git/git.git
6 years agoMerge branch 'ak/show-branch-usage-string'
Junio C Hamano [Wed, 21 Jan 2015 00:16:09 +0000 (16:16 -0800)]
Merge branch 'ak/show-branch-usage-string'

* ak/show-branch-usage-string:
  show-branch: fix indentation of usage string

6 years agoshow-branch: fix indentation of usage string
Ralf Thielow [Tue, 20 Jan 2015 19:30:28 +0000 (20:30 +0100)]
show-branch: fix indentation of usage string

Noticed-by: Jean-Noël Avila <jn.avila@free.fr>
Signed-off-by: Ralf Thielow <ralf.thielow@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoMerge branch 'jk/colors'
Junio C Hamano [Tue, 20 Jan 2015 23:57:22 +0000 (15:57 -0800)]
Merge branch 'jk/colors'

* jk/colors:
  parse_color: fix return value for numeric color values 0-8

6 years agoparse_color: fix return value for numeric color values 0-8
Jeff King [Tue, 20 Jan 2015 22:14:48 +0000 (17:14 -0500)]
parse_color: fix return value for numeric color values 0-8

When commit 695d95d refactored the color parsing, it missed
a "return 0" when parsing literal numbers 0-8 (which
represent basic ANSI colors), leading us to report these
colors as an error.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agogit-svn: make it play nicely with submodules
Ramkumar Ramachandra [Sat, 10 Jan 2015 14:55:11 +0000 (09:55 -0500)]
git-svn: make it play nicely with submodules

It's a simple matter of opening the directory specified in the gitfile.

[ew: tweaked check to avoid open() on directories]

Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
Signed-off-by: Eric Wong <normalperson@yhbt.net>
6 years agoGit::SVN: handle missing ref_id case correctly
Ramkumar Ramachandra [Sat, 10 Jan 2015 14:35:10 +0000 (09:35 -0500)]
Git::SVN: handle missing ref_id case correctly

ref_id should not match "refs/remotes/".

[ew: dropped initial hunk for GIT_SVN_ID at Ramkumar's request]

Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
Signed-off-by: Eric Wong <normalperson@yhbt.net>
6 years agoFifth batch for 2.3 cycle
Junio C Hamano [Wed, 14 Jan 2015 20:44:03 +0000 (12:44 -0800)]
Fifth batch for 2.3 cycle

Hopefully this will be the final feature update for 2.3-rc1

Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoMerge branch 'po/doc-core-ignorestat'
Junio C Hamano [Wed, 14 Jan 2015 20:41:38 +0000 (12:41 -0800)]
Merge branch 'po/doc-core-ignorestat'

* po/doc-core-ignorestat:
  doc: core.ignoreStat update, and clarify the --assume-unchanged effect
  doc: core.ignoreStat clarify the --assume-unchanged effect

6 years agoMerge branch 'rc/for-each-ref-tracking'
Junio C Hamano [Wed, 14 Jan 2015 20:39:02 +0000 (12:39 -0800)]
Merge branch 'rc/for-each-ref-tracking'

* rc/for-each-ref-tracking:
  for-each-ref: always check stat_tracking_info()'s return value

6 years agoMerge branch 'rh/autoconf-rhel3'
Junio C Hamano [Wed, 14 Jan 2015 20:37:21 +0000 (12:37 -0800)]
Merge branch 'rh/autoconf-rhel3'

Build update for older RHEL.

* rh/autoconf-rhel3:
  configure.ac: check for HMAC_CTX_cleanup
  configure.ac: check for clock_gettime and CLOCK_MONOTONIC
  configure.ac: check 'tv_nsec' field in 'struct stat'

6 years agoMerge branch 'ak/fewer-includes'
Junio C Hamano [Wed, 14 Jan 2015 20:37:19 +0000 (12:37 -0800)]
Merge branch 'ak/fewer-includes'

* ak/fewer-includes:
  cat-file: remove unused includes
  git.c: remove unnecessary #includes

6 years agoMerge branch 'ak/doc-add-v-n-options'
Junio C Hamano [Wed, 14 Jan 2015 20:37:13 +0000 (12:37 -0800)]
Merge branch 'ak/doc-add-v-n-options'

* ak/doc-add-v-n-options:
  Documentation: list long options for -v and -n

6 years agoMerge branch 'ak/show-branch-usage-string'
Junio C Hamano [Wed, 14 Jan 2015 20:37:07 +0000 (12:37 -0800)]
Merge branch 'ak/show-branch-usage-string'

* ak/show-branch-usage-string:
  show-branch: line-wrap show-branch usage

6 years agoMerge branch 'rh/test-color-avoid-terminfo-in-original-home'
Junio C Hamano [Wed, 14 Jan 2015 20:36:45 +0000 (12:36 -0800)]
Merge branch 'rh/test-color-avoid-terminfo-in-original-home'

We try to see if "tput" gives a useful result before switching TERM
to dumb and moving HOME to point to our fake location for stability
of the tests, and then use the command when coloring the output
from the tests, but there is no guarantee "tput" works after
switching HOME.

* rh/test-color-avoid-terminfo-in-original-home:
  test-lib.sh: do tests for color support after changing HOME
  test-lib: use 'test ...' instead of '[ ... ]'

6 years agoMerge branch 'tf/prompt-preserve-exit-status'
Junio C Hamano [Wed, 14 Jan 2015 20:35:48 +0000 (12:35 -0800)]
Merge branch 'tf/prompt-preserve-exit-status'

Using the exit status of the last command in the prompt, e.g.
PS1='$(__git_ps1) $? ', did not work well because the helper
function stomped on the exit status.

* tf/prompt-preserve-exit-status:
  git-prompt: preserve value of $? in all cases

6 years agoMerge branch 'rh/hide-prompt-in-ignored-directory'
Junio C Hamano [Wed, 14 Jan 2015 20:34:01 +0000 (12:34 -0800)]
Merge branch 'rh/hide-prompt-in-ignored-directory'

* rh/hide-prompt-in-ignored-directory:
  git-prompt.sh: allow to hide prompt for ignored pwd
  git-prompt.sh: if pc mode, immediately set PS1 to a plain prompt

6 years agoMerge branch 'mm/complete-rebase-autostash'
Junio C Hamano [Wed, 14 Jan 2015 20:33:57 +0000 (12:33 -0800)]
Merge branch 'mm/complete-rebase-autostash'

* mm/complete-rebase-autostash:
  git-completion: add --autostash for 'git rebase'

6 years agoMerge branch 'aw/doc-smtp-ssl-cert-path'
Junio C Hamano [Wed, 14 Jan 2015 20:33:50 +0000 (12:33 -0800)]
Merge branch 'aw/doc-smtp-ssl-cert-path'

A long overdue documentation update to match an age-old code
update.

