git
10 years agoMerge branch 'nd/log-show-linear-break'
Junio C Hamano [Thu, 3 Apr 2014 19:38:11 +0000 (12:38 -0700)] 
Merge branch 'nd/log-show-linear-break'

Attempts to show where a single-strand-of-pearls break in "git log"
output.

* nd/log-show-linear-break:
  log: add --show-linear-break to help see non-linear history
  object.h: centralize object flag allocation

10 years agoMerge branch 'ep/shell-command-substitution'
Junio C Hamano [Thu, 3 Apr 2014 19:38:04 +0000 (12:38 -0700)] 
Merge branch 'ep/shell-command-substitution'

* ep/shell-command-substitution:
  git-am.sh: use the $(...) construct for command substitution
  check-builtins.sh: use the $(...) construct for command substitution

10 years agoMerge branch 'ap/remote-hg-skip-null-bookmarks'
Junio C Hamano [Wed, 2 Apr 2014 21:18:23 +0000 (14:18 -0700)] 
Merge branch 'ap/remote-hg-skip-null-bookmarks'

* ap/remote-hg-skip-null-bookmarks:
  remote-hg: do not fail on invalid bookmarks

10 years agoMerge branch 'rs/pickaxe-i'
Junio C Hamano [Wed, 2 Apr 2014 21:18:19 +0000 (14:18 -0700)] 
Merge branch 'rs/pickaxe-i'

Allow the options -i/--regexp-ignore-case, --pickaxe-regex, and -S
to be used together and work as expected to perform a pickaxe
search using case-insensitive regular expression matching.

* rs/pickaxe-i:
  pickaxe: simplify kwset loop in contains()
  pickaxe: call strlen only when necessary in diffcore_pickaxe_count()
  pickaxe: move pickaxe() after pickaxe_match()
  pickaxe: merge diffcore_pickaxe_grep() and diffcore_pickaxe_count() into diffcore_pickaxe()
  pickaxe: honor -i when used with -S and --pickaxe-regex
  t4209: use helper functions to test --author
  t4209: use helper functions to test --grep
  t4209: factor out helper function test_log_icase()
  t4209: factor out helper function test_log()
  t4209: set up expectations up front

10 years agoRevert part of 384364b (Start preparing for Git 2.0, 2014-03-07)
Junio C Hamano [Tue, 1 Apr 2014 18:55:56 +0000 (11:55 -0700)] 
Revert part of 384364b (Start preparing for Git 2.0, 2014-03-07)

As we are not shipping with the submodule change, remove the
entry for it.

10 years agoRevert "submodule: explicit local branch creation in module_clone"
Junio C Hamano [Wed, 2 Apr 2014 21:15:36 +0000 (14:15 -0700)] 
Revert "submodule: explicit local branch creation in module_clone"

This reverts commit 23d25e48f5ead73c9ce233986f90791abec9f1e8, as it is
broken for users who haven't opted into the new feature of checking
out submodule.*.branch with update mode set to checkout.

10 years agoRevert "Merge branch 'wt/doc-submodule-name-path-confusion-2'"
Junio C Hamano [Tue, 1 Apr 2014 18:52:37 +0000 (11:52 -0700)] 
Revert "Merge branch 'wt/doc-submodule-name-path-confusion-2'"

This reverts commit 00d4ff1a69883e24b095f45251d99143b5bc0320, reversing
changes made to d3badc6eb0961382788c2670129d5ee133d079fd.

10 years agoUpdate draft release notes to 2.0
Junio C Hamano [Mon, 31 Mar 2014 23:41:23 +0000 (16:41 -0700)] 
Update draft release notes to 2.0

Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoMerge branch 'mm/status-porcelain-format-i18n-fix'
Junio C Hamano [Mon, 31 Mar 2014 23:31:25 +0000 (16:31 -0700)] 
Merge branch 'mm/status-porcelain-format-i18n-fix'

* mm/status-porcelain-format-i18n-fix:
  status: disable translation when --porcelain is used

10 years agoMerge branch 'an/branch-config-message'
Junio C Hamano [Mon, 31 Mar 2014 23:31:19 +0000 (16:31 -0700)] 
Merge branch 'an/branch-config-message'

* an/branch-config-message:
  branch.c: install_branch_config: simplify if chain

10 years agoMerge branch 'jk/tests-cleanup'
Junio C Hamano [Mon, 31 Mar 2014 23:31:17 +0000 (16:31 -0700)] 
Merge branch 'jk/tests-cleanup'

* jk/tests-cleanup:
  t0001: drop subshells just for "cd"
  t0001: drop useless subshells
  t0001: use test_must_fail
  t0001: use test_config_global
  t0001: use test_path_is_*
  t0001: make symlink reinit test more careful
  t: prefer "git config --file" to GIT_CONFIG
  t: prefer "git config --file" to GIT_CONFIG with test_must_fail
  t: stop using GIT_CONFIG to cross repo boundaries
  t: drop useless sane_unset GIT_* calls
  t/test-lib: drop redundant unset of GIT_CONFIG
  t/Makefile: stop setting GIT_CONFIG

10 years agoMerge branch 'wt/doc-submodule-name-path-confusion-2'
Junio C Hamano [Mon, 31 Mar 2014 23:31:16 +0000 (16:31 -0700)] 
Merge branch 'wt/doc-submodule-name-path-confusion-2'

* wt/doc-submodule-name-path-confusion-2:
  doc: submodule.*.branch config is keyed by name

10 years agoMerge branch 'wt/doc-submodule-name-path-confusion-1'
Junio C Hamano [Mon, 31 Mar 2014 23:31:13 +0000 (16:31 -0700)] 
Merge branch 'wt/doc-submodule-name-path-confusion-1'

* wt/doc-submodule-name-path-confusion-1:
  doc: submodule.* config are keyed by submodule names

10 years agoMerge branch 'mr/msvc-link-with-lcurl'
Junio C Hamano [Mon, 31 Mar 2014 23:31:07 +0000 (16:31 -0700)] 
Merge branch 'mr/msvc-link-with-lcurl'

* mr/msvc-link-with-lcurl:
  MSVC: allow linking with the cURL library

10 years agoMerge branch 'ib/rev-parse-parseopt-argh'
Junio C Hamano [Mon, 31 Mar 2014 23:30:59 +0000 (16:30 -0700)] 
Merge branch 'ib/rev-parse-parseopt-argh'

Teaches the "rev-parse --parseopt" mechanism used by scripted
Porcelains to parse command line options and give help text how to
supply argv-help (the placeholder string for an option parameter,
e.g. "key-id" in "--gpg-sign=<key-id>").

