git/git.git
3 years agoobject_array: add and use `object_array_pop()`
Martin Ågren [Fri, 22 Sep 2017 23:34:53 +0000 (01:34 +0200)]
object_array: add and use `object_array_pop()`

In a couple of places, we pop objects off an object array `foo` by
decreasing `foo.nr`. We access `foo.nr` in many places, but most if not
all other times we do so read-only, e.g., as we iterate over the array.
But when we change `foo.nr` behind the array's back, it feels a bit
nasty and looks like it might leak memory.

Leaks happen if the popped element has an allocated `name` or `path`.
At the moment, that is not the case. Still, 1) the object array might
gain more fields that want to be freed, 2) a code path where we pop
might start using names or paths, 3) one of these code paths might be
copied to somewhere where we do, and 4) using a dedicated function for
popping is conceptually cleaner.

Introduce and use `object_array_pop()` instead. Release memory in the
new function. Document that popping an object leaves the associated
elements in limbo.

The converted places were identified by grepping for "\.nr\>" and
looking for "--".

Make the new function return NULL on an empty array. This is consistent
with `pop_commit()` and allows the following:

while ((o = object_array_pop(&foo)) != NULL) {
// do something
}

But as noted above, we don't need to go out of our way to avoid reading
`foo.nr`. This is probably more readable:

while (foo.nr) {
... o = object_array_pop(&foo);
// do something
}

The name of `object_array_pop()` does not quite align with
`add_object_array()`. That is unfortunate. On the other hand, it matches
`object_array_clear()`. Arguably it's `add_...` that is the odd one out,
since it reads like it's used to "add" an "object array". For that
reason, side with `object_array_clear()`.

Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoobject_array: use `object_array_clear()`, not `free()`
Martin Ågren [Fri, 22 Sep 2017 23:34:52 +0000 (01:34 +0200)]
object_array: use `object_array_clear()`, not `free()`

Instead of freeing `foo.objects` for an object array `foo` (sometimes
conditionally), call `object_array_clear(&foo)`. This means we don't
poke as much into the implementation, which is already a good thing, but
also that we release the individual entries as well, thereby fixing at
least one memory-leak (in diff-lib.c).

If someone is holding on to a pointer to an element's `name` or `path`,
that is now a dangling pointer, i.e., we'd be turning an unpleasant
situation into an outright bug. To the best of my understanding no such
long-term pointers are being taken.

The way we handle `study` in builting/reflog.c still looks like it might
leak. That will be addressed in the next commit.

Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoleak_pending: use `object_array_clear()`, not `free()`
Martin Ågren [Fri, 22 Sep 2017 23:34:51 +0000 (01:34 +0200)]
leak_pending: use `object_array_clear()`, not `free()`

Setting `leak_pending = 1` tells `prepare_revision_walk()` not to
release the `pending` array, and makes that the caller's responsibility.
See 4a43d374f (revision: add leak_pending flag, 2011-10-01) and
353f5657a (bisect: use leak_pending flag, 2011-10-01).

Commit 1da1e07c8 (clean up name allocation in prepare_revision_walk,
2014-10-15) fixed a memory leak in `prepare_revision_walk()` by
switching from `free()` to `object_array_clear()`. However, where we use
the `leak_pending`-mechanism, we're still only calling `free()`.

Use `object_array_clear()` instead. Copy some helpful comments from
353f5657a to the other callers that we update to clarify the memory
responsibilities, and to highlight that the commits are not affected
when we clear the array -- it is indeed correct to both tidy up the
commit flags and clear the object array.

Document `leak_pending` in revision.h to help future users get this
right.

Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agocommit: fix memory leak in `reduce_heads()`
Martin Ågren [Fri, 22 Sep 2017 23:34:50 +0000 (01:34 +0200)]
commit: fix memory leak in `reduce_heads()`

We don't free the temporary scratch space we use with
`remove_redundant()`. Free it similar to how we do it in
`get_merge_bases_many_0()`.

Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agobuiltin/commit: fix memory leak in `prepare_index()`
Martin Ågren [Fri, 22 Sep 2017 23:34:49 +0000 (01:34 +0200)]
builtin/commit: fix memory leak in `prepare_index()`

Release `pathspec` and the string list `partial`.

When we clear the string list, make sure we do not free the `util`
pointers. That would result in double-freeing, since we set them up as
`item->util = item` in `list_paths()`.

Initialize the string list early, so that we can always release it. That
introduces some unnecessary overhead in various code paths, but means
there is one and only one way out of the function. If we ever accumulate
more things we need to free, it should be straightforward to do so.

Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoRelNotes: further fixes for 2.14.2 from the master front
Junio C Hamano [Sun, 10 Sep 2017 08:06:09 +0000 (17:06 +0900)]
RelNotes: further fixes for 2.14.2 from the master front

Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoMerge branch 'jt/doc-pack-objects-fix' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:10 +0000 (17:03 +0900)]
Merge branch 'jt/doc-pack-objects-fix' into maint

Doc updates.

* jt/doc-pack-objects-fix:
  Doc: clarify that pack-objects makes packs, plural

3 years agoMerge branch 'jn/vcs-svn-cleanup' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:09 +0000 (17:03 +0900)]
Merge branch 'jn/vcs-svn-cleanup' into maint

Code clean-up.

* jn/vcs-svn-cleanup:
  vcs-svn: move remaining repo_tree functions to fast_export.h
  vcs-svn: remove repo_delete wrapper function
  vcs-svn: remove custom mode constants
  vcs-svn: remove more unused prototypes and declarations

3 years agoMerge branch 'bc/vcs-svn-cleanup' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:08 +0000 (17:03 +0900)]
Merge branch 'bc/vcs-svn-cleanup' into maint

Code clean-up.

* bc/vcs-svn-cleanup:
  vcs-svn: rename repo functions to "svn_repo"
  vcs-svn: remove unused prototypes

3 years agoMerge branch 'jk/doc-the-this' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:07 +0000 (17:03 +0900)]
Merge branch 'jk/doc-the-this' into maint

Doc clean-up.

* jk/doc-the-this:
  doc: fix typo in sendemail.identity

3 years agoMerge branch 'rs/commit-h-single-parent-cleanup' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:07 +0000 (17:03 +0900)]
Merge branch 'rs/commit-h-single-parent-cleanup' into maint

Code clean-up.