* aw/doc-smtp-ssl-cert-path:
  correct smtp-ssl-cert-path description

6 years agoMerge branch 'sp/subtree-doc'
Junio C Hamano [Wed, 14 Jan 2015 20:33:45 +0000 (12:33 -0800)]
Merge branch 'sp/subtree-doc'

* sp/subtree-doc:
  subtree: fix AsciiDoc list item continuation

6 years agoMerge branch 'km/log-usage-string-i18n'
Junio C Hamano [Wed, 14 Jan 2015 20:32:39 +0000 (12:32 -0800)]
Merge branch 'km/log-usage-string-i18n'

* km/log-usage-string-i18n:
  log.c: fix translation markings

6 years agoMerge branch 'km/imap-send-libcurl-options'
Junio C Hamano [Wed, 14 Jan 2015 20:31:50 +0000 (12:31 -0800)]
Merge branch 'km/imap-send-libcurl-options'

Now imap-send learned to talk to the server using cURL library,
allow the same GIT_CURL_VERBOSE environment variable to control the
verbosity of the chattering.

* km/imap-send-libcurl-options:
  imap-send.c: set CURLOPT_USE_SSL to CURLUSESSL_TRY
  imap-send.c: support GIT_CURL_VERBOSE

6 years agoMerge branch 'jk/prune-packed-server-info'
Junio C Hamano [Wed, 14 Jan 2015 20:30:26 +0000 (12:30 -0800)]
Merge branch 'jk/prune-packed-server-info'

Fix recent breakage in Git 2.2 that started creating info/refs and
objects/info/packs files with permission bits tighter than user's
umask.

* jk/prune-packed-server-info:
  update-server-info: create info/* with mode 0666
  t1301: set umask in reflog sharedrepository=group test

6 years agoMerge branch 'js/remote-add-with-insteadof'
Junio C Hamano [Wed, 14 Jan 2015 20:29:47 +0000 (12:29 -0800)]
Merge branch 'js/remote-add-with-insteadof'

"git remote add $name $URL" is now allowed when "url.$URL.insteadOf"
is already defined.

* js/remote-add-with-insteadof:
  Add a regression test for 'git remote add <existing> <same-url>'
  git remote: allow adding remotes agreeing with url.<...>.insteadOf

6 years agogit-prompt: preserve value of $? in all cases
Tony Finch [Wed, 14 Jan 2015 10:06:28 +0000 (10:06 +0000)]
git-prompt: preserve value of $? in all cases

Signed-off-by: Tony Finch <dot@dotat.at>
Reviewed-by: SZEDER Gábor <szeder@ira.uka.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agodoc: core.ignoreStat update, and clarify the --assume-unchanged effect
Philip Oakley [Mon, 5 Jan 2015 22:22:53 +0000 (22:22 +0000)]
doc: core.ignoreStat update, and clarify the --assume-unchanged effect

The assume-unchanged bit, and consequently core.ignoreStat, can be
misunderstood. Be assertive about the expectation that file changes should
notified to Git.

Overhaul the general wording thus:
    1. direct description of what is ignored given first.
    2. example instruction of the user manual action required.
    3. use sideways indirection for assume-unchanged and update-index
       references.
    4. add a 'normally' to give leeway for the change detection.

Signed-off-by: Philip Oakley <philipoakley@iee.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agofor-each-ref: always check stat_tracking_info()'s return value
Raphael Kubo da Costa [Mon, 5 Jan 2015 09:58:55 +0000 (11:58 +0200)]
for-each-ref: always check stat_tracking_info()'s return value

The code handling %(upstream:track) and %(upstream:trackshort)
assumed that it always had a valid branch that had been sanitized
earlier in populate_value(), and thus did not check the return value
of the call to stat_tracking_info().

While there is indeed some sanitization code that basically
corresponds to stat_tracking_info() returning 0 (no base branch
set), the function can also return -1 when the base branch did exist
but has since then been deleted.

In this case, num_ours and num_theirs had undefined values and a
call to `git for-each-ref --format="%(upstream:track)"` could print
spurious values such as

  [behind -111794512]
  [ahead 38881640, behind 5103867]

even for repositories with one single commit.

Verify stat_tracking_info()'s return value and do not print anything
if it returns -1. This behavior also matches the documentation ("has
no effect if the ref does not have tracking information associated
with it").

Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Raphael Kubo da Costa <raphael.kubo.da.costa@intel.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoGit 2.3.0-rc0 v2.3.0-rc0
Junio C Hamano [Mon, 12 Jan 2015 22:12:42 +0000 (14:12 -0800)]
Git 2.3.0-rc0

Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoSync with 2.2.2
Junio C Hamano [Mon, 12 Jan 2015 22:08:42 +0000 (14:08 -0800)]
Sync with 2.2.2

6 years agoGit 2.2.2 v2.2.2
Junio C Hamano [Mon, 12 Jan 2015 22:06:12 +0000 (14:06 -0800)]
Git 2.2.2

Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoMerge branch 'jk/read-packed-refs-without-path-max' into maint
Junio C Hamano [Mon, 12 Jan 2015 22:02:54 +0000 (14:02 -0800)]
Merge branch 'jk/read-packed-refs-without-path-max' into maint

* jk/read-packed-refs-without-path-max:
  read_packed_refs: use skip_prefix instead of static array
  read_packed_refs: pass strbuf to parse_ref_line
  read_packed_refs: use a strbuf for reading lines

6 years agoMerge branch 'mg/add-ignore-errors' into maint
Junio C Hamano [Mon, 12 Jan 2015 22:02:19 +0000 (14:02 -0800)]
Merge branch 'mg/add-ignore-errors' into maint

