git/git.git
2 years agopath.h: make REPO_GIT_PATH_FUNC repository agnostic
Stefan Beller [Sat, 15 Dec 2018 00:09:41 +0000 (16:09 -0800)]
path.h: make REPO_GIT_PATH_FUNC repository agnostic

git_pathdup uses the_repository internally, but the macro
REPO_GIT_PATH_FUNC is specifically made for arbitrary repositories.
Switch to repo_git_path which works on arbitrary repositories.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit: prepare free_commit_buffer and release_commit_memory for any repo
Stefan Beller [Sat, 15 Dec 2018 00:09:40 +0000 (16:09 -0800)]
commit: prepare free_commit_buffer and release_commit_memory for any repo

Pass the object pool to free_commit_buffer and release_commit_memory,
such that we can eliminate access to 'the_repository'.

Also remove the TODO in release_commit_memory, as commit->util was
removed in 9d2c97016f (commit.h: delete 'util' field in struct commit,
2018-05-19)

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit-graph: convert remaining functions to handle any repo
Stefan Beller [Sat, 15 Dec 2018 00:09:39 +0000 (16:09 -0800)]
commit-graph: convert remaining functions to handle any repo

Convert all functions to handle arbitrary repositories in commit-graph.c
that are used by functions taking a repository argument already.

Notable exclusion is write_commit_graph and its local functions as that
only works on the_repository.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agosubmodule: don't add submodule as odb for push
Stefan Beller [Sat, 15 Dec 2018 00:09:38 +0000 (16:09 -0800)]
submodule: don't add submodule as odb for push

In push_submodule(), because we do not actually need access to objects
in the submodule, do not invoke add_submodule_odb().
(for_each_remote_ref_submodule() does not require access to those
objects, and the actual push is done by spawning another process,
which handles object access by itself.)

This code of push_submodule() is exercised in t5531 and continues
to work, showing that the submodule odbc is not needed.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agosubmodule: use submodule repos for object lookup
Stefan Beller [Sat, 15 Dec 2018 00:09:37 +0000 (16:09 -0800)]
submodule: use submodule repos for object lookup

This converts the 'show_submodule_header' function to use
the repository API properly, such that the submodule objects
are not added to the main object store.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agopretty: prepare format_commit_message to handle arbitrary repositories
Stefan Beller [Wed, 14 Nov 2018 00:13:00 +0000 (16:13 -0800)]
pretty: prepare format_commit_message to handle arbitrary repositories

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit: prepare logmsg_reencode to handle arbitrary repositories
Stefan Beller [Wed, 14 Nov 2018 00:12:59 +0000 (16:12 -0800)]
commit: prepare logmsg_reencode to handle arbitrary repositories

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit: prepare repo_unuse_commit_buffer to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:58 +0000 (16:12 -0800)]
commit: prepare repo_unuse_commit_buffer to handle any repo

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit: prepare get_commit_buffer to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:57 +0000 (16:12 -0800)]
commit: prepare get_commit_buffer to handle any repo

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit-reach: prepare in_merge_bases[_many] to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:56 +0000 (16:12 -0800)]
commit-reach: prepare in_merge_bases[_many] to handle any repo

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit-reach: prepare get_merge_bases to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:55 +0000 (16:12 -0800)]
commit-reach: prepare get_merge_bases to handle any repo

Similarly to previous patches, the get_merge_base functions are used
often in the code base, which makes migrating them hard.

Implement the new functions, prefixed with 'repo_' and hide the old
functions behind a wrapper macro.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit-reach.c: allow get_merge_bases_many_0 to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:54 +0000 (16:12 -0800)]
commit-reach.c: allow get_merge_bases_many_0 to handle any repo

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit-reach.c: allow remove_redundant to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:53 +0000 (16:12 -0800)]
commit-reach.c: allow remove_redundant to handle any repo

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit-reach.c: allow merge_bases_many to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:52 +0000 (16:12 -0800)]
commit-reach.c: allow merge_bases_many to handle any repo

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit-reach.c: allow paint_down_to_common to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:51 +0000 (16:12 -0800)]
commit-reach.c: allow paint_down_to_common to handle any repo

As the function is file local and not widely used, migrate it all at once.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agocommit: allow parse_commit* to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:50 +0000 (16:12 -0800)]
commit: allow parse_commit* to handle any repo

Just like the previous commit, parse_commit and friends are used a lot
and are found in new patches, so we cannot change their signature easily.

