-off-by: Miklos Vajna <vmiklos@frugalware.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoRemove redundant code, eliminate one static variable
Heikki Orsila [Fri, 23 May 2008 22:43:55 +0000 (01:43 +0300)]
Remove redundant code, eliminate one static variable

Signed-off-by: Heikki Orsila <heikki.orsila@iki.fi>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years ago"git diff": do not ignore index without --no-index
Junio C Hamano [Sat, 24 May 2008 05:28:56 +0000 (22:28 -0700)]
"git diff": do not ignore index without --no-index

Even if "foo" and/or "bar" does not exist in index, "git diff foo bar"
should not change behaviour drastically from "git diff foo bar baz" or
"git diff foo".  A feature that "sometimes works and is handy" is an
unreliable cute hack.

"git diff foo bar" outside a git repository continues to work as a more
colourful alternative to "diff -u" as before.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agodiff-files: do not play --no-index games
Junio C Hamano [Sat, 24 May 2008 01:15:03 +0000 (18:15 -0700)]
diff-files: do not play --no-index games

Being able to say "git diff A B" outside a git repository and getting a
colourful version of "diff -u A B" may be nice, but such a cute hack
should not give bogus results to scripts that want to give two paths,
either or both of which happen to have been removed from the work tree,
to "git diff-files".

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agotests: do not use implicit "git diff --no-index"
Junio C Hamano [Sat, 24 May 2008 05:28:56 +0000 (22:28 -0700)]
tests: do not use implicit "git diff --no-index"

As a general principle, we should not use "git diff" to validate the
results of what git command that is being tested has done.  We would not
know if we are testing the command in question, or locating a bug in the
cute hack of "git diff --no-index".

Rather use test_cmp for that purpose.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge git://git.kernel.org/pub/scm/gitk/gitk
Junio C Hamano [Sat, 24 May 2008 01:28:52 +0000 (18:28 -0700)]
Merge git://git.kernel.org/pub/scm/gitk/gitk

* git://git.kernel.org/pub/scm/gitk/gitk:
  gitk: Fix bug introduced by "gitk: Fix "wrong # coordinates" error on reload"
  gitk: Fix bug where current row number display stops working
  gitk: Move es.po where it belongs
  gitk: Fix "wrong # coordinates" error on reload

16 years agoMerge branch 'pb/push'
Junio C Hamano [Fri, 23 May 2008 23:06:07 +0000 (16:06 -0700)]
Merge branch 'pb/push'

* pb/push:
  add special "matching refs" refspec

16 years agoMerge branch 'bc/repack'
Junio C Hamano [Fri, 23 May 2008 23:06:01 +0000 (16:06 -0700)]
Merge branch 'bc/repack'

* bc/repack:
  Documentation/git-repack.txt: document new -A behaviour
  let pack-objects do the writing of unreachable objects as loose objects
  add a force_object_loose() function
  builtin-gc.c: deprecate --prune, it now really has no effect
  git-gc: always use -A when manually repacking
  repack: modify behavior of -A option to leave unreferenced objects unpacked

Conflicts:

builtin-pack-objects.c

16 years agoMerge branch 'sp/ignorecase'
Junio C Hamano [Fri, 23 May 2008 23:05:52 +0000 (16:05 -0700)]
Merge branch 'sp/ignorecase'

* sp/ignorecase:
  t0050: Fix merge test on case sensitive file systems
  t0050: Add test for case insensitive add
  t0050: Set core.ignorecase case to activate case insensitivity
  t0050: Test autodetect core.ignorecase
  git-init: autodetect core.ignorecase

16 years agoMerge branch 'maint'
Junio C Hamano [Fri, 23 May 2008 23:05:46 +0000 (16:05 -0700)]
Merge branch 'maint'

* maint:
  rev-parse --symbolic-full-name: don't print '^' if SHA1 is not a ref
  Add missing "short" alternative to --date in rev-list-options.txt
  git-show.txt: Not very stubby these days.
  Clarify repack -n documentation

16 years agoAdd log.date config variable
Heikki Orsila [Thu, 22 May 2008 15:24:07 +0000 (18:24 +0300)]
Add log.date config variable

log.date config variable sets the default date-time mode for the log
command. Setting log.date value is similar to using git log's --date
option.