* mg/add-ignore-errors:
  add: ignore only ignored files

6 years agoMerge branch 'mh/find-uniq-abbrev' into maint
Junio C Hamano [Mon, 12 Jan 2015 22:02:05 +0000 (14:02 -0800)]
Merge branch 'mh/find-uniq-abbrev' into maint

* mh/find-uniq-abbrev:
  sha1_name: avoid unnecessary sha1 lookup in find_unique_abbrev

6 years agoMerge branch 'jk/approxidate-avoid-y-d-m-over-future-dates' into maint
Junio C Hamano [Mon, 12 Jan 2015 22:01:18 +0000 (14:01 -0800)]
Merge branch 'jk/approxidate-avoid-y-d-m-over-future-dates' into maint

* jk/approxidate-avoid-y-d-m-over-future-dates:
  approxidate: allow ISO-like dates far in the future
  pass TIME_DATE_NOW to approxidate future-check

6 years agoMerge branch 'rw/apply-does-not-take-ignore-date' into maint
Junio C Hamano [Mon, 12 Jan 2015 22:00:16 +0000 (14:00 -0800)]
Merge branch 'rw/apply-does-not-take-ignore-date' into maint

* rw/apply-does-not-take-ignore-date:
  git-am.txt: --ignore-date flag is not passed to git-apply

6 years agoMerge branch 'jk/for-each-reflog-ent-reverse' into maint
Junio C Hamano [Mon, 12 Jan 2015 20:19:17 +0000 (12:19 -0800)]
Merge branch 'jk/for-each-reflog-ent-reverse' into maint

* jk/for-each-reflog-ent-reverse:
  for_each_reflog_ent_reverse: turn leftover check into assertion
  for_each_reflog_ent_reverse: fix newlines on block boundaries

6 years agoMerge branch 'bc/fetch-thin-less-aggressive-in-normal-repository'
Junio C Hamano [Mon, 12 Jan 2015 19:38:56 +0000 (11:38 -0800)]
Merge branch 'bc/fetch-thin-less-aggressive-in-normal-repository'

Earlier we made "rev-list --object-edge" more aggressively list the
objects at the edge commits, in order to reduce number of objects
fetched into a shallow repository, but the change affected cases
other than "fetching into a shallow repository" and made it
unusably slow (e.g. fetching into a normal repository should not
have to suffer the overhead from extra processing).  Limit it to a
more specific case by introducing --objects-edge-aggressive, a new
option to rev-list.

* bc/fetch-thin-less-aggressive-in-normal-repository:
  pack-objects: use --objects-edge-aggressive for shallow repos
  rev-list: add an option to mark fewer edges as uninteresting
  Documentation: add missing article in rev-list-options.txt

6 years agoMerge branch 'sb/doc-submitting-patches-keep-notes'
Junio C Hamano [Mon, 12 Jan 2015 19:38:54 +0000 (11:38 -0800)]
Merge branch 'sb/doc-submitting-patches-keep-notes'

* sb/doc-submitting-patches-keep-notes:
  SubmittingPatches: explain rationale for using --notes with format-patch

6 years agoMerge branch 'rs/simplify-transport-get'
Junio C Hamano [Mon, 12 Jan 2015 19:38:51 +0000 (11:38 -0800)]
Merge branch 'rs/simplify-transport-get'

* rs/simplify-transport-get:
  transport: simplify duplicating a substring in transport_get() using xmemdupz()

6 years agoMerge branch 'rs/simplify-parsing-commit-tree-S'
Junio C Hamano [Mon, 12 Jan 2015 19:38:39 +0000 (11:38 -0800)]
Merge branch 'rs/simplify-parsing-commit-tree-S'

* rs/simplify-parsing-commit-tree-S:
  commit-tree: simplify parsing of option -S using skip_prefix()

6 years agoMerge branch 'rs/plug-strbuf-leak-in-merge'
Junio C Hamano [Mon, 12 Jan 2015 19:38:32 +0000 (11:38 -0800)]
Merge branch 'rs/plug-strbuf-leak-in-merge'

* rs/plug-strbuf-leak-in-merge:
  merge: release strbuf after use in suggest_conflicts()

6 years agoMerge branch 'rs/plug-strbuf-leak-in-lock-ref'
Junio C Hamano [Mon, 12 Jan 2015 19:38:30 +0000 (11:38 -0800)]
Merge branch 'rs/plug-strbuf-leak-in-lock-ref'

* rs/plug-strbuf-leak-in-lock-ref:
  refs: plug strbuf leak in lock_ref_sha1_basic()

6 years agoMerge branch 'es/checkout-index-temp'
Junio C Hamano [Mon, 12 Jan 2015 19:38:23 +0000 (11:38 -0800)]
Merge branch 'es/checkout-index-temp'

"git checkout-index --temp=$target $path" did not work correctly
for paths outside the current subdirectory in the project.

* es/checkout-index-temp:
  checkout-index: fix --temp relative path mangling
  t2004: demonstrate broken relative path printing
  t2004: standardize file naming in symlink test
  t2004: drop unnecessary write-tree/read-tree
  t2004: modernize style

6 years agoMerge branch 'cc/bisect-rev-parsing'
Junio C Hamano [Mon, 12 Jan 2015 19:38:14 +0000 (11:38 -0800)]
Merge branch 'cc/bisect-rev-parsing'

The logic in "git bisect bad HEAD" etc. to avoid forcing the test
of the common ancestor of bad and good commits was broken.

* cc/bisect-rev-parsing:
  bisect: add test to check that revs are properly parsed
  bisect: parse revs before passing them to check_expected_revs()

6 years agoDocumentation: list long options for -v and -n
Alexander Kuleshov [Fri, 9 Jan 2015 07:48:40 +0000 (13:48 +0600)]
Documentation: list long options for -v and -n

Signed-off-by: Alexander Kuleshov <kuleshovmail@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agocat-file: remove unused includes
Alexander Kuleshov [Thu, 8 Jan 2015 18:56:06 +0000 (00:56 +0600)]
cat-file: remove unused includes

 - "exec_cmd.h" became unnecessary at b931aa5a (Call builtin ls-tree
   in git-cat-file -p, 2006-05-26), when it changed an earlier code
   that delegated tree display to "ls-tree" via the run_command()
   API (hence needing "exec_cmd.h") to call cmd_ls_tree() directly.
   We should have removed the include in the same commit, but we
   forgot to do so.