Re-introduce these function prefixed with 'repo_' that take a repository
argument and keep the original as a shallow macro.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoobject: parse_object to honor its repository argument
Stefan Beller [Wed, 14 Nov 2018 00:12:49 +0000 (16:12 -0800)]
object: parse_object to honor its repository argument

In 8e4b0b6047 (object.c: allow parse_object to handle
arbitrary repositories, 2018-06-28), we forgot to pass the
repository down to the read_object_file.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoobject-store: prepare has_{sha1, object}_file to handle any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:48 +0000 (16:12 -0800)]
object-store: prepare has_{sha1, object}_file to handle any repo

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoobject-store: prepare read_object_file to deal with any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:47 +0000 (16:12 -0800)]
object-store: prepare read_object_file to deal with any repo

As read_object_file is a widely used function (which is also regularly used
in new code in flight between master..pu), changing its signature is painful
is hard, as other series in flight rely on the original signature. It would
burden the maintainer if we'd just change the signature.

Introduce repo_read_object_file which takes the repository argument, and
hide the original read_object_file as a macro behind
NO_THE_REPOSITORY_COMPATIBILITY_MACROS, similar to
e675765235 (diff.c: remove implicit dependency on the_index, 2018-09-21)

Add a coccinelle patch to convert existing callers, but do not apply
the resulting patch to keep the diff of this patch small.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoobject-store: allow read_object_file_extended to read from any repo
Stefan Beller [Wed, 14 Nov 2018 00:12:46 +0000 (16:12 -0800)]
object-store: allow read_object_file_extended to read from any repo

read_object_file_extended is not widely used, so migrate it all at once.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agopackfile: allow has_packed_and_bad to handle arbitrary repositories
Stefan Beller [Tue, 16 Oct 2018 23:35:33 +0000 (16:35 -0700)]
packfile: allow has_packed_and_bad to handle arbitrary repositories

has_packed_and_bad is not widely used, so just migrate it all at once.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agosha1_file: allow read_object to read objects in arbitrary repositories
Stefan Beller [Tue, 16 Oct 2018 23:35:32 +0000 (16:35 -0700)]
sha1_file: allow read_object to read objects in arbitrary repositories

Allow read_object (a file local functon in sha1_file) to
handle arbitrary repositories by passing the repository down
to oid_object_info_extended.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoFifth batch for 2.20
Junio C Hamano [Fri, 19 Oct 2018 04:52:51 +0000 (13:52 +0900)]
Fifth batch for 2.20

Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoMerge branch 'jt/cache-tree-allow-missing-object-in-partial-clone'
Junio C Hamano [Fri, 19 Oct 2018 04:34:08 +0000 (13:34 +0900)]
Merge branch 'jt/cache-tree-allow-missing-object-in-partial-clone'

In a partial clone that will lazily be hydrated from the
originating repository, we generally want to avoid "does this
object exist (locally)?" on objects that we deliberately omitted
when we created the clone.  The cache-tree codepath (which is used
to write a tree object out of the index) however insisted that the
object exists, even for paths that are outside of the partial
checkout area.  The code has been updated to avoid such a check.

* jt/cache-tree-allow-missing-object-in-partial-clone:
  cache-tree: skip some blob checks in partial clone

2 years agoMerge branch 'tb/filter-alternate-refs'
Junio C Hamano [Fri, 19 Oct 2018 04:34:08 +0000 (13:34 +0900)]
Merge branch 'tb/filter-alternate-refs'

When pushing into a repository that borrows its objects from an
alternate object store, "git receive-pack" that responds to the
push request on the other side lists the tips of refs in the
alternate to reduce the amount of objects transferred.  This
sometimes is detrimental when the number of refs in the alternate
is absurdly large, in which case the bandwidth saved in potentially
fewer objects transferred is wasted in excessively large ref
advertisement.  The alternate refs that are advertised are now
configurable with a pair of configuration variables.

* tb/filter-alternate-refs:
  transport.c: introduce core.alternateRefsPrefixes
  transport.c: introduce core.alternateRefsCommand
  transport.c: extract 'fill_alternate_refs_command'
  transport: drop refnames from for_each_alternate_ref

2 years agoMerge branch 'jt/avoid-ls-refs'
Junio C Hamano [Fri, 19 Oct 2018 04:34:07 +0000 (13:34 +0900)]
Merge branch 'jt/avoid-ls-refs'

Over some transports, fetching objects with an exact commit object
name can be done without first seeing the ref advertisements.  The
code has been optimized to exploit this.