* rs/commit-h-single-parent-cleanup:
  commit: remove unused inline function single_parent()

3 years agoMerge branch 'mg/format-ref-doc-fix' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:06 +0000 (17:03 +0900)]
Merge branch 'mg/format-ref-doc-fix' into maint

Doc fix.

* mg/format-ref-doc-fix:
  Documentation/git-for-each-ref: clarify peeling of tags for --format
  Documentation: use proper wording for ref format strings

3 years agoMerge branch 'sb/submodule-parallel-update' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:06 +0000 (17:03 +0900)]
Merge branch 'sb/submodule-parallel-update' into maint

Code clean-up.

* sb/submodule-parallel-update:
  submodule.sh: remove unused variable

3 years agoMerge branch 'hv/t5526-andand-chain-fix' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:05 +0000 (17:03 +0900)]
Merge branch 'hv/t5526-andand-chain-fix' into maint

Test fix.

* hv/t5526-andand-chain-fix:
  t5526: fix some broken && chains

3 years agoMerge branch 'sb/sha1-file-cleanup' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:04 +0000 (17:03 +0900)]
Merge branch 'sb/sha1-file-cleanup' into maint

Code clean-up.

* sb/sha1-file-cleanup:
  sha1_file: make read_info_alternates static

3 years agoMerge branch 'rs/t1002-do-not-use-sum' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:04 +0000 (17:03 +0900)]
Merge branch 'rs/t1002-do-not-use-sum' into maint

Test simplification.

* rs/t1002-do-not-use-sum:
  t1002: stop using sum(1)

3 years agoMerge branch 'ah/doc-empty-string-is-false' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:03 +0000 (17:03 +0900)]
Merge branch 'ah/doc-empty-string-is-false' into maint

Doc update.

* ah/doc-empty-string-is-false:
  doc: clarify "config --bool" behaviour with empty string

3 years agoMerge branch 'rs/merge-microcleanup' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:02 +0000 (17:03 +0900)]
Merge branch 'rs/merge-microcleanup' into maint

Code clean-up.

* rs/merge-microcleanup:
  merge: use skip_prefix()

3 years agoMerge branch 'rs/find-pack-entry-bisection' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:02 +0000 (17:03 +0900)]
Merge branch 'rs/find-pack-entry-bisection' into maint

Code clean-up.

* rs/find-pack-entry-bisection:
  sha1_file: avoid comparison if no packed hash matches the first byte

3 years agoMerge branch 'rs/apply-lose-prefix-length' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:01 +0000 (17:03 +0900)]
Merge branch 'rs/apply-lose-prefix-length' into maint

Code clean-up.

* rs/apply-lose-prefix-length:
  apply: remove prefix_length member from apply_state

3 years agoMerge branch 'rj/add-chmod-error-message' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:03:00 +0000 (17:03 +0900)]
Merge branch 'rj/add-chmod-error-message' into maint

Message fix.

* rj/add-chmod-error-message:
  builtin/add: add detail to a 'cannot chmod' error message

3 years agoMerge branch 'jk/hashcmp-memcmp' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:59 +0000 (17:02 +0900)]
Merge branch 'jk/hashcmp-memcmp' into maint

Code clean-up.

* jk/hashcmp-memcmp:
  hashcmp: use memcmp instead of open-coded loop

3 years agoMerge branch 'rs/t3700-clean-leftover' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:58 +0000 (17:02 +0900)]
Merge branch 'rs/t3700-clean-leftover' into maint

A test fix.

* rs/t3700-clean-leftover:
  t3700: fix broken test under !POSIXPERM

3 years agoMerge branch 'jc/perl-git-comment-typofix' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:57 +0000 (17:02 +0900)]
Merge branch 'jc/perl-git-comment-typofix' into maint

A comment fix.

* jc/perl-git-comment-typofix:
  perl/Git.pm: typofix in a comment

3 years agoMerge branch 'mf/no-dashed-subcommands' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:56 +0000 (17:02 +0900)]
Merge branch 'mf/no-dashed-subcommands' into maint

Code clean-up.

* mf/no-dashed-subcommands:
  scripts: use "git foo" not "git-foo"

3 years agoMerge branch 'ab/ref-filter-no-contains' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:56 +0000 (17:02 +0900)]
Merge branch 'ab/ref-filter-no-contains' into maint

A test fix.

* ab/ref-filter-no-contains:
  tests: don't give unportable ">" to "test" built-in, use -gt

3 years agoMerge branch 'rs/archive-excluded-directory' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:55 +0000 (17:02 +0900)]
Merge branch 'rs/archive-excluded-directory' into maint

"git archive" did not work well with pathspecs and the
export-ignore attribute.

We may want to resurrect the "we don't archive an empty directory"
bonus patch, but I do not mind merging the above early to 'next'
and leave it as a separate follow-up enhancement.
cf. <20170820090629.tumvqwzkromcykjf@sigill.intra.peff.net>

* rs/archive-excluded-directory:
  archive: don't queue excluded directories
  archive: factor out helper functions for handling attributes
  t5001: add tests for export-ignore attributes and exclude pathspecs

3 years agoMerge branch 'mg/killed-merge' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:55 +0000 (17:02 +0900)]
Merge branch 'mg/killed-merge' into maint

Killing "git merge --edit" before the editor returns control left
the repository in a state with MERGE_MSG but without MERGE_HEAD,
which incorrectly tells the subsequent "git commit" that there was
a squash merge in progress.  This has been fixed.

* mg/killed-merge:
  merge: save merge state earlier
  merge: split write_merge_state in two
  merge: clarify call chain
  Documentation/git-merge: explain --continue

3 years agoMerge branch 'tb/apply-with-crlf' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:55 +0000 (17:02 +0900)]
Merge branch 'tb/apply-with-crlf' into maint

"git apply" that is used as a better "patch -p1" failed to apply a
taken from a file with CRLF line endings to a file with CRLF line
endings.  The root cause was because it misused convert_to_git()
that tried to do "safe-crlf" processing by looking at the index
entry at the same path, which is a nonsense---in that mode, "apply"
is not working on the data in (or derived from) the index at all.
This has been fixed.