 - "diff.h" was added at e5fba602 (textconv: support for cat_file,
   2010-06-15), together with "userdiff.h", but "userdiff.h" can be
   included without including "diff.h"; the header was unnecessary
   from the beginning.

 - "tag.h" and "tree.h" were necessary since 8e440259 (Use blob_,
   commit_, tag_, and tree_type throughout., 2006-04-02) to check
   the type of object by comparing typename with tree_type and
   tag_type (pointers to extern strings).

   21666f1a (convert object type handling from a string to a number,
   2007-02-26) made these <type>_type strings unnecessary, and it
   could have switched to include "object.h", which is necessary to
   use typename(), but it forgot to do so.  Because "tag.h" and
   "tree.h" include "object.h", it did not need to explicitly
   include "object.h" in order to start using typename() itself.

   We do not even have to include "object.h" after removing these
   two #includes, because "builtin.h" includes "commit.h" which in
   turn includes "object.h" these days.  This happened at 7b9c0a69
   (git-commit-tree: make it usable from other builtins,
   2008-07-01).

Signed-off-by: Alexander Kuleshov <kuleshovmail@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agogit.c: remove unnecessary #includes
Alexander Kuleshov [Fri, 9 Jan 2015 10:12:33 +0000 (16:12 +0600)]
git.c: remove unnecessary #includes

"cache.h" and "commit.h" are already included via "builtin.h".

We started to include "quote.h" at 575ba9d6 (GIT_TRACE: show which
built-in/external commands are executed, 2006-06-25) that wanted to
use sq_quote_print().

When 6ce4e61f (Trace into a file or an open fd and refactor tracing
code., 2006-09-02) introduced trace.c API, the calls this file makes
to sq_quote_print() were replaced by calls to trace_argv_printf()
that are declared in "cache.h", which this file already includes.
We should have stopped including "quote.h" in that commit, but
forgot to do so.

Signed-off-by: Alexander Kuleshov <kuleshovmail@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoconfigure.ac: check for HMAC_CTX_cleanup
Reuben Hawkins [Thu, 8 Jan 2015 20:00:57 +0000 (12:00 -0800)]
configure.ac: check for HMAC_CTX_cleanup

OpenSSL version 0.9.6b and before defined the function HMAC_cleanup.
Newer versions define HMAC_CTX_cleanup.  Check for HMAC_CTX_cleanup and
fall back to HMAC_cleanup when the newer function is missing.

Signed-off-by: Reuben Hawkins <reubenhwk@gmail.com>
Reviewed-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoconfigure.ac: check for clock_gettime and CLOCK_MONOTONIC
Reuben Hawkins [Thu, 8 Jan 2015 20:00:56 +0000 (12:00 -0800)]
configure.ac: check for clock_gettime and CLOCK_MONOTONIC

Set or clear Makefile variables HAVE_CLOCK_GETTIME and
HAVE_CLOCK_MONOTONIC based upon results of the checks (overriding
default values from config.mak.uname).

CLOCK_MONOTONIC isn't available on RHEL3, but there are still RHEL3
systems being used in production.

Signed-off-by: Reuben Hawkins <reubenhwk@gmail.com>
Reviewed-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoconfigure.ac: check 'tv_nsec' field in 'struct stat'
Reuben Hawkins [Thu, 8 Jan 2015 20:00:55 +0000 (12:00 -0800)]
configure.ac: check 'tv_nsec' field in 'struct stat'

Detect 'tv_nsec' field in 'struct stat' and set Makefile variable
NO_NSEC appropriately.

A side-effect of the above detection is that we also determine
whether 'stat.st_mtimespec' is available, so, as a bonus, set the
Makefile variable USE_ST_TIMESPEC, as well.

Signed-off-by: Reuben Hawkins <reubenhwk@gmail.com>
Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Reviewed-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoshow-branch: line-wrap show-branch usage
Alexander Kuleshov [Thu, 8 Jan 2015 18:08:36 +0000 (00:08 +0600)]
show-branch: line-wrap show-branch usage

Signed-off-by: Alexander Kuleshov <kuleshovmail@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoFourth batch for 2.3 cycle
Junio C Hamano [Wed, 7 Jan 2015 21:12:54 +0000 (13:12 -0800)]
Fourth batch for 2.3 cycle

Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoMerge branch 'maint'
Junio C Hamano [Wed, 7 Jan 2015 21:28:29 +0000 (13:28 -0800)]
Merge branch 'maint'

* maint:
  is_hfs_dotgit: loosen over-eager match of \u{..47}

6 years agoMerge branch 'maint-2.1' into maint
Junio C Hamano [Wed, 7 Jan 2015 21:28:10 +0000 (13:28 -0800)]
Merge branch 'maint-2.1' into maint

* maint-2.1:
  is_hfs_dotgit: loosen over-eager match of \u{..47}

6 years agoMerge branch 'maint-2.0' into maint-2.1
Junio C Hamano [Wed, 7 Jan 2015 21:27:56 +0000 (13:27 -0800)]
Merge branch 'maint-2.0' into maint-2.1

* maint-2.0:
  is_hfs_dotgit: loosen over-eager match of \u{..47}

6 years agoMerge branch 'maint-1.9' into maint-2.0
Junio C Hamano [Wed, 7 Jan 2015 21:27:19 +0000 (13:27 -0800)]
Merge branch 'maint-1.9' into maint-2.0

* maint-1.9:
  is_hfs_dotgit: loosen over-eager match of \u{..47}

6 years agoMerge branch 'maint-1.8.5' into maint-1.9
Junio C Hamano [Wed, 7 Jan 2015 21:27:13 +0000 (13:27 -0800)]
Merge branch 'maint-1.8.5' into maint-1.9

* maint-1.8.5:
  is_hfs_dotgit: loosen over-eager match of \u{..47}

6 years agoMerge branch 'jk/dotgit-case-maint-1.8.5' into maint-1.8.5
Junio C Hamano [Wed, 7 Jan 2015 21:26:35 +0000 (13:26 -0800)]
Merge branch 'jk/dotgit-case-maint-1.8.5' into maint-1.8.5

* jk/dotgit-case-maint-1.8.5:
  is_hfs_dotgit: loosen over-eager match of \u{..47}

6 years agoMerge branch 'bw/maint-0090-awk-tweak'
Junio C Hamano [Wed, 7 Jan 2015 21:10:44 +0000 (13:10 -0800)]
Merge branch 'bw/maint-0090-awk-tweak'

* bw/maint-0090-awk-tweak:
  t0090: tweak awk statement for Solaris /usr/xpg4/bin/awk

6 years agoMerge branch 'jh/pre-push-sample-no-custom-ifs'
Junio C Hamano [Wed, 7 Jan 2015 21:10:40 +0000 (13:10 -0800)]
Merge branch 'jh/pre-push-sample-no-custom-ifs'

The sample pre-push hook used customized IFS=' ' for no good reason.