* jt/avoid-ls-refs:
  fetch: do not list refs if fetching only hashes
  transport: list refs before fetch if necessary
  transport: do not list refs if possible
  transport: allow skipping of ref listing

2 years agoMerge branch 'ds/commit-graph-leakfix'
Junio C Hamano [Fri, 19 Oct 2018 04:34:07 +0000 (13:34 +0900)]
Merge branch 'ds/commit-graph-leakfix'

Code clean-up.

* ds/commit-graph-leakfix:
  commit-graph: reduce initial oid allocation
  builtin/commit-graph.c: UNLEAK variables
  commit-graph: clean up leaked memory during write

2 years agoMerge branch 'jt/non-blob-lazy-fetch'
Junio C Hamano [Fri, 19 Oct 2018 04:34:07 +0000 (13:34 +0900)]
Merge branch 'jt/non-blob-lazy-fetch'

A partial clone that is configured to lazily fetch missing objects
will on-demand issue a "git fetch" request to the originating
repository to fill not-yet-obtained objects.  The request has been
optimized for requesting a tree object (and not the leaf blob
objects contained in it) by telling the originating repository that
no blobs are needed.

* jt/non-blob-lazy-fetch:
  fetch-pack: exclude blobs when lazy-fetching trees
  fetch-pack: avoid object flags if no_dependents

2 years agoMerge branch 'pw/diff-color-moved-ws-fix'
Junio C Hamano [Fri, 19 Oct 2018 04:34:06 +0000 (13:34 +0900)]
Merge branch 'pw/diff-color-moved-ws-fix'

Various fixes to "diff --color-moved-ws".

* pw/diff-color-moved-ws-fix:
  diff --color-moved: fix a memory leak
  diff --color-moved-ws: fix another memory leak
  diff --color-moved-ws: fix a memory leak
  diff --color-moved-ws: fix out of bounds string access
  diff --color-moved-ws: fix double free crash

2 years agoMerge branch 'rs/oidset-on-khash'
Junio C Hamano [Fri, 19 Oct 2018 04:34:06 +0000 (13:34 +0900)]
Merge branch 'rs/oidset-on-khash'

The oidset API was built on top of the oidmap API which in turn is
on the hashmap API.  Replace the implementation to build on top of
the khash API and gain performance.

* rs/oidset-on-khash:
  oidset: uninline oidset_init()
  oidset: use khash
  khash: factor out kh_release_*
  fetch-pack: load tip_oids eagerly iff needed
  fetch-pack: factor out is_unmatched_ref()

2 years agoMerge branch 'rs/grep-no-recursive'
Junio C Hamano [Fri, 19 Oct 2018 04:34:05 +0000 (13:34 +0900)]
Merge branch 'rs/grep-no-recursive'

Unlike "grep", "git grep" by default recurses to the whole tree.
The command learned "git grep --recursive" option, so that "git
grep --no-recursive" can serve as a synonym to setting the
max-depth to 0.

* rs/grep-no-recursive:
  grep: add -r/--[no-]recursive

2 years agoMerge branch 'nd/help-commands-verbose-by-default'
Junio C Hamano [Fri, 19 Oct 2018 04:34:05 +0000 (13:34 +0900)]
Merge branch 'nd/help-commands-verbose-by-default'

"git help -a" and "git help -av" give different pieces of
information, and generally the "verbose" version is more friendly
to the new users.  "git help -a" by default now uses the more
verbose output (with "--no-verbose", you can go back to the
original).  Also "git help -av" now lists aliases and external
commands, which it did not used to.

* nd/help-commands-verbose-by-default:
  help -a: improve and make --verbose default

2 years agoMerge branch 'jc/how-to-document-api'
Junio C Hamano [Fri, 19 Oct 2018 04:34:05 +0000 (13:34 +0900)]
Merge branch 'jc/how-to-document-api'

Doc update.

* jc/how-to-document-api:
  CodingGuidelines: document the API in *.h files

2 years agoMerge branch 'sm/show-superproject-while-conflicted'
Junio C Hamano [Fri, 19 Oct 2018 04:34:04 +0000 (13:34 +0900)]
Merge branch 'sm/show-superproject-while-conflicted'

A corner-case bugfix.