* ib/rev-parse-parseopt-argh:
  t1502: protect runs of SPs used in the indentation
  rev-parse --parseopt: option argument name hints

10 years agoMerge branch 'dp/makefile-charset-lib-doc'
Junio C Hamano [Mon, 31 Mar 2014 23:30:57 +0000 (16:30 -0700)] 
Merge branch 'dp/makefile-charset-lib-doc'

* dp/makefile-charset-lib-doc:
  Makefile: describe CHARSET_LIB better

10 years agoMerge branch 'js/userdiff-cc'
Junio C Hamano [Mon, 31 Mar 2014 23:30:54 +0000 (16:30 -0700)] 
Merge branch 'js/userdiff-cc'

Improves the pattern to match the hunk-header for C/C++.

* js/userdiff-cc:
  userdiff: have 'cpp' hunk header pattern catch more C++ anchor points
  t4018: test cases showing that the cpp pattern misses many anchor points
  t4018: test cases for the built-in cpp pattern
  t4018: reduce test files for pattern compilation tests
  t4018: convert custom pattern test to the new infrastructure
  t4018: convert java pattern test to the new infrastructure
  t4018: convert perl pattern tests to the new infrastructure
  t4018: an infrastructure to test hunk headers
  userdiff: support unsigned and long long suffixes of integer constants
  userdiff: support C++ ->* and .* operators in the word regexp

10 years agoMerge branch 'dw/doc-status-no-longer-shows-pound-prefix'
Junio C Hamano [Mon, 31 Mar 2014 23:30:52 +0000 (16:30 -0700)] 
Merge branch 'dw/doc-status-no-longer-shows-pound-prefix'

* dw/doc-status-no-longer-shows-pound-prefix:
  doc: status, remove leftover statement about '#' prefix

10 years agoMerge branch 'ca/doc-config-third-party'
Junio C Hamano [Mon, 31 Mar 2014 23:30:49 +0000 (16:30 -0700)] 
Merge branch 'ca/doc-config-third-party'

* ca/doc-config-third-party:
  config.txt: third-party tools may and do use their own variables

10 years agoMerge branch 'hs/simplify-bit-setting-in-fsck-tree'
Junio C Hamano [Mon, 31 Mar 2014 23:30:44 +0000 (16:30 -0700)] 
Merge branch 'hs/simplify-bit-setting-in-fsck-tree'

* hs/simplify-bit-setting-in-fsck-tree:
  fsck: use bitwise-or assignment operator to set flag

10 years agoMerge branch 'dt/tests-with-env-not-subshell'
Junio C Hamano [Mon, 31 Mar 2014 23:30:40 +0000 (16:30 -0700)] 
Merge branch 'dt/tests-with-env-not-subshell'

* dt/tests-with-env-not-subshell:
  tests: use "env" to run commands with temporary env-var settings

10 years agoUpdate draft release notes to 2.0
Junio C Hamano [Fri, 28 Mar 2014 20:56:29 +0000 (13:56 -0700)] 
Update draft release notes to 2.0

Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoMerge branch 'ys/fsck-commit-parsing'
Junio C Hamano [Fri, 28 Mar 2014 20:51:24 +0000 (13:51 -0700)] 
Merge branch 'ys/fsck-commit-parsing'

* ys/fsck-commit-parsing:
  fsck.c:fsck_commit(): use skip_prefix() to verify and skip constant
  fsck.c:fsck_ident(): ident points at a const string

10 years agoMerge branch 'bg/rebase-off-of-previous-branch'
Junio C Hamano [Fri, 28 Mar 2014 20:51:20 +0000 (13:51 -0700)] 
Merge branch 'bg/rebase-off-of-previous-branch'

* bg/rebase-off-of-previous-branch:
  rebase: allow "-" short-hand for the previous branch

10 years agoMerge branch 'bp/commit-p-editor'
Junio C Hamano [Fri, 28 Mar 2014 20:51:11 +0000 (13:51 -0700)] 
Merge branch 'bp/commit-p-editor'

When it is not necessary to edit a commit log message (e.g. "git
commit -m" is given a message without specifying "-e"), we used to
disable the spawning of the editor by overriding GIT_EDITOR, but
this means all the uses of the editor, other than to edit the
commit log message, are also affected.

* bp/commit-p-editor:
  run-command: mark run_hook_with_custom_index as deprecated
  merge hook tests: fix and update tests
  merge: fix GIT_EDITOR override for commit hook
  commit: fix patch hunk editing with "commit -p -m"
  test patch hunk editing with "commit -p -m"
  merge hook tests: use 'test_must_fail' instead of '!'
  merge hook tests: fix missing '&&' in test

10 years agoMerge branch 'ah/doc-gitk-config'
Junio C Hamano [Fri, 28 Mar 2014 20:51:08 +0000 (13:51 -0700)] 
Merge branch 'ah/doc-gitk-config'

* ah/doc-gitk-config:
  Documentation/gitk: document the location of the configulation file

10 years agoMerge branch 'fr/add-interactive-argv-array'
Junio C Hamano [Fri, 28 Mar 2014 20:51:04 +0000 (13:51 -0700)] 
Merge branch 'fr/add-interactive-argv-array'

* fr/add-interactive-argv-array:
  add: use struct argv_array in run_add_interactive()

10 years agoMerge branch 'jk/subtree-prefix'
Junio C Hamano [Fri, 28 Mar 2014 20:50:59 +0000 (13:50 -0700)] 
Merge branch 'jk/subtree-prefix'

A stray environment variable $prefix could have leaked into and
affected the behaviour of the "subtree" script.