* jh/pre-push-sample-no-custom-ifs:
  pre-push.sample: remove unnecessary and misleading IFS=' '

6 years agoMerge branch 'tf/prompt-preserve-exit-status'
Junio C Hamano [Wed, 7 Jan 2015 21:09:35 +0000 (13:09 -0800)]
Merge branch 'tf/prompt-preserve-exit-status'

Using the exit status of the last command in the prompt, e.g.
PS1='$(__git_ps1) $? ', did not work well because the helper
function stomped on the exit status.

* tf/prompt-preserve-exit-status:
  git-prompt: preserve value of $? inside shell prompt

6 years agoMerge branch 'sb/dco-indentation-fix'
Junio C Hamano [Wed, 7 Jan 2015 21:09:32 +0000 (13:09 -0800)]
Merge branch 'sb/dco-indentation-fix'

* sb/dco-indentation-fix:
  Documentation/SubmittingPatches: unify whitespace/tabs for the DCO

6 years agoMerge branch 'bb/update-unicode-table'
Junio C Hamano [Wed, 7 Jan 2015 21:09:04 +0000 (13:09 -0800)]
Merge branch 'bb/update-unicode-table'

Simplify the procedure to generate unicode table.

* bb/update-unicode-table:
  update_unicode.sh: delete the command group
  update_unicode.sh: make the output structure visible
  update_unicode.sh: shorten uniset invocation path
  update_unicode.sh: set UNICODE_DIR only once
  update_unicode.sh: simplify output capture

6 years agoMerge branch 'es/squelch-openssl-warnings-on-macosx'
Junio C Hamano [Wed, 7 Jan 2015 21:08:30 +0000 (13:08 -0800)]
Merge branch 'es/squelch-openssl-warnings-on-macosx'

Squelch useless compiler warnings on Mac OS X.

* es/squelch-openssl-warnings-on-macosx:
  git-compat-util: suppress unavoidable Apple-specific deprecation warnings

6 years agoMerge branch 'sb/t5400-remove-unused'
Junio C Hamano [Wed, 7 Jan 2015 21:08:26 +0000 (13:08 -0800)]
Merge branch 'sb/t5400-remove-unused'

* sb/t5400-remove-unused:
  t5400: remove dead code

6 years agoMerge branch 'lh/send-email-hide-x-mailer'
Junio C Hamano [Wed, 7 Jan 2015 21:07:27 +0000 (13:07 -0800)]
Merge branch 'lh/send-email-hide-x-mailer'

"git send-email" normally identifies itself via X-Mailer: header
in the message it sends out.  A new command line flag allows the
user to squelch the header.

* lh/send-email-hide-x-mailer:
  test/send-email: --[no-]xmailer tests
  send-email: add --[no-]xmailer option

6 years agoMerge branch 'rd/send-email-2047-fix'
Junio C Hamano [Wed, 7 Jan 2015 21:06:47 +0000 (13:06 -0800)]
Merge branch 'rd/send-email-2047-fix'

"git send-email" did not handle RFC 2047 encoded headers quite
right.

* rd/send-email-2047-fix:
  send-email: handle adjacent RFC 2047-encoded words properly
  send-email: align RFC 2047 decoding more closely with the spec

6 years agoMerge branch 'pd/completion-filenames-fix'
Junio C Hamano [Wed, 7 Jan 2015 21:06:36 +0000 (13:06 -0800)]
Merge branch 'pd/completion-filenames-fix'

The top-of-the-file instruction for completion scripts (in contrib/)
did not name the files correctly.

* pd/completion-filenames-fix:
  Update documentation occurrences of filename .sh

6 years agoMerge branch 'jk/add-i-read-error'
Junio C Hamano [Wed, 7 Jan 2015 21:05:58 +0000 (13:05 -0800)]
Merge branch 'jk/add-i-read-error'

"git add -i" did not notice when the interactive command input
stream went away and kept asking.

* jk/add-i-read-error:
  add--interactive: leave main loop on read error

6 years agoMerge branch 'jk/approxidate-avoid-y-d-m-over-future-dates'
Junio C Hamano [Wed, 7 Jan 2015 21:01:16 +0000 (13:01 -0800)]
Merge branch 'jk/approxidate-avoid-y-d-m-over-future-dates'

Traditionally we tried to avoid interpreting date strings given by
the user as future dates, e.g. GIT_COMMITTER_DATE=2014-12-10 when
used early November 2014 was taken as "October 12, 2014" because it
is likely that a date in the future, December 10, is a mistake.

Loosen this and do not tiebreak by future-ness of the date when

(1) ISO-like format is used, and
(2) the string can make sense interpreted as both y-m-d and y-d-m.

* jk/approxidate-avoid-y-d-m-over-future-dates:
  approxidate: allow ISO-like dates far in the future
  pass TIME_DATE_NOW to approxidate future-check

6 years agoMerge branch 'br/imap-send-via-libcurl'
Junio C Hamano [Wed, 7 Jan 2015 20:58:05 +0000 (12:58 -0800)]
Merge branch 'br/imap-send-via-libcurl'

Newer libCurl knows how to talk IMAP; "git imap-send" has been
updated to use this instead of a hand-rolled OpenSSL calls.