* tb/apply-with-crlf:
  apply: file commited with CRLF should roundtrip diff and apply
  convert: add SAFE_CRLF_KEEP_CRLF

3 years agoMerge branch 'cc/subprocess-handshake-missing-capabilities' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:55 +0000 (17:02 +0900)]
Merge branch 'cc/subprocess-handshake-missing-capabilities' into maint

When handshake with a subprocess filter notices that the process
asked for an unknown capability, Git did not report what program
the offending subprocess was running.  This has been corrected.

We may want a follow-up fix to tighten the error checking, though.

* cc/subprocess-handshake-missing-capabilities:
  sub-process: print the cmd when a capability is unsupported

3 years agoMerge branch 'as/grep-quiet-no-match-exit-code-fix' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:55 +0000 (17:02 +0900)]
Merge branch 'as/grep-quiet-no-match-exit-code-fix' into maint

"git grep -L" and "git grep --quiet -L" reported different exit
codes; this has been corrected.

* as/grep-quiet-no-match-exit-code-fix:
  git-grep: correct exit code with --quiet and -L

3 years agoMerge branch 'kd/stash-with-bash-4.4' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:54 +0000 (17:02 +0900)]
Merge branch 'kd/stash-with-bash-4.4' into maint

bash 4.4 or newer gave a warning on NUL byte in command
substitution done in "git stash"; this has been squelched.

* kd/stash-with-bash-4.4:
  stash: prevent warning about null bytes in input

3 years agoMerge branch 'rs/win32-syslog-leakfix' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:54 +0000 (17:02 +0900)]
Merge branch 'rs/win32-syslog-leakfix' into maint

Memory leak in an error codepath has been plugged.

* rs/win32-syslog-leakfix:
  win32: plug memory leak on realloc() failure in syslog()

3 years agoMerge branch 'rs/unpack-entry-leakfix' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:53 +0000 (17:02 +0900)]
Merge branch 'rs/unpack-entry-leakfix' into maint

Memory leak in an error codepath has been plugged.

* rs/unpack-entry-leakfix:
  sha1_file: release delta_stack on error in unpack_entry()

3 years agoMerge branch 'rs/fsck-obj-leakfix' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:52 +0000 (17:02 +0900)]
Merge branch 'rs/fsck-obj-leakfix' into maint

Memory leak in an error codepath has been plugged.

* rs/fsck-obj-leakfix:
  fsck: free buffers on error in fsck_obj()

3 years agoMerge branch 'ur/svn-local-zone' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:52 +0000 (17:02 +0900)]
Merge branch 'ur/svn-local-zone' into maint

"git svn" used with "--localtime" option did not compute the tz
offset for the timestamp in question and instead always used the
current time, which has been corrected.

* ur/svn-local-zone:
  git svn fetch: Create correct commit timestamp when using --localtime

3 years agoMerge branch 'pw/am-signoff' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:51 +0000 (17:02 +0900)]
Merge branch 'pw/am-signoff' into maint

"git am -s" has been taught that some input may end with a trailer
block that is not Signed-off-by: and it should refrain from adding
an extra blank line before adding a new sign-off in such a case.

* pw/am-signoff:
  am: fix signoff when other trailers are present

3 years agoMerge branch 'rs/in-obsd-basename-dirname-take-const' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:51 +0000 (17:02 +0900)]
Merge branch 'rs/in-obsd-basename-dirname-take-const' into maint

Portability fix.

* rs/in-obsd-basename-dirname-take-const:
  test-path-utils: handle const parameter of basename and dirname

3 years agoMerge branch 'rs/t4062-obsd' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:51 +0000 (17:02 +0900)]
Merge branch 'rs/t4062-obsd' into maint

Test portability fix.

* rs/t4062-obsd:
  t4062: use less than 256 repetitions in regex

3 years agoMerge branch 'rs/obsd-getcwd-workaround' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:50 +0000 (17:02 +0900)]
Merge branch 'rs/obsd-getcwd-workaround' into maint

Test portability fix for BSDs.

* rs/obsd-getcwd-workaround:
  t0001: skip test with restrictive permissions if getpwd(3) respects them

3 years agoMerge branch 'bw/clone-recursive-quiet' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:49 +0000 (17:02 +0900)]
Merge branch 'bw/clone-recursive-quiet' into maint

"git clone --recurse-submodules --quiet" did not pass the quiet
option down to submodules.

* bw/clone-recursive-quiet:
  clone: teach recursive clones to respect -q

3 years agoMerge branch 'pw/sequence-rerere-autoupdate' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:49 +0000 (17:02 +0900)]
Merge branch 'pw/sequence-rerere-autoupdate' into maint

Commands like "git rebase" accepted the --rerere-autoupdate option
from the command line, but did not always use it.  This has been
fixed.

* pw/sequence-rerere-autoupdate:
  cherry-pick/revert: reject --rerere-autoupdate when continuing
  cherry-pick/revert: remember --rerere-autoupdate
  t3504: use test_commit
  rebase -i: honor --rerere-autoupdate
  rebase: honor --rerere-autoupdate
  am: remember --rerere-autoupdate setting

3 years agoMerge branch 'bw/push-options-recursively-to-submodules' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:49 +0000 (17:02 +0900)]
Merge branch 'bw/push-options-recursively-to-submodules' into maint

"git push --recurse-submodules $there HEAD:$target" was not
propagated down to the submodules, but now it is.

* bw/push-options-recursively-to-submodules:
  submodule--helper: teach push-check to handle HEAD

3 years agoMerge branch 'ma/pager-per-subcommand-action' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:48 +0000 (17:02 +0900)]
Merge branch 'ma/pager-per-subcommand-action' into maint

The "tag.pager" configuration variable was useless for those who
actually create tag objects, as it interfered with the use of an
editor.  A new mechanism has been introduced for commands to enable
pager depending on what operation is being carried out to fix this,
and then "git tag -l" is made to run pager by default.

If this works out OK, I think there are low-hanging fruits in
other commands like "git branch" that outputs long list in one mode
while taking input in another.