* jk/subtree-prefix:
  subtree: initialize "prefix" variable

10 years agoMerge branch 'jk/pack-bitmap-progress'
Junio C Hamano [Fri, 28 Mar 2014 20:50:56 +0000 (13:50 -0700)] 
Merge branch 'jk/pack-bitmap-progress'

The progress output while repacking and transferring objects showed
an apparent large silence while writing the objects out of existing
packfiles, when the reachability bitmap was in use.

* jk/pack-bitmap-progress:
  pack-objects: show reused packfile objects in "Counting objects"
  pack-objects: show progress for reused packfiles

10 years agoMerge branch 'jk/pack-bitmap'
Junio C Hamano [Fri, 28 Mar 2014 20:50:50 +0000 (13:50 -0700)] 
Merge branch 'jk/pack-bitmap'

Instead of dying when asked to (re)pack with the reachability
bitmap when a bitmap cannot be built, just (re)pack without
producing a bitmap in such a case, with a warning.

* jk/pack-bitmap:
  pack-objects: turn off bitmaps when skipping objects

10 years agodoc: submodule.*.branch config is keyed by name
W. Trevor King [Thu, 27 Mar 2014 21:06:20 +0000 (14:06 -0700)] 
doc: submodule.*.branch config is keyed by name

Ever since 941987a5 (git-submodule: give submodules proper names,
2007-06-11) introduced the ability to move a submodule from one path
to another inside its superproject tree without losing its identity,
we should have consistently used submodule.<name>.* to access
settings related to the named submodule.

Signed-off-by: W. Trevor King <wking@tremily.us>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agodoc: submodule.* config are keyed by submodule names
W. Trevor King [Thu, 27 Mar 2014 21:06:20 +0000 (14:06 -0700)] 
doc: submodule.* config are keyed by submodule names

Ever since 941987a5 (git-submodule: give submodules proper names,
2007-06-11) introduced the ability to move a submodule from one path
to another inside its superproject tree without losing its identity,
we should have consistently used submodule.<name>.* to access
settings related to the named submodule.

Reported-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: W. Trevor King <wking@tremily.us>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoMSVC: allow linking with the cURL library
Marat Radchenko [Thu, 27 Mar 2014 07:34:27 +0000 (11:34 +0400)] 
MSVC: allow linking with the cURL library

Teach the clink.pl script that -lcurl is a request to link with the
cURL library, and drop NO_CURL from config.mak.uname for the MSVC
platform.

Signed-off-by: Marat Radchenko <marat@slonopotamus.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agostatus: disable translation when --porcelain is used
Matthieu Moy [Thu, 20 Mar 2014 12:12:41 +0000 (13:12 +0100)] 
status: disable translation when --porcelain is used

"git status --branch --porcelain" displays the status of the branch
(ahead, behind, gone), and used gettext to translate the string.

Use hardcoded strings when --porcelain is used, but keep the gettext
translation for "git status --short" which is essentially the same, but
meant to be read by a human.

Reported-by: Anarky <ghostanarky@gmail.com>
Signed-off-by: Matthieu Moy <Matthieu.Moy@imag.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agolog: add --show-linear-break to help see non-linear history
Nguyễn Thái Ngọc Duy [Tue, 25 Mar 2014 13:23:27 +0000 (20:23 +0700)] 
log: add --show-linear-break to help see non-linear history

Option explanation is in rev-list-options.txt. The interaction with -z
is left undecided.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoobject.h: centralize object flag allocation
Nguyễn Thái Ngọc Duy [Tue, 25 Mar 2014 13:23:26 +0000 (20:23 +0700)] 
object.h: centralize object flag allocation

While the field "flags" is mainly used by the revision walker, it is
also used in many other places. Centralize the whole flag allocation to
one place for a better overview (and easier to move flags if we have
too).

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agogit-am.sh: use the $(...) construct for command substitution
Elia Pinto [Tue, 25 Mar 2014 17:22:22 +0000 (10:22 -0700)] 
git-am.sh: use the $(...) construct for command substitution

The Git CodingGuidelines prefer the $(...) construct for command
substitution instead of using the backquotes `...`.

The backquoted form is the traditional method for command
substitution, and is supported by POSIX.  However, all but the
simplest uses become complicated quickly.  In particular, embedded
command substitutions and/or the use of double quotes require
careful escaping with the backslash character.

The patch was generated by:

    for _f in $(find . -name "*.sh")
    do
      sed -i 's@`\(.*\)`@$(\1)@g' ${_f}
    done

and then carefully proof-read.

Signed-off-by: Elia Pinto <gitter.spiros@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agocheck-builtins.sh: use the $(...) construct for command substitution
Elia Pinto [Tue, 25 Mar 2014 17:22:22 +0000 (10:22 -0700)] 
check-builtins.sh: use the $(...) construct for command substitution

The Git CodingGuidelines prefer the $(...) construct for command
substitution instead of using the backquotes `...`.

The backquoted form is the traditional method for command
substitution, and is supported by POSIX.  However, all but the
simplest uses become complicated quickly.  In particular, embedded
command substitutions and/or the use of double quotes require
careful escaping with the backslash character.

The patch was generated by:

    for _f in $(find . -name "*.sh")
    do
      sed -i 's@`\(.*\)`@$(\1)@g' ${_f}
    done

and then carefully proof-read.