* sm/show-superproject-while-conflicted:
  rev-parse: --show-superproject-working-tree should work during a merge

2 years agoMerge branch 'jt/fetch-tips-in-partial-clone'
Junio C Hamano [Fri, 19 Oct 2018 04:34:04 +0000 (13:34 +0900)]
Merge branch 'jt/fetch-tips-in-partial-clone'

"git fetch $repo $object" in a partial clone did not correctly
fetch the asked-for object that is referenced by an object in
promisor packfile, which has been fixed.

* jt/fetch-tips-in-partial-clone:
  fetch: in partial clone, check presence of targets
  connected: document connectivity in partial clones

2 years agoMerge branch 'nd/status-refresh-progress'
Junio C Hamano [Fri, 19 Oct 2018 04:34:03 +0000 (13:34 +0900)]
Merge branch 'nd/status-refresh-progress'

"git status" learns to show progress bar when refreshing the index
takes a long time.

* nd/status-refresh-progress:
  status: show progress bar if refreshing the index takes too long

2 years agoMerge branch 'bp/read-cache-parallel'
Junio C Hamano [Fri, 19 Oct 2018 04:34:03 +0000 (13:34 +0900)]
Merge branch 'bp/read-cache-parallel'

A new extension to the index file has been introduced, which allows
the file to be read in parallel.

* bp/read-cache-parallel:
  read-cache: load cache entries on worker threads
  ieot: add Index Entry Offset Table (IEOT) extension
  read-cache: load cache extensions on a worker thread
  config: add new index.threads config setting
  eoie: add End of Index Entry (EOIE) extension
  read-cache: clean up casting and byte decoding
  read-cache.c: optimize reading index format v4

2 years agoMerge branch 'bp/rename-test-env-var'
Junio C Hamano [Fri, 19 Oct 2018 04:34:03 +0000 (13:34 +0900)]
Merge branch 'bp/rename-test-env-var'

Some environment variables that control the runtime options of Git
used during tests are getting renamed for consistency.

* bp/rename-test-env-var:
  t0000: do not get self-test disrupted by environment warnings
  preload-index: update GIT_FORCE_PRELOAD_TEST support
  read-cache: update TEST_GIT_INDEX_VERSION support
  fsmonitor: update GIT_TEST_FSMONITOR support
  preload-index: use git_env_bool() not getenv() for customization
  t/README: correct spelling of "uncommon"

2 years agoMerge branch 'ss/wt-status-committable'
Junio C Hamano [Fri, 19 Oct 2018 04:34:02 +0000 (13:34 +0900)]
Merge branch 'ss/wt-status-committable'

Code clean-up in the internal machinery used by "git status" and
"git commit --dry-run".

* ss/wt-status-committable:
  roll wt_status_state into wt_status and populate in the collect phase
  wt-status.c: set the committable flag in the collect phase
  t7501: add test of "commit --dry-run --short"
  wt-status: rename commitable to committable
  wt-status.c: move has_unmerged earlier in the file

2 years agoMerge branch 'nd/the-index'
Junio C Hamano [Fri, 19 Oct 2018 04:34:02 +0000 (13:34 +0900)]
Merge branch 'nd/the-index'

Various codepaths in the core-ish part learn to work on an
arbitrary in-core index structure, not necessarily the default
instance "the_index".

* nd/the-index: (23 commits)
  revision.c: reduce implicit dependency the_repository
  revision.c: remove implicit dependency on the_index
  ws.c: remove implicit dependency on the_index
  tree-diff.c: remove implicit dependency on the_index
  submodule.c: remove implicit dependency on the_index
  line-range.c: remove implicit dependency on the_index
  userdiff.c: remove implicit dependency on the_index
  rerere.c: remove implicit dependency on the_index
  sha1-file.c: remove implicit dependency on the_index
  patch-ids.c: remove implicit dependency on the_index
  merge.c: remove implicit dependency on the_index
  merge-blobs.c: remove implicit dependency on the_index
  ll-merge.c: remove implicit dependency on the_index
  diff-lib.c: remove implicit dependency on the_index
  read-cache.c: remove implicit dependency on the_index
  diff.c: remove implicit dependency on the_index
  grep.c: remove implicit dependency on the_index
  diff.c: remove the_index dependency in textconv() functions
  blame.c: rename "repo" argument to "r"
  combine-diff.c: remove implicit dependency on the_index
  ...