* ma/pager-per-subcommand-action:
  git.c: ignore pager.* when launching builtin as dashed external
  tag: change default of `pager.tag` to "on"
  tag: respect `pager.tag` in list-mode only
  t7006: add tests for how git tag paginates
  git.c: provide setup_auto_pager()
  git.c: let builtins opt for handling `pager.foo` themselves
  builtin.h: take over documentation from api-builtin.txt

3 years agoMerge branch 'jk/rev-list-empty-input' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:48 +0000 (17:02 +0900)]
Merge branch 'jk/rev-list-empty-input' into maint

"git log --tag=no-such-tag" showed log starting from HEAD, which
has been fixed---it now shows nothing.

* jk/rev-list-empty-input:
  revision: do not fallback to default when rev_input_given is set
  rev-list: don't show usage when we see empty ref patterns
  revision: add rev_input_given flag
  t6018: flesh out empty input/output rev-list tests

3 years agoMerge branch 'st/lib-gpg-kill-stray-agent' into maint
Junio C Hamano [Sun, 10 Sep 2017 08:02:48 +0000 (17:02 +0900)]
Merge branch 'st/lib-gpg-kill-stray-agent' into maint

Some versions of GnuPG fails to kill gpg-agent it auto-spawned
and such a left-over agent can interfere with a test.  Work it
around by attempting to kill one before starting a new test.

* st/lib-gpg-kill-stray-agent:
  t: lib-gpg: flush gpg agent on startup

3 years agoPrepare for 2.14.2
Junio C Hamano [Wed, 23 Aug 2017 21:36:03 +0000 (14:36 -0700)]
Prepare for 2.14.2

Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoMerge branch 'jt/t1450-fsck-corrupt-packfile' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:52 +0000 (14:33 -0700)]
Merge branch 'jt/t1450-fsck-corrupt-packfile' into maint

A test update.

* jt/t1450-fsck-corrupt-packfile:
  tests: ensure fsck fails on corrupt packfiles

3 years agoMerge branch 'jb/t8008-cleanup' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:52 +0000 (14:33 -0700)]
Merge branch 'jb/t8008-cleanup' into maint

Code clean-up.

* jb/t8008-cleanup:
  t8008: rely on rev-parse'd HEAD instead of sha1 value

3 years agoMerge branch 'jt/subprocess-handshake' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:52 +0000 (14:33 -0700)]
Merge branch 'jt/subprocess-handshake' into maint

Code cleanup.

* jt/subprocess-handshake:
  sub-process: refactor handshake to common function
  Documentation: migrate sub-process docs to header
  convert: add "status=delayed" to filter process protocol
  convert: refactor capabilities negotiation
  convert: move multiple file filter error handling to separate function
  convert: put the flags field before the flag itself for consistent style
  t0021: write "OUT <size>" only on success
  t0021: make debug log file name configurable
  t0021: keep filter log files on comparison

3 years agoMerge branch 'dc/fmt-merge-msg-microcleanup' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:52 +0000 (14:33 -0700)]
Merge branch 'dc/fmt-merge-msg-microcleanup' into maint

Code cleanup.

* dc/fmt-merge-msg-microcleanup:
  fmt-merge-msg: fix coding style

3 years agoMerge branch 'ah/doc-wserrorhighlight' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:51 +0000 (14:33 -0700)]
Merge branch 'ah/doc-wserrorhighlight' into maint

Doc update.

* ah/doc-wserrorhighlight:
  doc: add missing values "none" and "default" for diff.wsErrorHighlight

3 years agoMerge branch 'cc/ref-is-hidden-microcleanup' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:50 +0000 (14:33 -0700)]
Merge branch 'cc/ref-is-hidden-microcleanup' into maint

Code cleanup.

* cc/ref-is-hidden-microcleanup:
  refs: use skip_prefix() in ref_is_hidden()

3 years agoMerge branch 'js/run-process-parallel-api-fix' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:49 +0000 (14:33 -0700)]
Merge branch 'js/run-process-parallel-api-fix' into maint

API fix.

* js/run-process-parallel-api-fix:
  run_processes_parallel: change confusing task_cb convention

3 years agoMerge branch 'rs/pack-objects-pbase-cleanup' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:48 +0000 (14:33 -0700)]
Merge branch 'rs/pack-objects-pbase-cleanup' into maint

Code clean-up.

* rs/pack-objects-pbase-cleanup:
  pack-objects: remove unnecessary NULL check

3 years agoMerge branch 'jt/fsck-code-cleanup' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:48 +0000 (14:33 -0700)]
Merge branch 'jt/fsck-code-cleanup' into maint

Code clean-up.

* jt/fsck-code-cleanup:
  fsck: cleanup unused variable
  object: remove "used" field from struct object
  fsck: remove redundant parse_tree() invocation

3 years agoMerge branch 'rs/stat-data-unaligned-reads-fix' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:47 +0000 (14:33 -0700)]
Merge branch 'rs/stat-data-unaligned-reads-fix' into maint

Code clean-up.

* rs/stat-data-unaligned-reads-fix:
  dir: support platforms that require aligned reads

3 years agoMerge branch 'rs/move-array' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:46 +0000 (14:33 -0700)]
Merge branch 'rs/move-array' into maint

Code clean-up.

* rs/move-array:
  ls-files: don't try to prune an empty index
  apply: use COPY_ARRAY and MOVE_ARRAY in update_image()
  use MOVE_ARRAY
  add MOVE_ARRAY

3 years agoMerge branch 'rs/bswap-ubsan-fix' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:46 +0000 (14:33 -0700)]
Merge branch 'rs/bswap-ubsan-fix' into maint

Code clean-up.

* rs/bswap-ubsan-fix:
  bswap: convert get_be16, get_be32 and put_be32 to inline functions
  bswap: convert to unsigned before shifting in get_be32

3 years agoMerge branch 'dl/credential-cache-socket-in-xdg-cache' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:45 +0000 (14:33 -0700)]
Merge branch 'dl/credential-cache-socket-in-xdg-cache' into maint

A recently added test for the "credential-cache" helper revealed
that EOF detection done around the time the connection to the cache
daemon is torn down were flaky.  This was fixed by reacting to
ECONNRESET and behaving as if we got an EOF.