* br/imap-send-via-libcurl:
  git-imap-send: use libcurl for implementation

6 years agoMerge branch 'br/imap-send-verbosity'
Junio C Hamano [Wed, 7 Jan 2015 20:57:03 +0000 (12:57 -0800)]
Merge branch 'br/imap-send-verbosity'

* br/imap-send-verbosity:
  imap-send: use parse options API to determine verbosity

6 years agoMerge branch 'nd/lockfile-absolute'
Junio C Hamano [Wed, 7 Jan 2015 20:56:01 +0000 (12:56 -0800)]
Merge branch 'nd/lockfile-absolute'

The lockfile API can get confused which file to clean up when the
process moved the $cwd after creating a lockfile.

* nd/lockfile-absolute:
  lockfile.c: store absolute path

6 years agoMerge branch 'jc/merge-bases'
Junio C Hamano [Wed, 7 Jan 2015 20:55:05 +0000 (12:55 -0800)]
Merge branch 'jc/merge-bases'

The get_merge_bases*() API was easy to misuse by careless
copy&paste coders, leaving object flags tainted in the commits that
needed to be traversed.

* jc/merge-bases:
  get_merge_bases(): always clean-up object flags
  bisect: clean flags after checking merge bases

6 years agoMerge branch 'jc/strbuf-add-lines-avoid-sp-ht-sequence'
Junio C Hamano [Wed, 7 Jan 2015 20:49:19 +0000 (12:49 -0800)]
Merge branch 'jc/strbuf-add-lines-avoid-sp-ht-sequence'

The commented output used to blindly add a SP before the payload
line, resulting in "# \t<indented text>\n" when the payload began
with a HT.  Instead, produce "#\t<indented text>\n".

* jc/strbuf-add-lines-avoid-sp-ht-sequence:
  strbuf_add_commented_lines(): avoid SP-HT sequence in commented lines

6 years agoMerge branch 'jc/diff-b-m'
Junio C Hamano [Wed, 7 Jan 2015 20:44:41 +0000 (12:44 -0800)]
Merge branch 'jc/diff-b-m'

Fix long-standing bug in "diff -B -M" output.

* jc/diff-b-m:
  diff -B -M: fix output for "copy and then rewrite" case

6 years agoMerge branch 'jc/clone-borrow'
Junio C Hamano [Wed, 7 Jan 2015 20:42:13 +0000 (12:42 -0800)]
Merge branch 'jc/clone-borrow'

Allow "git clone --reference" to be used more safely.

* jc/clone-borrow:
  clone: --dissociate option to mark that reference is only temporary

6 years agoMerge branch 'jc/checkout-local-track-report'
Junio C Hamano [Wed, 7 Jan 2015 20:40:59 +0000 (12:40 -0800)]
Merge branch 'jc/checkout-local-track-report'

The report from "git checkout" on a branch that builds on another
local branch by setting its branch.*.merge to branch name (not a
full refname) incorrectly said that the upstream is gone.

* jc/checkout-local-track-report:
  checkout: report upstream correctly even with loosely defined branch.*.merge

6 years agogit-completion: add --autostash for 'git rebase'
Matthieu Moy [Wed, 7 Jan 2015 12:40:31 +0000 (13:40 +0100)]
git-completion: add --autostash for 'git rebase'

This option was added in 58794775 (rebase: implement
--[no-]autostash and rebase.autostash, 2013-05-12).

Completion of "--autosquash" has been there, but this was not;
addition of this would require people completing "--autosquash" to
type a bit more than before.

Signed-off-by: Matthieu Moy <Matthieu.Moy@imag.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agocorrect smtp-ssl-cert-path description
Adam Williamson [Tue, 6 Jan 2015 19:01:00 +0000 (11:01 -0800)]
correct smtp-ssl-cert-path description

The git-send-email documentation was never updated to reflect
the change made in 01645b74 to use the SSL library's default
CA trust store rather than /etc/ssl/certs as a hardcoded
default CApath. This corrects that, and also tweaks the rest
of the text a bit to explain more accurately what is required
for a valid CApath / CAfile.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agotest-lib.sh: do tests for color support after changing HOME
Richard Hansen [Tue, 6 Jan 2015 22:57:51 +0000 (17:57 -0500)]
test-lib.sh: do tests for color support after changing HOME

If ncurses needs ~/.terminfo for the current $TERM, then tput will
succeed before changing HOME to $TRASH_DIRECTORY but fail afterward.
Move the tests that determine whether there is color support after
changing HOME so that color=t is set if and only if tput would succeed
when say_color() is run.

Note that color=t is now set after --no-color is processed, so the
condition to set color=t has changed:  it is now set only if
color has not already been set to the empty string by --no-color.

This disables color support for those that need ~/.terminfo for
their TERM, but it's better than filling the screen with:

    tput: unknown terminal "custom-terminal-name-here"

An alternative would be to symlink or copy the user's terminfo
database into $TRASH_DIRECTORY, but this is tricky due to the lack of
a standard name for the terminfo database (for example, instead of a
~/.terminfo directory, NetBSD uses a ~/.terminfo.cdb database file).

Signed-off-by: Richard Hansen <rhansen@bbn.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agotest-lib: use 'test ...' instead of '[ ... ]'
Richard Hansen [Tue, 6 Jan 2015 22:57:50 +0000 (17:57 -0500)]
test-lib: use 'test ...' instead of '[ ... ]'

(see Documentation/CodingGuidelines)

Signed-off-by: Richard Hansen <rhansen@bbn.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agogit-prompt.sh: allow to hide prompt for ignored pwd
Jess Austin [Wed, 7 Jan 2015 01:22:27 +0000 (20:22 -0500)]
git-prompt.sh: allow to hide prompt for ignored pwd

Optionally set __git_ps1 to display nothing when present working
directory is ignored, triggered by the new environment variable
GIT_PS1_HIDE_IF_PWD_IGNORED. This environment variable may be
overridden on any repository by setting bash.hideIfPwdIgnored to
"false". In the absence of GIT_PS1_HIDE_IF_PWD_IGNORED this change
has no effect.

Many people manage e.g. dotfiles in their home directory with git.
This causes the prompt generated by __git_ps1 to refer to that "top
level" repo while working in any descendant directory. That can be
distracting, so this patch helps one shut off that noise.