2 years agoMerge branch 'nd/complete-fetch-multiple-args'
Junio C Hamano [Fri, 19 Oct 2018 04:34:01 +0000 (13:34 +0900)]
Merge branch 'nd/complete-fetch-multiple-args'

Teach bash completion that "git fetch --multiple" only takes remote
names as arguments and no refspecs.

* nd/complete-fetch-multiple-args:
  completion: support "git fetch --multiple"

2 years agoFourth batch for 2.20
Junio C Hamano [Tue, 16 Oct 2018 07:21:17 +0000 (16:21 +0900)]
Fourth batch for 2.20

Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoMerge branch 'sf/complete-stash-list'
Junio C Hamano [Tue, 16 Oct 2018 07:16:09 +0000 (16:16 +0900)]
Merge branch 'sf/complete-stash-list'

The completion script (in contrib/) learned to complete a handful of
options "git stash list" command takes.

* sf/complete-stash-list:
  git-completion.bash: add completion for stash list

2 years agoMerge branch 'mw/doc-typofixes'
Junio C Hamano [Tue, 16 Oct 2018 07:16:09 +0000 (16:16 +0900)]
Merge branch 'mw/doc-typofixes'

Typofixes.

* mw/doc-typofixes:
  docs: typo: s/isimilar/similar/
  docs: graph: remove unnecessary `graph_update()' call
  docs: typo: s/go/to/

2 years agoMerge branch 'js/mingw-wants-vista-or-above'
Junio C Hamano [Tue, 16 Oct 2018 07:16:08 +0000 (16:16 +0900)]
Merge branch 'js/mingw-wants-vista-or-above'

The minimum version of Windows supported by Windows port fo Git is
now set to Vista.

* js/mingw-wants-vista-or-above:
  mingw: bump the minimum Windows version to Vista
  mingw: set _WIN32_WINNT explicitly for Git for Windows
  compat/poll: prepare for targeting Windows Vista

2 years agoMerge branch 'rs/sequencer-oidset-insert-avoids-dups'
Junio C Hamano [Tue, 16 Oct 2018 07:16:08 +0000 (16:16 +0900)]
Merge branch 'rs/sequencer-oidset-insert-avoids-dups'

Code clean-up.

* rs/sequencer-oidset-insert-avoids-dups:
  sequencer: use return value of oidset_insert()

2 years agoMerge branch 'jk/oideq-hasheq-cleanup'
Junio C Hamano [Tue, 16 Oct 2018 07:16:07 +0000 (16:16 +0900)]
Merge branch 'jk/oideq-hasheq-cleanup'

Code clean-up.

* jk/oideq-hasheq-cleanup:
  more oideq/hasheq conversions

2 years agoMerge branch 'ma/mailing-list-address-in-git-help'
Junio C Hamano [Tue, 16 Oct 2018 07:16:07 +0000 (16:16 +0900)]
Merge branch 'ma/mailing-list-address-in-git-help'

Doc update.

* ma/mailing-list-address-in-git-help:
  git doc: direct bug reporters to mailing list archive

2 years agoMerge branch 'nd/packobjectshook-doc-fix'
Junio C Hamano [Tue, 16 Oct 2018 07:16:07 +0000 (16:16 +0900)]
Merge branch 'nd/packobjectshook-doc-fix'

Doc update.

* nd/packobjectshook-doc-fix:
  config.txt: correct the note about uploadpack.packObjectsHook

2 years agoMerge branch 'rt/rebase-typofix'
Junio C Hamano [Tue, 16 Oct 2018 07:16:06 +0000 (16:16 +0900)]
Merge branch 'rt/rebase-typofix'

Typofix.

* rt/rebase-typofix:
  git-rebase.sh: fix typos in error messages

2 years agoMerge branch 'ma/t1400-undebug-test'
Junio C Hamano [Tue, 16 Oct 2018 07:16:06 +0000 (16:16 +0900)]
Merge branch 'ma/t1400-undebug-test'

Test fix.

* ma/t1400-undebug-test:
  t1400: drop debug `echo` to actually execute `test`

2 years agoMerge branch 'ma/commit-graph-docs'
Junio C Hamano [Tue, 16 Oct 2018 07:16:05 +0000 (16:16 +0900)]
Merge branch 'ma/commit-graph-docs'

Doc update.