* dl/credential-cache-socket-in-xdg-cache:
  credential-cache: interpret an ECONNRESET as an EOF

3 years agoMerge branch 'hb/gitweb-project-list' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:44 +0000 (14:33 -0700)]
Merge branch 'hb/gitweb-project-list' into maint

When a directory is not readable, "gitweb" fails to build the
project list.  Work this around by skipping such a directory.

It might end up hiding a problem under the rug and a better
solution might be to loudly complain to the administrator pointing
out the problematic directory, but this will at least make it
"work".

* hb/gitweb-project-list:
  gitweb: skip unreadable subdirectories

3 years agoMerge branch 'ks/commit-abort-on-empty-message-fix' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:44 +0000 (14:33 -0700)]
Merge branch 'ks/commit-abort-on-empty-message-fix' into maint

"git commit" when seeing an totally empty message said "you did not
edit the message", which is clearly wrong.  The message has been
corrected.

* ks/commit-abort-on-empty-message-fix:
  commit: check for empty message before the check for untouched template

3 years agoMerge branch 'jk/reflog-walk' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:43 +0000 (14:33 -0700)]
Merge branch 'jk/reflog-walk' into maint

Numerous bugs in walking of reflogs via "log -g" and friends have
been fixed.

* jk/reflog-walk:
  reflog-walk: apply --since/--until to reflog dates
  reflog-walk: stop using fake parents
  rev-list: check reflog_info before showing usage
  get_revision_1(): replace do-while with an early return
  log: do not free parents when walking reflog
  log: clarify comment about reflog cycles
  revision: disallow reflog walking with revs->limited
  t1414: document some reflog-walk oddities

3 years agoMerge branch 'jc/http-sslkey-and-ssl-cert-are-paths' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:43 +0000 (14:33 -0700)]
Merge branch 'jc/http-sslkey-and-ssl-cert-are-paths' into maint

The http.{sslkey,sslCert} configuration variables are to be
interpreted as a pathname that honors "~[username]/" prefix, but
weren't, which has been fixed.

* jc/http-sslkey-and-ssl-cert-are-paths:
  http.c: http.sslcert and http.sslkey are both pathnames

3 years agoMerge branch 'jk/ref-filter-colors' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:42 +0000 (14:33 -0700)]
Merge branch 'jk/ref-filter-colors' into maint

"%C(color name)" in the pretty print format always produced ANSI
color escape codes, which was an early design mistake.  They now
honor the configuration (e.g. "color.ui = never") and also tty-ness
of the output medium.

* jk/ref-filter-colors:
  ref-filter: consult want_color() before emitting colors
  pretty: respect color settings for %C placeholders
  rev-list: pass diffopt->use_colors through to pretty-print
  for-each-ref: load config earlier
  color: check color.ui in git_default_config()
  ref-filter: pass ref_format struct to atom parsers
  ref-filter: factor out the parsing of sorting atoms
  ref-filter: make parse_ref_filter_atom a private function
  ref-filter: provide a function for parsing sort options
  ref-filter: move need_color_reset_at_eol into ref_format
  ref-filter: abstract ref format into its own struct
  ref-filter: simplify automatic color reset
  t: use test_decode_color rather than literal ANSI codes
  docs/for-each-ref: update pointer to color syntax
  check return value of verify_ref_format()

3 years agoMerge branch 'js/git-gui-msgfmt-on-windows' into maint
Junio C Hamano [Wed, 23 Aug 2017 21:33:42 +0000 (14:33 -0700)]
Merge branch 'js/git-gui-msgfmt-on-windows' into maint

Because recent Git for Windows do come with a real msgfmt, the
build procedure for git-gui has been updated to use it instead of a
hand-rolled substitute.

* js/git-gui-msgfmt-on-windows:
  git-gui (MinGW): make use of MSys2's msgfmt
  git gui: allow for a long recentrepo list
  git gui: de-dup selected repo from recentrepo history
  git gui: cope with duplicates in _get_recentrepo
  git-gui: remove duplicate entries from .gitconfig's gui.recentrepo

3 years agovcs-svn: move remaining repo_tree functions to fast_export.h
Jonathan Nieder [Wed, 23 Aug 2017 00:04:47 +0000 (17:04 -0700)]
vcs-svn: move remaining repo_tree functions to fast_export.h

These used to be for manipulating the in-memory repo_tree structure,
but nowadays they are convenience wrappers to handle a few git-vs-svn
mismatches:

 1. Git does not track empty directories but Subversion does.  When
    looking up a path in git that Subversion thinks exists and finding
    nothing, we can safely assume that the path represents a
    directory.  This is needed when a later Subversion revision
    modifies that directory.

 2. Subversion allows deleting a file by copying.  In Git fast-import
    we have to handle that more explicitly as a deletion.

These are details of the tool's interaction with git fast-import.
Move them to fast_export.c, where other such details are handled.

This way the function names do not start with a repo_ prefix that
would clash with the repository object introduced in
v2.14.0-rc0~38^2~16 (repository: introduce the repository object,
2017-06-22) or an svn_ prefix that would clash with libsvn (in case
someone wants to link this code with libsvn some day).

Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agovcs-svn: remove repo_delete wrapper function
Jonathan Nieder [Wed, 23 Aug 2017 00:02:15 +0000 (17:02 -0700)]
vcs-svn: remove repo_delete wrapper function

Since v1.7.10-rc0~118^2~4^2~4^2~3 (vcs-svn: pass paths through to
fast-import, 2010-12-13) this is an alias for fast_export_delete.
Remove the unnecessary layer of indirection.

No functional change intended.

Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agovcs-svn: remove custom mode constants
Jonathan Nieder [Wed, 23 Aug 2017 00:01:34 +0000 (17:01 -0700)]
vcs-svn: remove custom mode constants

In the rest of Git, these modes are spelled as S_IFDIR,
S_IFREG | 0644, S_IFREG | 0755, and S_IFLNK.  Use the same constants
in svn-fe for simplicity and consistency.

No functional change intended.

Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agovcs-svn: remove more unused prototypes and declarations
Jonathan Nieder [Wed, 23 Aug 2017 00:00:57 +0000 (17:00 -0700)]
vcs-svn: remove more unused prototypes and declarations

I forgot to remove these in v1.7.10-rc0~118^2~4^2~5^2~4 (vcs-svn:
eliminate repo_tree structure, 2010-12-10).