Signed-off-by: Elia Pinto <gitter.spiros@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoremote-hg: do not fail on invalid bookmarks
Max Horn [Fri, 21 Mar 2014 11:36:36 +0000 (12:36 +0100)] 
remote-hg: do not fail on invalid bookmarks

Mercurial can have bookmarks pointing to "nullid" (the empty root
revision), while Git can not have references to it. When cloning or
fetching from a Mercurial repository that has such a bookmark, the
import failed because git-remote-hg was not be able to create the
corresponding reference.

Warn the user about the invalid reference, and do not advertise these
bookmarks as head refs, but otherwise continue the import. In
particular, we still keep track of the fact that the remote repository
has a bookmark of the given name, in case the user wants to modify that
bookmark.

Also add some test cases for this issue.

Reported-by: Antoine Pelisse <apelisse@gmail.com>
Signed-off-by: Max Horn <max@quendi.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoUpdate draft release notes to 2.0
Junio C Hamano [Tue, 25 Mar 2014 19:01:39 +0000 (12:01 -0700)] 
Update draft release notes to 2.0

Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoMerge branch 'ss/test-on-mingw-rsync-path-no-absolute'
Junio C Hamano [Tue, 25 Mar 2014 18:08:35 +0000 (11:08 -0700)] 
Merge branch 'ss/test-on-mingw-rsync-path-no-absolute'

* ss/test-on-mingw-rsync-path-no-absolute:
  t5510: Do not use $(pwd) when fetching / pushing / pulling via rsync

10 years agoMerge branch 'bb/diff-no-index-dotdot'
Junio C Hamano [Tue, 25 Mar 2014 18:08:31 +0000 (11:08 -0700)] 
Merge branch 'bb/diff-no-index-dotdot'

* bb/diff-no-index-dotdot:
  diff-no-index: replace manual "."/".." check with is_dot_or_dotdot()
  diff-no-index: rename read_directory()

10 years agoMerge branch 'us/printf-not-echo'
Junio C Hamano [Tue, 25 Mar 2014 18:08:26 +0000 (11:08 -0700)] 
Merge branch 'us/printf-not-echo'

* us/printf-not-echo:
  test-lib.sh: do not "echo" caller-supplied strings
  rebase -i: do not "echo" random user-supplied strings

10 years agoMerge branch 'rr/doc-merge-strategies'
Junio C Hamano [Tue, 25 Mar 2014 18:08:23 +0000 (11:08 -0700)] 
Merge branch 'rr/doc-merge-strategies'

* rr/doc-merge-strategies:
  Documentation/merge-strategies: avoid hyphenated commands

10 years agoMerge branch 'nd/index-pack-error-message'
Junio C Hamano [Tue, 25 Mar 2014 18:08:19 +0000 (11:08 -0700)] 
Merge branch 'nd/index-pack-error-message'

* nd/index-pack-error-message:
  index-pack: report error using the correct variable

10 years agoMerge branch 'jk/lib-terminal-lazy'
Junio C Hamano [Tue, 25 Mar 2014 18:08:09 +0000 (11:08 -0700)] 
Merge branch 'jk/lib-terminal-lazy'

The test helper lib-terminal always run an actual test_expect_* when
included, which screwed up with the use of skil-all that may have to
be done later.

* jk/lib-terminal-lazy:
  t/lib-terminal: make TTY a lazy prerequisite

10 years agoMerge branch 'dm/configure-iconv-locale-charset'
Junio C Hamano [Tue, 25 Mar 2014 18:07:51 +0000 (11:07 -0700)] 
Merge branch 'dm/configure-iconv-locale-charset'

* dm/configure-iconv-locale-charset:
  configure.ac: link with -liconv for locale_charset()

10 years agoMerge branch 'nd/commit-editor-cleanup'
Junio C Hamano [Tue, 25 Mar 2014 18:07:47 +0000 (11:07 -0700)] 
Merge branch 'nd/commit-editor-cleanup'

"git commit --cleanup=<mode>" learned a new mode, scissors.

* nd/commit-editor-cleanup:
  commit: add --cleanup=scissors
  wt-status.c: move cut-line print code out to wt_status_add_cut_line
  wt-status.c: make cut_line[] const to shrink .data section a bit

10 years agoMerge branch 'jk/warn-on-object-refname-ambiguity'
Junio C Hamano [Tue, 25 Mar 2014 18:07:36 +0000 (11:07 -0700)] 
Merge branch 'jk/warn-on-object-refname-ambiguity'

* jk/warn-on-object-refname-ambiguity:
  rev-list: disable object/refname ambiguity check with --stdin
  cat-file: restore warn_on_object_refname_ambiguity flag
  cat-file: fix a minor memory leak in batch_objects
  cat-file: refactor error handling of batch_objects

10 years agoMerge branch 'mh/remove-subtree-long-pathname-fix'
Junio C Hamano [Tue, 25 Mar 2014 18:07:09 +0000 (11:07 -0700)] 
Merge branch 'mh/remove-subtree-long-pathname-fix'

* mh/remove-subtree-long-pathname-fix:
  entry.c: fix possible buffer overflow in remove_subtree()
  checkout_entry(): use the strbuf throughout the function

10 years agoMerge branch 'nd/indent-fix-connect-c'
Junio C Hamano [Tue, 25 Mar 2014 18:07:05 +0000 (11:07 -0700)] 
Merge branch 'nd/indent-fix-connect-c'

* nd/indent-fix-connect-c:
  connect.c: SP after "}", not TAB

10 years agoMerge branch 'jk/mv-submodules-fix'
Junio C Hamano [Tue, 25 Mar 2014 18:02:01 +0000 (11:02 -0700)] 
Merge branch 'jk/mv-submodules-fix'

"git mv" that moves a submodule forgot to adjust the array that uses
to keep track of which submodules were to be moved to update its
configuration.