Signed-off-by: Jess Austin <jess.austin@gmail.com>
Signed-off-by: Richard Hansen <rhansen@bbn.com>
Reviewed-by: Richard Hansen <rhansen@bbn.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agogit-prompt.sh: if pc mode, immediately set PS1 to a plain prompt
Richard Hansen [Wed, 7 Jan 2015 01:22:26 +0000 (20:22 -0500)]
git-prompt.sh: if pc mode, immediately set PS1 to a plain prompt

At the beginning of __git_ps1, right after determining that the
function is running in pc mode, set PS1 to a plain (undecorated)
prompt.  This makes it possible to simply return early without having
to set PS1 if the prompt should not be decorated.

Signed-off-by: Richard Hansen <rhansen@bbn.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoSubmittingPatches: explain rationale for using --notes with format-patch
Eric Sunshine [Tue, 30 Dec 2014 23:30:30 +0000 (18:30 -0500)]
SubmittingPatches: explain rationale for using --notes with format-patch

While here, also change grammatically poor "three dash lines" to
"three-dash line".

Suggested-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agosubtree: fix AsciiDoc list item continuation
Steffen Prohaska [Sun, 4 Jan 2015 10:54:47 +0000 (11:54 +0100)]
subtree: fix AsciiDoc list item continuation

List items must be continued with '+' (see [asciidoc]).

[asciidoc] AsciiDoc user guide 17.7. List Item Continuation
    <http://www.methods.co.nz/asciidoc/userguide.html#X15>

Signed-off-by: Steffen Prohaska <prohaska@zib.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoupdate-server-info: create info/* with mode 0666
Jeff King [Tue, 6 Jan 2015 03:50:49 +0000 (22:50 -0500)]
update-server-info: create info/* with mode 0666

Prior to d38379e (make update-server-info more robust, 2014-09-13),
we used a straight "fopen" to create the info/refs and
objects/info/packs files, which creates the file using mode 0666
(less the default umask).

In d38379e, we switched to creating the file with mkstemp to get a
unique filename. But mkstemp also uses the more restrictive 0600
mode to create the file. This was an unintended side effect that we
did not want, and causes problems when the repository is served by a
different user than the one running update-server-info (it is not
readable by a dumb http server running as `www`, for example).

We can fix this by using git_mkstemp_mode and specifying 0666 to
make sure that the umask is honored.

Note that we could also say "just use core.sharedrepository", as we
do call adjust_shared_perm on the result before renaming it into
place.  But that should not be necessary as long as everybody
involved is using permissive umask to allow HTTP server to read
necessary files.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoimap-send.c: set CURLOPT_USE_SSL to CURLUSESSL_TRY
Kyle J. McKay [Tue, 6 Jan 2015 11:20:37 +0000 (03:20 -0800)]
imap-send.c: set CURLOPT_USE_SSL to CURLUSESSL_TRY

According to the cURL documentation for the CURLOPT_USE_SSL option,
it is only used with plain text protocols that get upgraded to SSL
using the STARTTLS command.

The server.use_ssl variable is only set when we are using a protocol
that is already SSL/TLS (i.e. imaps), so setting CURLOPT_USE_SSL
when the server.use_ssl variable is set has no effect whatsoever.

Instead, set CURLOPT_USE_SSL to CURLUSESSL_TRY when the server.use_ssl
variable is NOT set so that cURL will attempt to upgrade the plain
text connection to SSL/TLS using STARTTLS in that case.

This much more closely matches the behavior of the non-cURL code path.

Signed-off-by: Kyle J. McKay <mackyle@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoimap-send.c: support GIT_CURL_VERBOSE
Kyle J. McKay [Tue, 6 Jan 2015 11:00:55 +0000 (03:00 -0800)]
imap-send.c: support GIT_CURL_VERBOSE

When using git-imap-send to send via cURL, support setting
the GIT_CURL_VERBOSE environment variable to enable cURL's
verbose mode.

The existing http.c code already supports this and does
it by simply checking to see whether or not the environment
variable exists -- it does not examine the value at all.

For consistency, enable CURLOPT_VERBOSE when GIT_CURL_VERBOSE
is set by using the exact same test that http.c does.

Signed-off-by: Kyle J. McKay <mackyle@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agolog.c: fix translation markings
Kyle J. McKay [Tue, 6 Jan 2015 10:34:18 +0000 (02:34 -0800)]
log.c: fix translation markings

The parse_options API expects an array of alternative usage lines
to which it automatically ads the language-appropriate "or" when
displaying.  Each of these options is marked for translation with N_
and then later translated when gettext is called on each element
of the array.

Since the N_ macro just expands to its argument, if two N_-marked
strings appear next to each other without being separated by anything
else such as a comma, the preprocessor will join them into one string.

In that case two separate strings get marked for translation, but at
runtime they have been joined into a single string passed to gettext
which then fails to get translated because the combined string was
never marked for translation.

Fix this by properly separating the two N_ marked strings with
a comma and removing the embedded "\n" and "   or:" that are
properly supplied by the parse_options API.

Signed-off-by: Kyle J. McKay <mackyle@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agot1301: set umask in reflog sharedrepository=group test
Jeff King [Tue, 6 Jan 2015 03:49:43 +0000 (22:49 -0500)]
t1301: set umask in reflog sharedrepository=group test

The t1301 script sets the umask globally before many of the
tests. Most of the tests that care about the umask then set
it explicitly at the start of the test. However, one test
does not, and relies on the 077 umask setting from earlier
tests. This is fragile and can break if another test is
added in between. Let's be more explicit.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agorefs: plug strbuf leak in lock_ref_sha1_basic()
René Scharfe [Wed, 24 Dec 2014 00:18:45 +0000 (01:18 +0100)]
refs: plug strbuf leak in lock_ref_sha1_basic()

Don't just reset, but release the resource held by the local
variable that is about to go out of scope.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agois_hfs_dotgit: loosen over-eager match of \u{..47}
Jeff King [Tue, 23 Dec 2014 08:45:36 +0000 (03:45 -0500)]
is_hfs_dotgit: loosen over-eager match of \u{..47}

Our is_hfs_dotgit function relies on the hackily-implemented
next_hfs_char to give us the next character that an HFS+
filename comparison would look at. It's hacky because it
doesn't implement the full case-folding table of HFS+; it
gives us just enough to see if the path matches ".git".