This finishes what was started in commit 36f63b50 (vcs-svn: remove
unused prototypes, 2017-08-21).

Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoDoc: clarify that pack-objects makes packs, plural
Jonathan Tan [Wed, 23 Aug 2017 00:40:10 +0000 (17:40 -0700)]
Doc: clarify that pack-objects makes packs, plural

The documentation for pack-objects describes that it creates "a packed
archive of objects", which is confusing because it may create multiple
packs if --max-pack-size is set. Update the documentation to clarify
this, and explaining in which cases such a feature would be useful.

Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agomerge: save merge state earlier
Michael J Gruber [Wed, 23 Aug 2017 12:10:45 +0000 (14:10 +0200)]
merge: save merge state earlier

If the `git merge` process is killed while waiting for the editor to
finish, the merge state is lost but the prepared merge msg and tree is kept.
So, a subsequent `git commit` creates a squashed merge even when the
user asked for proper merge commit originally.

Demonstrate the problem with a test crafted after the in t7502. The test
requires EXECKEEPSPID (thus does not run under MINGW).

Save the merge state earlier (in the non-squash case) so that it does
not get lost. This makes the test pass.

Reported-by: hIpPy <hippy2981@gmail.com>
Signed-off-by: Michael J Gruber <git@grubix.eu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agomerge: split write_merge_state in two
Michael J Gruber [Wed, 23 Aug 2017 12:10:44 +0000 (14:10 +0200)]
merge: split write_merge_state in two

write_merge_state() writes out the merge heads, mode, and msg. But we
may want to write out heads, mode without the msg. So, split out heads
(+mode) into a separate function write_merge_heads() that is called by
write_merge_state().

No funtional change so far, except when these non-atomic writes are
interrupted: we write heads-mode-msg now when we used to write
heads-msg-mode.

Signed-off-by: Michael J Gruber <git@grubix.eu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agomerge: clarify call chain
Michael J Gruber [Wed, 23 Aug 2017 12:10:43 +0000 (14:10 +0200)]
merge: clarify call chain

prepare_to_commit() cannot be reached in the non-squash case:
It is called by merge_trivial() and finish_automerge() only, but the
calls to the latter are somewhat hard to track:

If option_commit is not set, the code in cmd_merge() uses a fake
conflict return code (ret=1) to avoid writing the tree, which also
avoids setting automerge_was_ok (just as in the proper ret==1 case), so
that finish_automerge() is not called.

To ensure that no code change breaks that assumption, safe-guard
prepare_to_commit() by a BUG() statement.

Suggested-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Michael J Gruber <git@grubix.eu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoDocumentation/git-merge: explain --continue
Michael J Gruber [Mon, 21 Aug 2017 12:53:14 +0000 (14:53 +0200)]
Documentation/git-merge: explain --continue

Currently, 'git merge --continue' is mentioned but not explained.

Explain it.

Signed-off-by: Michael J Gruber <git@grubix.eu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agovcs-svn: rename repo functions to "svn_repo"
brian m. carlson [Mon, 21 Aug 2017 00:00:18 +0000 (00:00 +0000)]
vcs-svn: rename repo functions to "svn_repo"

There were several functions in the Subversion code that started with
"repo_".  This namespace is also used by the Git struct repository code.
Rename these functions to start with "svn_repo" to avoid any future
conflicts.

Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agovcs-svn: remove unused prototypes
brian m. carlson [Mon, 21 Aug 2017 00:00:17 +0000 (00:00 +0000)]
vcs-svn: remove unused prototypes

The Subversion code had prototypes for several functions which were not
ever defined or used.  These functions all had names starting with
"repo_", some of which conflict with those in repository.h.  To avoid
the conflict, remove those unused prototypes.

Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agodoc: fix typo in sendemail.identity
Jeff King [Sun, 20 Aug 2017 09:46:57 +0000 (05:46 -0400)]
doc: fix typo in sendemail.identity

Saying "the this" is an obvious typo. But while we're here,
let's polish the English on the second half of the sentence,
too.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoapply: file commited with CRLF should roundtrip diff and apply
Torsten Bögershausen [Sat, 19 Aug 2017 11:28:01 +0000 (13:28 +0200)]
apply: file commited with CRLF should roundtrip diff and apply

When a file had been commited with CRLF but now .gitattributes say
"* text=auto" (or core.autocrlf is true), the following does not
roundtrip, `git apply` fails:

    printf "Added line\r\n" >>file &&
    git diff >patch &&
    git checkout -- . &&
    git apply patch

Before applying the patch, the file from working tree is converted
into the index format (clean filter, CRLF conversion, ...).  Here,
when commited with CRLF, the line endings should not be converted.

Note that `git apply --index` or `git apply --cache` doesn't call
convert_to_git() because the source material is already in index
format.

Analyze the patch if there is a) any context line with CRLF, or b)
if any line with CRLF is to be removed.  In this case the patch file
`patch` has mixed line endings, for a) it looks like this:

    diff --git a/one b/one
    index 533790e..c30dea8 100644
    --- a/one
    +++ b/one
    @@ -1 +1,2 @@
     a\r
    +b\r

And for b) it looks like this:

    diff --git a/one b/one
    index 533790e..485540d 100644
    --- a/one
    +++ b/one
    @@ -1 +1 @@
    -a\r
    +b\r

If `git apply` detects that the patch itself has CRLF, (look at the
line " a\r" or "-a\r" above), the new flag crlf_in_old is set in
"struct patch" and two things will happen:

    - read_old_data() will not convert CRLF into LF by calling
      convert_to_git(..., SAFE_CRLF_KEEP_CRLF);
    - The WS_CR_AT_EOL bit is set in the "white space rule",
      CRLF are no longer treated as white space.

While at there, make it clear that read_old_data() in apply.c knows
what it wants convert_to_git() to do with respect to CRLF.  In fact,
this codepath is about applying a patch to a file in the filesystem,
which may not exist in the index, or may exist but may not match
what is recorded in the index, or in the extreme case, we may not
even be in a Git repository.  If convert_to_git() peeked at the
index while doing its work, it *would* be a bug.

Pass NULL instead of &the_index to convert_to_git() to make sure we
catch future bugs to clarify this.