* jk/mv-submodules-fix:
  mv: prevent mismatched data when ignoring errors.
  builtin/mv: fix out of bounds write

10 years agoMerge branch 'cp/am-patch-format-doc'
Junio C Hamano [Tue, 25 Mar 2014 18:01:31 +0000 (11:01 -0700)] 
Merge branch 'cp/am-patch-format-doc'

* cp/am-patch-format-doc:
  Documentation/git-am: typofix
  Documentation/git-am: Document supported --patch-format options

10 years agopickaxe: simplify kwset loop in contains()
René Scharfe [Sat, 22 Mar 2014 17:16:00 +0000 (18:16 +0100)] 
pickaxe: simplify kwset loop in contains()

Inlining the variable "found" actually makes the code shorter and
easier to read.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agopickaxe: call strlen only when necessary in diffcore_pickaxe_count()
René Scharfe [Sat, 22 Mar 2014 17:15:59 +0000 (18:15 +0100)] 
pickaxe: call strlen only when necessary in diffcore_pickaxe_count()

We need to determine the search term's length only when fixed-string
matching is used; regular expression compilation takes a NUL-terminated
string directly.  Only call strlen() in the former case.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agopickaxe: move pickaxe() after pickaxe_match()
René Scharfe [Sat, 22 Mar 2014 17:15:58 +0000 (18:15 +0100)] 
pickaxe: move pickaxe() after pickaxe_match()

pickaxe() calls pickaxe_match(); moving the definition of the former
after the latter allows us to do without an explicit function
declaration.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agopickaxe: merge diffcore_pickaxe_grep() and diffcore_pickaxe_count() into diffcore_pic...
René Scharfe [Sat, 22 Mar 2014 17:15:57 +0000 (18:15 +0100)] 
pickaxe: merge diffcore_pickaxe_grep() and diffcore_pickaxe_count() into diffcore_pickaxe()

diffcore_pickaxe_count() initializes the regular expression or kwset for
the search term, calls pickaxe() with the callback has_changes() and
cleans up afterwards.  diffcore_pickaxe_grep() does the same, only it
doesn't support kwset and uses the callback diff_grep() instead.  Merge
the two functions to form the new diffcore_pickaxe() and thus get rid of
the duplicate regex setup and cleanup code.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agopickaxe: honor -i when used with -S and --pickaxe-regex
René Scharfe [Sat, 22 Mar 2014 17:15:56 +0000 (18:15 +0100)] 
pickaxe: honor -i when used with -S and --pickaxe-regex

accccde4 (pickaxe: allow -i to search in patch case-insensitively)
allowed case-insenitive matching for -G and -S, but for the latter
only if fixed string matching is used.  Allow it for -S and regular
expression matching as well to make the support complete.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4209: use helper functions to test --author
René Scharfe [Sat, 22 Mar 2014 17:15:55 +0000 (18:15 +0100)] 
t4209: use helper functions to test --author

Also add tests for case sensitive and non-matching cases.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4209: use helper functions to test --grep
René Scharfe [Sat, 22 Mar 2014 17:15:54 +0000 (18:15 +0100)] 
t4209: use helper functions to test --grep

Also add tests for non-matching cases.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4209: factor out helper function test_log_icase()
René Scharfe [Sat, 22 Mar 2014 17:15:53 +0000 (18:15 +0100)] 
t4209: factor out helper function test_log_icase()

Reduce code duplication by introducing test_log_icase() that runs the
same test with both --regexp-ignore-case and -i.  The specification of
the four basic test scenarios (matching/nomatching combined with case
sensitive/insensitive) becomes easier to read and write.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4209: factor out helper function test_log()
René Scharfe [Sat, 22 Mar 2014 17:15:52 +0000 (18:15 +0100)] 
t4209: factor out helper function test_log()

Twelve tests in t4209 follow the same simple pattern for description,
git log call and checking.  Extract that shared logic into a helper
function named test_log.  Test specifications become a lot more
compact, new tests can be added more easily.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agobranch.c: install_branch_config: simplify if chain
Adam [Mon, 10 Mar 2014 05:32:01 +0000 (01:32 -0400)] 
branch.c: install_branch_config: simplify if chain

Simplify if chain in install_branch_config().

Signed-off-by: Adam <Adam@sigterm.info>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4209: set up expectations up front
René Scharfe [Sat, 22 Mar 2014 17:15:51 +0000 (18:15 +0100)] 
t4209: set up expectations up front

Instead of creating an expect file for each test, build three files with
the possible valid values during setup and use them in the tests.  This
shortens the test code and saves nine calls to git rev-parse.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot1502: protect runs of SPs used in the indentation
Junio C Hamano [Sun, 23 Mar 2014 22:26:36 +0000 (15:26 -0700)] 
t1502: protect runs of SPs used in the indentation

The expected output from the argument help use runs of SPs to align
the description of each option; a careless use of --whitespace=fix
can turn leading parts of them into appropriate number of HTs.
Prevent such a breakage by prefixing all the expected lines with
leading vertical bars in the original and stripping them with a
small sed script.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agorev-parse --parseopt: option argument name hints
Ilya Bobyr [Sat, 22 Mar 2014 09:47:34 +0000 (02:47 -0700)] 
rev-parse --parseopt: option argument name hints

Built-in commands can specify names for option arguments when usage text
is generated for a command.  sh based commands should be able to do the
same.

Option argument name hint is any text that comes after [*=?!] after the
argument name up to the first whitespace.