Signed-off-by: Heikki Orsila <heikki.orsila@iki.fi>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMerge branch 'maint-1.5.4' into maint v1.5.5.2
Junio C Hamano [Fri, 23 May 2008 22:52:43 +0000 (15:52 -0700)]
Merge branch 'maint-1.5.4' into maint

* maint-1.5.4:
  rev-parse --symbolic-full-name: don't print '^' if SHA1 is not a ref

16 years agorev-parse --symbolic-full-name: don't print '^' if SHA1 is not a ref
Johannes Sixt [Fri, 23 May 2008 14:13:05 +0000 (16:13 +0200)]
rev-parse --symbolic-full-name: don't print '^' if SHA1 is not a ref

The intention of --symbolic-full-name is to not print anything if a
revision is not an exact ref. But this command:

    $ git-rev-parse --symbolic-full-name --not master~1

still emitted a sole '^' to stdout (provided that there's no other ref at
master~1). This fixes it.

Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMake git add -n and git -u -n output consistent
Gustaf Hendeby [Thu, 22 May 2008 21:59:42 +0000 (23:59 +0200)]
Make git add -n and git -u -n output consistent

Output format from "git add -n $path" lists path to blobs that are going
to be added on a single line, separated with SP.  On the other hand, the
suggested "git add -u -n" shows one path per line, like "add '<file>'\n".
Of course, these two are inconsistent.

Plain "git add -n" can afford to only say names of paths, as all it does
is to add (update).  However, "git add -u" needs to be able to express
"remove" somehow.  So if we need to have them formatted the same way, we
need to unify with the "git add -n -u" format.  Incidentally, this is
consistent with how 'update-index' says it.

This changes the output from "git add -n $paths" but as a general
principle, output from Porcelain commands is a fair game for improvements
and not for script consumption.

Signed-off-by: Gustaf Hendeby <hendeby@isy.liu.se>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agochange quoting in test t1006-cat-file.sh
Michele Ballabio [Fri, 23 May 2008 14:19:43 +0000 (16:19 +0200)]
change quoting in test t1006-cat-file.sh

Signed-off-by: Michele Ballabio <barra_cuda@katamail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobuiltin-cat-file.c: use parse_options()
Michele Ballabio [Fri, 23 May 2008 14:19:42 +0000 (16:19 +0200)]
builtin-cat-file.c: use parse_options()

This simplifies the option parsing.

Signed-off-by: Michele Ballabio <barra_cuda@katamail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-svn: Speed up fetch
Adam Roben [Fri, 23 May 2008 14:19:41 +0000 (16:19 +0200)]
git-svn: Speed up fetch

We were spending a lot of time forking/execing git-cat-file and
git-hash-object. We now maintain a global Git repository object in order to use
Git.pm's more efficient hash_and_insert_object and cat_blob methods.

Signed-off-by: Adam Roben <aroben@apple.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoGit.pm: Add hash_and_insert_object and cat_blob
Adam Roben [Fri, 23 May 2008 14:19:40 +0000 (16:19 +0200)]
Git.pm: Add hash_and_insert_object and cat_blob

These functions are more efficient ways of executing `git hash-object -w` and
`git cat-file blob` when you are dealing with many files/objects.

Signed-off-by: Adam Roben <aroben@apple.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoGit.pm: Add command_bidi_pipe and command_close_bidi_pipe
Adam Roben [Fri, 23 May 2008 14:19:39 +0000 (16:19 +0200)]
Git.pm: Add command_bidi_pipe and command_close_bidi_pipe

command_bidi_pipe hands back the stdin and stdout file handles from the
executed command. command_close_bidi_pipe closes these handles and terminates
the process.

Signed-off-by: Adam Roben <aroben@apple.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-hash-object: Add --stdin-paths option
Adam Roben [Fri, 23 May 2008 14:19:38 +0000 (16:19 +0200)]
git-hash-object: Add --stdin-paths option

This allows multiple paths to be specified on stdin.