* ma/commit-graph-docs:
  Doc: refer to the "commit-graph file" with dash
  git-commit-graph.txt: refer to "*commit*-graph file"
  git-commit-graph.txt: typeset more in monospace
  git-commit-graph.txt: fix bullet lists

2 years agoMerge branch 'dz/credential-doc-url-matching-rules'
Junio C Hamano [Tue, 16 Oct 2018 07:16:05 +0000 (16:16 +0900)]
Merge branch 'dz/credential-doc-url-matching-rules'

Doc update.

* dz/credential-doc-url-matching-rules:
  doc: clarify gitcredentials path component matching

2 years agoMerge branch 'en/status-multiple-renames-to-the-same-target-fix'
Junio C Hamano [Tue, 16 Oct 2018 07:16:05 +0000 (16:16 +0900)]
Merge branch 'en/status-multiple-renames-to-the-same-target-fix'

The code in "git status" sometimes hit an assertion failure.  This
was caused by a structure that was reused without cleaning the data
used for the first run, which has been corrected.

* en/status-multiple-renames-to-the-same-target-fix:
  commit: fix erroneous BUG, 'multiple renames on the same target? how?'

2 years agoMerge branch 'ds/reachable-final-cleanup'
Junio C Hamano [Tue, 16 Oct 2018 07:16:04 +0000 (16:16 +0900)]
Merge branch 'ds/reachable-final-cleanup'

Code already in 'master' is further cleaned-up by this patch.

* ds/reachable-final-cleanup:
  commit-reach: cleanups in can_all_from_reach...

2 years agoMerge branch 'jk/check-everything-connected-is-long-gone'
Junio C Hamano [Tue, 16 Oct 2018 07:16:04 +0000 (16:16 +0900)]
Merge branch 'jk/check-everything-connected-is-long-gone'

Comment fix.

* jk/check-everything-connected-is-long-gone:
  receive-pack: update comment with check_everything_connected

2 years agoMerge branch 'jn/gc-auto'
Junio C Hamano [Tue, 16 Oct 2018 07:16:02 +0000 (16:16 +0900)]
Merge branch 'jn/gc-auto'

"gc --auto" ended up calling exit(-1) upon error, which has been
corrected to use exit(1).  Also the error reporting behaviour when
daemonized has been updated to exit with zero status when stopping
due to a previously discovered error (which implies there is no
point running gc to improve the situation); we used to exit with
failure in such a case.

* jn/gc-auto:
  gc: do not return error for prior errors in daemonized mode

2 years agoMerge branch 'jn/gc-auto-prep'
Junio C Hamano [Tue, 16 Oct 2018 07:16:02 +0000 (16:16 +0900)]
Merge branch 'jn/gc-auto-prep'

Code clean-up.

* jn/gc-auto-prep:
  gc: exit with status 128 on failure
  gc: improve handling of errors reading gc.log

2 years agoMerge branch 'md/test-cleanup'
Junio C Hamano [Tue, 16 Oct 2018 07:16:01 +0000 (16:16 +0900)]
Merge branch 'md/test-cleanup'

Various test scripts have been updated for style and also correct
handling of exit status of various commands.

* md/test-cleanup:
  tests: order arguments to git-rev-list properly
  t9109: don't swallow Git errors upstream of pipes
  tests: don't swallow Git errors upstream of pipes
  t/*: fix ordering of expected/observed arguments
  tests: standardize pipe placement
  Documentation: add shell guidelines
  t/README: reformat Do, Don't, Keep in mind lists

2 years agoMerge branch 'fe/doc-updates'
Junio C Hamano [Tue, 16 Oct 2018 07:16:01 +0000 (16:16 +0900)]
Merge branch 'fe/doc-updates'

Doc updates.

* fe/doc-updates:
  git-describe.1: clarify that "human readable" is also git-readable
  git-column.1: clarify initial description, provide examples
  git-archimport.1: specify what kind of Arch we're talking about

2 years agoMerge branch 'jn/mailmap-update'
Junio C Hamano [Tue, 16 Oct 2018 07:16:01 +0000 (16:16 +0900)]
Merge branch 'jn/mailmap-update'

The mailmap file update.

* jn/mailmap-update:
  mailmap: consistently normalize brian m. carlson's name

2 years agoMerge branch 'tg/t5551-with-curl-7.61.1'
Junio C Hamano [Tue, 16 Oct 2018 07:16:00 +0000 (16:16 +0900)]
Merge branch 'tg/t5551-with-curl-7.61.1'

Test update.