At the end of next_hfs_char, we use tolower() to convert our
32-bit code point to lowercase. Our tolower() implementation
only takes an 8-bit char, though; it throws away the upper
24 bits. This means we can't have any false negatives for
is_hfs_dotgit. We only care about matching 7-bit ASCII
characters in ".git", and we will correctly process 'G' or
'g'.

However, we _can_ have false positives. Because we throw
away the upper bits, code point \u{0147} (for example) will
look like 'G' and get downcased to 'g'. It's not known
whether a sequence of code points whose truncation ends up
as ".git" is meaningful in any language, but it does not
hurt to be more accurate here. We can just pass out the full
32-bit code point, and compare it manually to the upper and
lowercase characters we care about.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agobisect: add test to check that revs are properly parsed
Christian Couder [Thu, 25 Dec 2014 18:25:33 +0000 (19:25 +0100)]
bisect: add test to check that revs are properly parsed

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agobisect: parse revs before passing them to check_expected_revs()
Christian Couder [Thu, 25 Dec 2014 18:25:32 +0000 (19:25 +0100)]
bisect: parse revs before passing them to check_expected_revs()

When running for example "git bisect bad HEAD" or
"git bisect good master", the parameter passed to
"git bisect (bad|good)" has to be parsed into a
commit hash before checking if it is the expected
commit or not.

We could do that in is_expected_rev() or in
check_expected_revs(), but it is already done in
bisect_state(). Let's just store the hash values
that result from this parsing, and then reuse
them after all the parsing is done.

This way we can also use a for loop over these
values to call bisect_write() on them, instead of
using eval.

Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agocheckout-index: fix --temp relative path mangling
Eric Sunshine [Wed, 24 Dec 2014 09:43:16 +0000 (04:43 -0500)]
checkout-index: fix --temp relative path mangling

checkout-index --temp only properly prints relative paths which are
descendants of the current directory. Paths in ancestor or sibling
directories (or their children) are often printed in mangled form. For
example:

    mkdir a bbb &&
    >file &&
    >bbb/file &&
    git update-index --add file bbb/file &&
    cd a &&
    git checkout-index --temp ../file ../bbb/file

prints:

    .merge_file_ooblek  le
    .merge_file_igloo0  b/file

rather than the correct:

    .merge_file_ooblek  ../file
    .merge_file_igloo0  ../bbb/file

Internally, given the above example, checkout-index prefixes each input
argument with the name of the current directory ("a/", in this case),
and then assumes that it can simply skip forward by strlen("a/") bytes
to recover the original name. This works for files in the current
directory or its descendants, but fails for files in ancestors or
siblings (or their children) due to path normalization.

For instance, given "../file", "a/" is prepended, giving "a/../file".
Path normalization folds out "a/../", resulting in "file". Attempting
to recover the original name by skipping strlen("a/") bytes gives the
incorrect "le" rather than the desired "../file".

Fix this by taking advantage of write_name_quoted_relative() to recover
the original name properly, rather than assuming that it can be
recovered by skipping strlen(prefix) bytes.

As a bonus, this also fixes a bug in which checkout-index --temp
accessed and printed memory beyond the end-of-string. For instance,
within a subdirectory named "subdirectory", and given argument
"../file", prefixing would give "subdirectory/../file", which would
become "file" after normalization. checkout-index would then attempt to
recover the original name by skipping strlen("subdirectory/") bytes of
"file", which placed it well beyond end-of-string. Despite this error,
it often appeared to give the correct result, but only due to an
accident of implementation which left an apparently correct copy of the
path in memory following the normalized value. In particular, handed
"subdirectory/../file", in-place processing by normalize_path_copy_len()
resulted in "file\0rectory/../file". When checkout-index skipped
strlen("subdirectory/") bytes, it ended up back at "../file" and thus
appeared to give the correct answer, despite being past end-of-string.

Reported-by: Russ Cox <rsc@golang.org>
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agot2004: demonstrate broken relative path printing
Eric Sunshine [Wed, 24 Dec 2014 09:43:15 +0000 (04:43 -0500)]
t2004: demonstrate broken relative path printing

checkout-index --temp only properly prints relative paths which are
descendants of the current directory. Paths in ancestor or sibling
directories (or their children) are often printed in mangled form. For
example:

    mkdir a bbb &&
    >file &&
    >bbb/file &&
    git update-index --add file bbb/file &&
    cd a &&
    git checkout-index --temp ../file ../bbb/file

prints:

    .merge_file_ooblek  le
    .merge_file_igloo0  b/file

rather than the correct:

    .merge_file_ooblek  ../file
    .merge_file_igloo0  ../bbb/file

Unfortunately, testing is complicated slightly by relative paths
sometimes _appearing_ to be printed correctly, but this is an accident
of implementation in which a "correct" copy of the string exists in
memory beyond the end of the real string, and that "correct" copy gets
printed. This test takes care to avoid the accidentally "correct"
behavior by testing with a filename longer than the directory name in
which checkout-index is invoked.

Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agot2004: standardize file naming in symlink test
Eric Sunshine [Wed, 24 Dec 2014 09:43:14 +0000 (04:43 -0500)]
t2004: standardize file naming in symlink test

Update "symlink" test to use the common file naming scheme so that its
temporary files can be cleaned up by the "rm -f path*" idiom employed by
other tests in this script.

Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agot2004: drop unnecessary write-tree/read-tree
Eric Sunshine [Wed, 24 Dec 2014 09:43:13 +0000 (04:43 -0500)]
t2004: drop unnecessary write-tree/read-tree

Unlike earlier tests which reference several trees prepared by "setup",
no other tests utilize the tree from the "symlink" test, so there is no
need to write it (or read it back immediately).

Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agot2004: modernize style
Eric Sunshine [Wed, 24 Dec 2014 09:43:12 +0000 (04:43 -0500)]
t2004: modernize style

In particular:

* indent test body
* place test description on same line as test_expect_*
* place closing quote on its own line
* name output file "actual" rather than "out"
* name setup test "setup" rather than "preparation"

Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years agoMerge branch 'for-junio' of git://bogomips.org/git-svn
Junio C Hamano [Mon, 29 Dec 2014 18:15:22 +0000 (10:15 -0800)]
Merge branch 'for-junio' of git://bogomips.org/git-svn

* 'for-junio' of git://bogomips.org/git-svn:
  git-svn: support for git-svn propset