Update the test in t4124: split one test case into 3:

    - Detect the " a\r" line in the patch
    - Detect the "-a\r" line in the patch
    - Use LF in repo and CLRF in the worktree.

Reported-by: Anthony Sottile <asottile@umich.edu>
Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agocommit: remove unused inline function single_parent()
René Scharfe [Sat, 19 Aug 2017 08:12:45 +0000 (10:12 +0200)]
commit: remove unused inline function single_parent()

53b2c823f6 (revision walker: mini clean-up) added the function in 2007,
but it was never used, so we should be able to get rid of it now.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoarchive: don't queue excluded directories
René Scharfe [Sat, 19 Aug 2017 05:32:37 +0000 (07:32 +0200)]
archive: don't queue excluded directories

Reject directories with the attribute export-ignore already while
queuing them.  This prevents read_tree_recursive() from descending into
them and this avoids write_archive_entry() rejecting them later on,
which queue_or_write_archive_entry() is not prepared for.

Borrow the existing strbuf to build the full path to avoid string
copies and extra allocations; just make sure we restore the original
value before moving on.

Keep checking any other attributes in write_archive_entry() as before,
but avoid checking them twice.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoarchive: factor out helper functions for handling attributes
René Scharfe [Sat, 19 Aug 2017 05:29:43 +0000 (07:29 +0200)]
archive: factor out helper functions for handling attributes

Add helpers for accessing attributes that encapsulate the details of how
to retrieve their values.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agot5001: add tests for export-ignore attributes and exclude pathspecs
René Scharfe [Sat, 19 Aug 2017 05:28:54 +0000 (07:28 +0200)]
t5001: add tests for export-ignore attributes and exclude pathspecs

Demonstrate mishandling of the attribute export-ignore by git archive
when used together with pathspecs.  Wildcard pathspecs can even cause it
to abort.  And a directory excluded without a wildcard is still included
as an empty folder in the archive.

Test-case-by: David Adam <zanchey@ucc.gu.uwa.edu.au>
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoDocumentation/git-for-each-ref: clarify peeling of tags for --format
Michael J Gruber [Fri, 18 Aug 2017 14:51:23 +0000 (16:51 +0200)]
Documentation/git-for-each-ref: clarify peeling of tags for --format

`*` in format strings means peeling of tag objects so that object field
names refer to the object that the tag object points at, instead of the
tag object itself.

Currently, this is documented using grammar that is clearly inspired by
classical latin, though missing more than an article in order to be
classical english.

Try and straighten that explanation out a bit.

Signed-off-by: Michael J Gruber <git@grubix.eu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoDocumentation: use proper wording for ref format strings
Michael J Gruber [Fri, 18 Aug 2017 14:51:22 +0000 (16:51 +0200)]
Documentation: use proper wording for ref format strings

Various commands list refs and allow to use a format string for the
output that interpolates from the ref as well as the object it points
at (for-each-ref; branch and tag in list mode).

Currently, the documentation talks about interpolating from the object.
This is confusing because a ref points to an object but not vice versa,
so the object cannot possible know %(refname), for example. Thus, this is
wrong independent of refs being objects (one day, maybe) or not.

Change the wording to make this clearer (and distinguish it from formats
for the log family).

Signed-off-by: Michael J Gruber <git@grubix.eu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agogit-grep: correct exit code with --quiet and -L
Anthony Sottile [Fri, 18 Aug 2017 01:38:51 +0000 (18:38 -0700)]
git-grep: correct exit code with --quiet and -L

The handling of `status_only` no longer interferes with the handling of
`unmatch_name_only`.  `--quiet` no longer affects the exit code when using
`-L`/`--files-without-match`.

Signed-off-by: Anthony Sottile <asottile@umich.edu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agot5526: fix some broken && chains
Heiko Voigt [Thu, 17 Aug 2017 10:36:13 +0000 (12:36 +0200)]
t5526: fix some broken && chains

Signed-off-by: Heiko Voigt <hvoigt@hvoigt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agosubmodule.sh: remove unused variable
Stefan Beller [Wed, 16 Aug 2017 22:50:42 +0000 (15:50 -0700)]
submodule.sh: remove unused variable

This could have been part of 48308681b0 (git submodule update: have a
dedicated helper for cloning, 2016-02-29).

Signed-off-by: Stefan Beller <sbeller@google.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoconvert: add SAFE_CRLF_KEEP_CRLF
Torsten Bögershausen [Sun, 13 Aug 2017 08:51:04 +0000 (10:51 +0200)]
convert: add SAFE_CRLF_KEEP_CRLF

When convert_to_git() is called, the caller may want to keep CRLF to
be kept as CRLF (and not converted into LF).

This will be used in the next commit, when apply works with files
that have CRLF and patches are applied onto these files.

Add the new value "SAFE_CRLF_KEEP_CRLF" to safe_crlf.

Prepare convert_to_git() to be able to run the clean filter, skip
the CRLF conversion and run the ident filter.

Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agosub-process: print the cmd when a capability is unsupported
Christian Couder [Wed, 16 Aug 2017 12:40:36 +0000 (14:40 +0200)]
sub-process: print the cmd when a capability is unsupported

In handshake_capabilities() we use warning() when a capability
is not supported, so the exit code of the function is 0 and no
further error is shown. This is a problem because the warning
message doesn't tell us which subprocess cmd failed.

On the contrary if we cannot write a packet from this function,
we use error() and then subprocess_start() outputs:

    initialization for subprocess '<cmd>' failed

so we can know which subprocess cmd failed.

Let's improve the warning() message, so that we can know which
subprocess cmd failed.

Helped-by: Lars Schneider <larsxschneider@gmail.com>
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agosha1_file: make read_info_alternates static
Stefan Beller [Tue, 15 Aug 2017 20:13:19 +0000 (13:13 -0700)]
sha1_file: make read_info_alternates static

read_info_alternates is not used from outside, so let's make it static.

We have to declare the function before link_alt_odb_entry instead of
moving the code around, link_alt_odb_entry calls read_info_alternates,
which in turn calls link_alt_odb_entry.