Signed-off-by: Adam Roben <aroben@apple.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoAdd more tests for git hash-object
Adam Roben [Fri, 23 May 2008 14:19:37 +0000 (16:19 +0200)]
Add more tests for git hash-object

Signed-off-by: Adam Roben <aroben@apple.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoMove git-hash-object tests from t5303 to t1007
Adam Roben [Fri, 23 May 2008 14:19:36 +0000 (16:19 +0200)]
Move git-hash-object tests from t5303 to t1007

This is a more appropriate location according to t/README.

Signed-off-by: Adam Roben <aroben@apple.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoUpdated status to show 'Not currently on any branch' in red
Chris Parsons [Thu, 22 May 2008 12:50:02 +0000 (08:50 -0400)]
Updated status to show 'Not currently on any branch' in red

This provides additional warning to users when attempting to
commit to a detached HEAD. It is configurable in color.status.nobranch.

Signed-off-by: Chris Parsons <chris@edendevelopment.co.uk>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoAdd test for cloning with "--reference" repo being a subset of source repo
Johan Herland [Thu, 22 May 2008 22:03:00 +0000 (18:03 -0400)]
Add test for cloning with "--reference" repo being a subset of source repo

The first test in this series tests "git clone -l -s --reference B A C",
where repo B is a superset of repo A (A has one commit, B has the same
commit plus another). In this case, all objects to be cloned are already
present in B.

However, we should also test the case where the "--reference" repo is a
_subset_ of the source repo (e.g. "git clone -l -s --reference A B C"),
i.e. some objects are not available in the "--reference" repo, and will
have to be found in the source repo.

Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoAdd a test for another combination of --reference
Johan Herland [Thu, 22 May 2008 22:03:05 +0000 (18:03 -0400)]
Add a test for another combination of --reference

In this case, the reference repository has some useful loose objects,
but not all useful objects, and we make sure that we can find the
objects we fetch from the repository we're cloning in the new
repository, instead of potentially being distracted by the reference
repository.

Doing the wrong thing in a builtin-clone implementation would lead to
this looking for an object in the wrong place, not finding it (because
it's only in the right place), and crashing.

Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoTest that --reference actually suppresses fetching referenced objects
Daniel Barkalow [Thu, 22 May 2008 22:03:08 +0000 (18:03 -0400)]
Test that --reference actually suppresses fetching referenced objects

Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoAdd missing "short" alternative to --date in rev-list-options.txt
Heikki Orsila [Thu, 22 May 2008 15:24:41 +0000 (18:24 +0300)]
Add missing "short" alternative to --date in rev-list-options.txt

Signed-off-by: Heikki Orsila <heikki.orsila@iki.fi>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-show.txt: Not very stubby these days.
Jon Loeliger [Thu, 22 May 2008 19:59:25 +0000 (14:59 -0500)]
git-show.txt: Not very stubby these days.

Signed-off-by: Jon Loeliger <jdl@freescale.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agoClarify repack -n documentation
Shawn O. Pearce [Thu, 22 May 2008 12:47:19 +0000 (08:47 -0400)]
Clarify repack -n documentation

While repacking a local repository a coworker thought the -n option
was necessary to git-repack to keep it from updating some unknown
file on the central server we all share.  Explaining further what
the option is (not) doing helps to make it clear the option does
not impact any remote repositories the user may have configured.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agopull --rebase: exit early when the working directory is dirty
Johannes Schindelin [Wed, 21 May 2008 11:32:16 +0000 (12:32 +0100)]
pull --rebase: exit early when the working directory is dirty

When rebasing fails during "pull --rebase", you cannot just clean up the
working directory and call "pull --rebase" again, since the remote branch
was already fetched.

Therefore, die early when the working directory is dirty.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogitweb: Convert string to internal form before chopping in chop_str
Anders Waldenborg [Wed, 21 May 2008 11:44:43 +0000 (13:44 +0200)]
gitweb: Convert string to internal form before chopping in chop_str

Fix chop_str not to cut in middle of utf8 multibyte chars. Without
this fix at least author name in short log may cut in middle of a
multibyte char. When the result comes to esc_html to_utf8 is called
again, which doesn't find valid utf8 and decodes using
$fallback_encoding making it even worse.