Signed-off-by: Ilya Bobyr <ilya.bobyr@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoMakefile: describe CHARSET_LIB better
Дилян Палаузов [Tue, 11 Mar 2014 22:37:33 +0000 (23:37 +0100)] 
Makefile: describe CHARSET_LIB better

The original explanation was not even grammatically correct or
readable.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agouserdiff: have 'cpp' hunk header pattern catch more C++ anchor points
Johannes Sixt [Fri, 21 Mar 2014 21:07:22 +0000 (22:07 +0100)] 
userdiff: have 'cpp' hunk header pattern catch more C++ anchor points

The hunk header pattern 'cpp' is intended for C and C++ source code, but
it is actually not particularly useful for the latter, and even misses
some use-cases for the former.

The parts of the pattern have the following flaws:

- The first part matches an identifier followed immediately by a colon
  and arbitrary text and is intended to reject goto labels and C++
  access specifiers (public, private, protected). But this pattern also
  rejects C++ constructs, which look like this:

    MyClass::MyClass()
    MyClass::~MyClass()
    MyClass::Item MyClass::Find(...

- The second part matches an identifier followed by a list of qualified
  names (i.e. identifiers separated by the C++ scope operator '::')
  separated by space or '*' followed by an opening parenthesis (with
  space between the tokens). It matches function declarations like

    struct item* get_head(...
    int Outer::Inner::Func(...

  Since the pattern requires at least two identifiers, GNU-style
  function definitions are ignored:

    void
    func(...

  Moreover, since the pattern does not allow punctuation other than '*',
  the following C++ constructs are not recognized:

  . template definitions:
      template<class T> int func(T arg)

  . functions returning references:
      const string& get_message()

  . functions returning templated types:
      vector<int> foo()

  . operator definitions:
      Value operator+(Value l, Value r)

- The third part of the pattern finally matches compound definitions.
  But it forgets about unions and namespaces, and also skips single-line
  definitions

    struct random_iterator_tag {};

  because no semicolon can occur on the line.

Change the first pattern to require a colon at the end of the line
(except for trailing space and comments), so that it does not reject
constructor or destructor definitions.

Notice that all interesting anchor points begin with an identifier or
keyword. But since there is a large variety of syntactical constructs
after the first "word", the simplest is to require only this word and
accept everything else. Therefore, this boils down to a line that begins
with a letter or underscore (optionally preceded by the C++ scope
operator '::' to accept functions returning a type anchored at the
global namespace). Replace the second and third part by a single pattern
that picks such a line.

This has the following desirable consequence:

- All constructs mentioned above are recognized.

and the following likely desirable consequences:

- Definitions of global variables and typedefs are recognized:

    int num_entries = 0;
    extern const char* help_text;
    typedef basic_string<wchar_t> wstring;

- Commonly used marco-ized boilerplate code is recognized:

    BEGIN_MESSAGE_MAP(CCanvas,CWnd)
    Q_DECLARE_METATYPE(MyStruct)
    PATTERNS("tex",...)

  (The last one is from this very patch.)

but also the following possibly undesirable consequence:

- When a label is not on a line by itself (except for a comment) it is
  no longer rejected, but can appear as a hunk header if it occurs at
  the beginning of a line:

    next:;

IMO, the benefits of the change outweigh the (possible) regressions by a
large margin.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4018: test cases showing that the cpp pattern misses many anchor points
Johannes Sixt [Fri, 21 Mar 2014 21:07:21 +0000 (22:07 +0100)] 
t4018: test cases showing that the cpp pattern misses many anchor points

Most of the tests show C++ code, but there is also a union definition and
a GNU style function definition that are not recognized.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4018: test cases for the built-in cpp pattern
Johannes Sixt [Fri, 21 Mar 2014 21:07:20 +0000 (22:07 +0100)] 
t4018: test cases for the built-in cpp pattern

A later patch changes the built-in cpp pattern. These test cases
demonstrate aspects of the pattern that we do not want to change.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4018: reduce test files for pattern compilation tests
Johannes Sixt [Fri, 21 Mar 2014 21:07:19 +0000 (22:07 +0100)] 
t4018: reduce test files for pattern compilation tests

All test cases that need a file with specific text patterns have been
converted to utilize texts in the t4018/ directory. The remaining tests
in the test script deal only with the validity of the regular
expressions. These tests do not depend on the contents of files that
'git diff' is invoked on. Remove the largish here-document and use only
tiny files.

While we are touching these tests, convert grep to test_i18ngrep as the
texts checked for may undergo translation in the future.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4018: convert custom pattern test to the new infrastructure
Johannes Sixt [Fri, 21 Mar 2014 21:07:18 +0000 (22:07 +0100)] 
t4018: convert custom pattern test to the new infrastructure

For the test case "matches to end of line", extend the pattern by a few
wildcards so that the pattern captures the "RIGHT" token, which is needed
for verification, without mentioning it in the pattern.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4018: convert java pattern test to the new infrastructure
Johannes Sixt [Fri, 21 Mar 2014 21:07:17 +0000 (22:07 +0100)] 
t4018: convert java pattern test to the new infrastructure

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4018: convert perl pattern tests to the new infrastructure
Johannes Sixt [Fri, 21 Mar 2014 21:07:16 +0000 (22:07 +0100)] 
t4018: convert perl pattern tests to the new infrastructure

There is one subtlety: The old test case 'perl pattern gets full line of
POD header' does not have its own new test case, but the feature is
tested nevertheless by placing the RIGHT tag at the end of the expected
hunk header in t4018/perl-skip-sub-in-pod.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot4018: an infrastructure to test hunk headers
Johannes Sixt [Fri, 21 Mar 2014 21:07:15 +0000 (22:07 +0100)] 
t4018: an infrastructure to test hunk headers

Add an infrastructure that simplifies adding new tests of the hunk
header regular expressions.