Signed-off-by: Stefan Beller <sbeller@google.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agot1002: stop using sum(1)
René Scharfe [Mon, 14 Aug 2017 20:16:34 +0000 (22:16 +0200)]
t1002: stop using sum(1)

sum(1) is a command for calculating checksums of the contents of files.
It was part of early editions of Unix ("Research Unix", 1972/1973, [1]).
cksum(1) appeared in 4.4BSD (1993) as a replacement [2], and became part
of POSIX.1-2008 [3].  OpenBSD 5.6 (2014) removed sum(1).

We only use sum(1) in t1002 to check for changes in three files.  On
MinGW we use md5sum(1) instead.  We could switch to the standard command
cksum(1) for all platforms; MinGW comes with GNU coreutils now, which
provides sum(1), cksum(1) and md5sum(1).  Use our standard method for
checking for file changes instead: test_cmp.

It's more convenient because it shows differences nicely, it's faster on
MinGW because we have a special implementation there based only on
shell-internal commands, it's simpler as it allows us to avoid stripping
out unnecessary entries from the checksum file using grep(1), and it's
more consistent with the rest of the test suite.

We already compare changed files with their expected new contents using
diff(1), so we don't need to check with "test_must_fail test_cmp" if
they differ from their original state.  A later patch could convert the
direct diff(1) calls to test_cmp as well.

With all sum(1) calls gone, remove the MinGW-specific implementation
from test-lib.sh as well.

[1] http://minnie.tuhs.org/cgi-bin/utree.pl?file=V3/man/man1/sum.1
[2] http://minnie.tuhs.org/cgi-bin/utree.pl?file=4.4BSD/usr/share/man/cat1/cksum.0
[3] http://pubs.opengroup.org/onlinepubs/9699919799/utilities/cksum.html

Signed-off-by: René Scharfe <l.s.r@web.de>
Reviewed-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agodoc: clarify "config --bool" behaviour with empty string
Andreas Heiduk [Mon, 14 Aug 2017 22:12:18 +0000 (00:12 +0200)]
doc: clarify "config --bool" behaviour with empty string

`git config --bool xxx.yyy` returns `true` for `[xxx]yyy` but
`false` for `[xxx]yyy=` or `[xxx]yyy=""`.  This is tested in
t1300-repo-config.sh since 09bc098c2.

Signed-off-by: Andreas Heiduk <asheiduk@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agostash: prevent warning about null bytes in input
Kevin Daudt [Mon, 14 Aug 2017 21:43:33 +0000 (23:43 +0200)]
stash: prevent warning about null bytes in input

The `no_changes` function calls the `untracked_files` function through
command substitution. `untracked_files` will return null bytes because it
runs ls-files with the '-z' option.

Bash since version 4.4 warns about these null bytes. As they are not
required for the test that is being done, make sure `untracked_files`
does not output null bytes when not required.

This is achieved by adding a parameter to the `untracked_files` function to
specify wither `-z` should be passed to ls-files or not.

This warning is triggered when running git stash save -u resulting in
two warnings:

    git-stash: line 43: warning: command substitution: ignored null byte
    in input

Signed-off-by: Kevin Daudt <me@ikke.info>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agosha1_file: release delta_stack on error in unpack_entry()
René Scharfe [Thu, 10 Aug 2017 09:42:21 +0000 (11:42 +0200)]
sha1_file: release delta_stack on error in unpack_entry()

When unpack_entry() encounters a broken packed object, it returns early.
It adjusts the reference count of the pack window, but leaks the buffer
for a big delta stack in case the small automatic one was not enough.
Jump to the cleanup code at end instead, which takes care of that.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agofsck: free buffers on error in fsck_obj()
René Scharfe [Thu, 10 Aug 2017 09:42:10 +0000 (11:42 +0200)]
fsck: free buffers on error in fsck_obj()

Move the code for releasing tree buffers and commit buffers in
fsck_obj() to the end of the function and make sure it's executed no
matter of an error is encountered or not.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agowin32: plug memory leak on realloc() failure in syslog()
René Scharfe [Thu, 10 Aug 2017 10:23:45 +0000 (12:23 +0200)]
win32: plug memory leak on realloc() failure in syslog()

If realloc() fails then the original buffer is still valid.  Free it
before exiting the function.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agomerge: use skip_prefix()
René Scharfe [Thu, 10 Aug 2017 16:47:55 +0000 (18:47 +0200)]
merge: use skip_prefix()

Get rid of a magic string length constant by using skip_prefix() instead
of starts_with().

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agohashcmp: use memcmp instead of open-coded loop
Jeff King [Wed, 9 Aug 2017 10:16:45 +0000 (06:16 -0400)]
hashcmp: use memcmp instead of open-coded loop

In 1a812f3a70 (hashcmp(): inline memcmp() by hand to
optimize, 2011-04-28), it was reported that an open-coded
loop outperformed memcmp() for comparing sha1s.

Discussion[1] a few years later in 2013 showed that this
depends on your libc's version of memcmp(). In particular,
glibc 2.13 optimized their memcmp around 2011. Here are
current timings with glibc 2.24 (best-of-five, on
linux.git):

  [before this patch, open-coded]
  $ time git rev-list --objects --all
  real 0m35.357s
  user 0m35.016s
  sys 0m0.340s

  [after this patch, memcmp]
  real 0m32.930s
  user 0m32.630s
  sys 0m0.300s

Now that we've had 6 years for that version of glibc to
make its way onto people's machines, it's worth revisiting
our benchmarks and switching to memcmp().

It may be that there are other non-glibc systems where
memcmp() isn't as well optimized. But since our single data
point in favor of open-coding was on a now-ancient glibc, we
should probably assume the system memcmp is good unless
proven otherwise. We may end up with a SLOW_MEMCMP Makefile
knob, but we can hold off on that until we actually find
such a system in practice.

[1] https://public-inbox.org/git/20130318073229.GA5551@sigill.intra.peff.net/

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years agoapply: remove prefix_length member from apply_state
René Scharfe [Wed, 9 Aug 2017 15:54:46 +0000 (17:54 +0200)]
apply: remove prefix_length member from apply_state

Use a NULL-and-NUL check to see if we have a prefix and consistently use
C string functions on it instead of storing its length in a member of
struct apply_state.  This avoids strlen() calls and simplifies the code.

Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>