This also have the nice side effect that it actually tries to show the
first 10 _characters_, not the number of characters that happened to fit
into 10 bytes.

Signed-off-by: Anders Waldenborg <anders@0x63.nu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobisect: use a detached HEAD to bisect
Christian Couder [Thu, 22 May 2008 23:28:57 +0000 (01:28 +0200)]
bisect: use a detached HEAD to bisect

When "git bisect" was first written, it was not possible to
checkout a detached HEAD. The detached feature appeared latter.

That's why before this patch the "git bisect" process used a
"bisect" branch to checkout new revisions to be tested (and also
a "new-bisect" one to check if the checkouts could work).

This patch makes "git bisect" checkout revisions to be tested on
a detached HEAD. This simplifies the code a bit.

The tests to check that "git bisect" does not start if a
"bisect" or a "new-bisect" branch exists are removed as they
are not relevant any more.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobisect: trap critical errors in "bisect_start"
Christian Couder [Thu, 22 May 2008 22:39:22 +0000 (00:39 +0200)]
bisect: trap critical errors in "bisect_start"

Before this patch, when using "git bisect start" with mistaken revs
or when the checkout of the branch we want to test failed, we exited
after having written files like ".git/BISECT_START",
".git/BISECT_NAMES" and after having written "refs/bisect/bad" and
"refs/bisect/good-*" refs.

With this patch we trap all errors that can happen when writing the
new state and when we are in "bisect_next". So that we can try to
clean up everything in case of problems, using "bisect_clean_state".

This patch also contains a "bisect_write" cleanup to make it exit
on error and return 0 otherwise.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobisect: fix left over "BISECT_START" file when starting with junk rev
Christian Couder [Thu, 22 May 2008 22:38:59 +0000 (00:38 +0200)]
bisect: fix left over "BISECT_START" file when starting with junk rev

Before this patch, when using for example:

$ git bisect start <stuff1> <stuff2>

with <stuff1> or <stuff2> that cannot be parsed as a revision, we
could leave a ".git/BISECT_START" file, from a previous
"git bisect start", alone.

This patch makes sure that it does not happen by removing the
"BISECT_START" file in "bisect_clean_state" and then always writing
it again at the end of "bisect_start".

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agobisect: add test cases to check that "git bisect start" is atomic
Christian Couder [Thu, 22 May 2008 22:38:46 +0000 (00:38 +0200)]
bisect: add test cases to check that "git bisect start" is atomic

This patch adds some test cases to check that "git bisect start"
doesn't leave us in a bad state, especially when it fails.

These test cases show that "git bisect start" is not atomic when it
fails and leave some files like .git/BISECT_START, and in some
cases some refs, over.

The test failures should be fixed in latter commits.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
16 years agogit-gui: Handle workdir detection when CYGWIN=nowinsymlinks
Shawn O. Pearce [Wed, 21 May 2008 20:40:10 +0000 (16:40 -0400)]
git-gui: Handle workdir detection when CYGWIN=nowinsymlinks

If the user has put nowinsymlinks into their CYGWIN environment
variable any symlinks created by a Cygwin process (e.g. ln -s)
will not have the ".lnk" suffix.  In this case workdir is still
a workdir, but our detection of looking for "info.lnk" fails
as the symlink is actually a normal file called "info".

Instead we just always use Cygwin's test executable to see if
info/exclude is a file.  If it is, we assume from there on it
can be read by git-ls-files --others and is thus safe to use
on the command line.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
16 years agogitk: Fix bug introduced by "gitk: Fix "wrong # coordinates" error on reload"
Paul Mackerras [Thu, 22 May 2008 11:12:19 +0000 (21:12 +1000)]
gitk: Fix bug introduced by "gitk: Fix "wrong # coordinates" error on reload"

Commit 94503a66c56c935e77a8fbe3622f1f56b7134ccc ("gitk: Fix "wrong #
coordinates" error on reload") was correct as far as it went, but
introduced a problem because it didn't also clear out boldrows and
boldnamerows in clear_display.  This resulted in Tcl errors after
scrolling through the graph for a while if some rows were highlighted.
This fixes it.

Signed-off-by: Paul Mackerras <paulus@samba.org>