To add new tests, a file with the syntax to test can be dropped in the
directory t4018. The README file explains how a test file must contain;
the README itself tests the default behavior.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agouserdiff: support unsigned and long long suffixes of integer constants
Johannes Sixt [Fri, 21 Mar 2014 21:07:14 +0000 (22:07 +0100)] 
userdiff: support unsigned and long long suffixes of integer constants

Do not split constants such as 123U, 456ll, 789UL at the first U or
second L.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agouserdiff: support C++ ->* and .* operators in the word regexp
Johannes Sixt [Fri, 21 Mar 2014 21:07:13 +0000 (22:07 +0100)] 
userdiff: support C++ ->* and .* operators in the word regexp

The character sequences ->* and .* are valid C++ operators. Keep them
together in --word-diff mode.

Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot0001: drop subshells just for "cd"
Jeff King [Thu, 20 Mar 2014 23:23:06 +0000 (19:23 -0400)] 
t0001: drop subshells just for "cd"

Many tests do something like:

  (
mkdir foo &&
cd foo &&
git init
  )

You can do the same these days with "git init foo", which
makes the tests shorter and simpler to read.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot0001: drop useless subshells
Jeff King [Thu, 20 Mar 2014 23:21:25 +0000 (19:21 -0400)] 
t0001: drop useless subshells

Many tests use subshells, but don't actually change the
shell environment. They were probably cargo-culted from
earlier tests which did need subshells. Drop the useless
ones.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot0001: use test_must_fail
Jeff King [Thu, 20 Mar 2014 23:19:50 +0000 (19:19 -0400)] 
t0001: use test_must_fail

We've hand-rolled several "if" statements looking for
failures. We can use test_must_fail here, which is shorter
and more robust.

Note that we modify the commands slightly (to use "git init
foo" rather than "cd foo && git init") to avoid dealing with
a subshell, but this should not affect the outcome.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot0001: use test_config_global
Jeff King [Thu, 20 Mar 2014 23:18:12 +0000 (19:18 -0400)] 
t0001: use test_config_global

We hand-set several config options using :

  git config -f $HOME/.gitconfig ...

Instead, we can use "test_config_global". Not only is this
more readable, but it cleans up for us so that subsequent
tests aren't polluted by our settings.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot0001: use test_path_is_*
Jeff King [Thu, 20 Mar 2014 23:17:35 +0000 (19:17 -0400)] 
t0001: use test_path_is_*

t0001 predates the test_path_is_* helpers, and uses "test
-f" and "test -d" directly. Using the helpers provides
better debugging output, and are a little more robust.
As opposed to "! test -d", test_path_is_missing will
actually makes sure the path does not exist at all.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot0001: make symlink reinit test more careful
Jeff King [Thu, 20 Mar 2014 23:17:15 +0000 (19:17 -0400)] 
t0001: make symlink reinit test more careful

In the final test of t0001, we have a repo whose .git is a
symlink to a directory "here", and we use
"--separate-git-dir" to migrate that to a .git file pointing
to a different directory. We check that the data is migrated
to the new directory and that .git looks like a git-file.

We also check that "here" is not a directory, which is
slightly misleading. It should not be a directory, but
neither should it be gone. It is the actual resting place of
the git-file, and .git remains a symlink to it.

Let's check that more explicitly, both to make our test more
robust, and to make further cleanups in this area more
obvious.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot: prefer "git config --file" to GIT_CONFIG
Jeff King [Thu, 20 Mar 2014 23:17:01 +0000 (19:17 -0400)] 
t: prefer "git config --file" to GIT_CONFIG

Doing:

  GIT_CONFIG=foo git config ...

is equivalent to:

  git config --file=foo ...

The latter is easier to read and slightly less error-prone,
because of issues with one-shot variables and shell
functions (e.g., you cannot use the former with
test_must_fail).

Note that we explicitly leave one case in t1300 which checks
the same operation on both GIT_CONFIG and "git config
--file". They are equivalent in the code these days, but
this will make sure it remains so.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot: prefer "git config --file" to GIT_CONFIG with test_must_fail
Jeff King [Thu, 20 Mar 2014 23:15:54 +0000 (19:15 -0400)] 
t: prefer "git config --file" to GIT_CONFIG with test_must_fail

This lets us get rid of an extra "env" invocation in the
middle, and is slightly more readable.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot: stop using GIT_CONFIG to cross repo boundaries
Jeff King [Thu, 20 Mar 2014 23:15:24 +0000 (19:15 -0400)] 
t: stop using GIT_CONFIG to cross repo boundaries

Some tests want to check or set config in another
repository. E.g., t1000 creates repositories and makes sure
that their core.bare and core.worktree settings are what we
expect. We can do this with:

  GIT_CONFIG=$repo/.git/config git config ...

but it better shows the intent to just enter the repository
and let "git config" do the normal lookups:

  (cd $repo && git config ...)

In theory, this would cause us to use an extra subshell, but
in all such cases, we are actually already in a subshell.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot: drop useless sane_unset GIT_* calls
Jeff King [Thu, 20 Mar 2014 23:14:33 +0000 (19:14 -0400)] 
t: drop useless sane_unset GIT_* calls

Several test scripts manually unset GIT_CONFIG and other
GIT_* variables. These are generally taken care of for us by
test-lib.sh already.

Unsetting these is not only useless, but can be confusing to
a reader, who may wonder why some tests in a script unset
them and others do not (t0001 is particularly guilty of this
inconsistency, probably because many of its tests predate
the test-lib.sh environment-cleansing).

Note that we cannot always get rid of such unsetting. For
example, t9130 can drop the GIT_CONFIG unset, but not the
GIT_DIR one, because lib-git-svn.sh sets the latter. And in
t1000, we unset GIT_TEMPLATE_DIR, which is explicitly
initialized by test-lib.sh.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot/test-lib: drop redundant unset of GIT_CONFIG
Jeff King [Thu, 20 Mar 2014 23:13:36 +0000 (19:13 -0400)] 
t/test-lib: drop redundant unset of GIT_CONFIG

This is already handled by the mass GIT_* unsetting added by
95a1d12 (tests: scrub environment of GIT_* variables,
2011-03-15).

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agot/Makefile: stop setting GIT_CONFIG
Jeff King [Thu, 20 Mar 2014 23:13:21 +0000 (19:13 -0400)] 
t/Makefile: stop setting GIT_CONFIG

Once upon a time, the setting of GIT_CONFIG in the
environment could affect how tests ran. Commit 9c3796f (Fix
setting config variables with an alternative GIT_CONFIG,
2006-06-20) unconditionally set GIT_CONFIG in the Makefile
when running tests to give us a known starting point.

This is insufficient for running the tests outside of the
Makefile, however, and 8565d2d (Make tests independent of
global config files, 2007-02-15) later set GIT_CONFIG
directly in test-lib.sh. At that point the Makefile setting
was redundant, but we never removed it. Let's do so now.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
10 years agoUpdate draft release notes to 2.0
Junio C Hamano [Fri, 21 Mar 2014 20:41:27 +0000 (13:41 -0700)] 
Update draft release notes to 2.0

10 years agoMerge branch 'lt/request-pull'
Junio C Hamano [Fri, 21 Mar 2014 19:50:44 +0000 (12:50 -0700)] 
Merge branch 'lt/request-pull'

Discard the accumulated "heuristics" to guess from which branch the
result wants to be pulled from and make sure what the end user
specified is not second-guessed by "git request-pull", to avoid
mistakes.

* lt/request-pull:
  request-pull: documentation updates
  request-pull: resurrect "pretty refname" feature
  request-pull: test updates
  request-pull: pick up tag message as before
  request-pull: allow "local:remote" to specify names on both ends
  request-pull: more strictly match local/remote branches

10 years agoMerge branch 'es/sh-i18n-envsubst'
Junio C Hamano [Fri, 21 Mar 2014 19:50:39 +0000 (12:50 -0700)] 
Merge branch 'es/sh-i18n-envsubst'

* es/sh-i18n-envsubst:
  sh-i18n--envsubst: retire unused string_list_member()

10 years agoMerge branch 'nd/upload-pack-shallow'
Junio C Hamano [Fri, 21 Mar 2014 19:49:08 +0000 (12:49 -0700)] 
Merge branch 'nd/upload-pack-shallow'

Serving objects from a shallow repository needs to write a
temporary file to be used, but the serving upload-pack may not have
write access to the repository which is meant to be read-only.

Instead feed these temporary shallow bounds from the standard input
of pack-objects so that we do not have to use a temporary file.

* nd/upload-pack-shallow:
  upload-pack: send shallow info over stdin to pack-objects

10 years agoMerge branch 'jn/wt-status'
Junio C Hamano [Fri, 21 Mar 2014 19:48:59 +0000 (12:48 -0700)] 
Merge branch 'jn/wt-status'

Unify the codepaths that format new/modified/changed sections and
conflicted paths in the "git status" output and make it possible to
properly internationalize their output.

* jn/wt-status:
  wt-status: lift the artificual "at least 20 columns" floor
  wt-status: i18n of section labels
  wt-status: extract the code to compute width for labels
  wt-status: make full label string to be subject to l10n

10 years agoMerge branch 'jc/stash-pop-not-popped'
Junio C Hamano [Fri, 21 Mar 2014 19:48:51 +0000 (12:48 -0700)] 
Merge branch 'jc/stash-pop-not-popped'

"stash pop", upon failing to apply the stash, refrains from
discarding the stash to avoid information loss.  Be more explicit
in the error message.

The wording may want to get a bit more bikeshedding.

* jc/stash-pop-not-popped:
  stash pop: mention we did not drop the stash upon failing to apply

10 years agoMerge branch 'dk/skip-prefix-scan-only-once'
Junio C Hamano [Fri, 21 Mar 2014 19:47:41 +0000 (12:47 -0700)] 
Merge branch 'dk/skip-prefix-scan-only-once'

Update implementation of skip_prefix() to scan only once; given
that most "prefix" arguments to the inline function are constant
strings whose strlen() can be determined at the compile time, this
might actually make things worse with a compiler with sufficient
intelligence.

* dk/skip-prefix-scan-only-once:
  skip_prefix(): scan prefix only once

10 years agoMerge branch 'nd/tag-version-sort'
Junio C Hamano [Fri, 21 Mar 2014 19:47:38 +0000 (12:47 -0700)] 
Merge branch 'nd/tag-version-sort'

Allow v1.9.0 sorted before v1.10.0 in "git tag --list" output.

* nd/tag-version-sort:
  tag: support --sort=<spec>

10 years agoMerge branch 'jk/shallow-update-fix'
Junio C Hamano [Fri, 21 Mar 2014 19:33:29 +0000 (12:33 -0700)] 
Merge branch 'jk/shallow-update-fix'

Serving objects from a shallow repository needs to write a
new file to hold the temporary shallow boundaries but it was not
cleaned when we exit due to die() or a signal.

* jk/shallow-update-fix:
  shallow: verify shallow file after taking lock
  shallow: automatically clean up shallow tempfiles
  shallow: use stat_validity to check for up-to-date file

10 years agoMerge branch 'tc/commit-dry-run-exit-status-tests'
Junio C Hamano [Fri, 21 Mar 2014 19:33:25 +0000 (12:33 -0700)] 
Merge branch 'tc/commit-dry-run-exit-status-tests'

* tc/commit-dry-run-exit-status-tests:
  demonstrate git-commit --dry-run exit code behaviour

10 years agoconfig.txt: third-party tools may and do use their own variables
Chris Angelico [Fri, 21 Mar 2014 04:07:08 +0000 (15:07 +1100)] 
config.txt: third-party tools may and do use their own variables

Signed-off-by: Chris Angelico <rosuav@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>