* tg/t5551-with-curl-7.61.1:
  t5551: compare sorted cookies files
  t5551: move setup code inside test_expect blocks

2 years agoMerge branch 'en/merge-cleanup'
Junio C Hamano [Tue, 16 Oct 2018 07:16:00 +0000 (16:16 +0900)]
Merge branch 'en/merge-cleanup'

Code clean-up.

* en/merge-cleanup:
  merge-recursive: rename merge_file_1() and merge_content()
  merge-recursive: remove final remaining caller of merge_file_one()
  merge-recursive: avoid wrapper function when unnecessary and wasteful
  merge-recursive: set paths correctly when three-way merging content

2 years agoMerge branch 'rj/header-check'
Junio C Hamano [Tue, 16 Oct 2018 07:16:00 +0000 (16:16 +0900)]
Merge branch 'rj/header-check'

Header files clean-up.

* rj/header-check:
  delta-islands.h: add missing forward declarations (hdr-check)
  midx.h: add missing forward declarations (hdr-check)
  refs/refs-internal.h: add missing declarations (hdr-check)
  refs/packed-backend.h: add missing declaration (hdr-check)
  refs/ref-cache.h: add missing declarations (hdr-check)
  ewah/ewok_rlw.h: add missing include (hdr-check)
  json-writer.h: add missing include (hdr-check)
  Makefile: add a hdr-check target

2 years agoMerge branch 'ma/config-doc-update'
Junio C Hamano [Tue, 16 Oct 2018 07:16:00 +0000 (16:16 +0900)]
Merge branch 'ma/config-doc-update'

Doc update.

* ma/config-doc-update:
  git-config.txt: fix 'see: above' note
  Doc: use `--type=bool` instead of `--bool`

2 years agoMerge branch 'jk/delta-islands-with-bitmap-reuse-delta-fix'
Junio C Hamano [Tue, 16 Oct 2018 07:16:00 +0000 (16:16 +0900)]
Merge branch 'jk/delta-islands-with-bitmap-reuse-delta-fix'

Fix interactions between two recent topics.

* jk/delta-islands-with-bitmap-reuse-delta-fix:
  pack-objects: handle island check for "external" delta base

2 years agoMerge branch 'tq/refs-internal-comment-fix'
Junio C Hamano [Tue, 16 Oct 2018 07:15:59 +0000 (16:15 +0900)]
Merge branch 'tq/refs-internal-comment-fix'

Fix for typo in a sample code in comment.

* tq/refs-internal-comment-fix:
  refs: docstring typo

2 years agoMerge branch 'ts/alias-of-alias'
Junio C Hamano [Tue, 16 Oct 2018 07:15:59 +0000 (16:15 +0900)]
Merge branch 'ts/alias-of-alias'

An alias that expands to another alias has so far been forbidden,
but now it is allowed to create such an alias.

* ts/alias-of-alias:
  t0014: introduce an alias testing suite
  alias: show the call history when an alias is looping
  alias: add support for aliases of an alias

2 years agoMerge branch 'ds/commit-graph-with-grafts'
Junio C Hamano [Tue, 16 Oct 2018 07:15:59 +0000 (16:15 +0900)]
Merge branch 'ds/commit-graph-with-grafts'

The recently introduced commit-graph auxiliary data is incompatible
with mechanisms such as replace & grafts that "breaks" immutable
nature of the object reference relationship.  Disable optimizations
based on its use (and updating existing commit-graph) when these
incompatible features are in use in the repository.

* ds/commit-graph-with-grafts:
  commit-graph: close_commit_graph before shallow walk
  commit-graph: not compatible with uninitialized repo
  commit-graph: not compatible with grafts
  commit-graph: not compatible with replace objects
  test-repository: properly init repo
  commit-graph: update design document
  refs.c: upgrade for_each_replace_ref to be a each_repo_ref_fn callback
  refs.c: migrate internal ref iteration to pass thru repository argument

2 years agoMerge branch 'ab/commit-graph-progress'
Junio C Hamano [Tue, 16 Oct 2018 07:15:58 +0000 (16:15 +0900)]
Merge branch 'ab/commit-graph-progress'

Generation of (experimental) commit-graph files have so far been
fairly silent, even though it takes noticeable amount of time in a
meaningfully large repository.  The users will now see progress
output.

* ab/commit-graph-progress:
  gc: fix regression in 7b0f229222 impacting --quiet
  commit-graph verify: add progress output
  commit-graph write: add progress output

2 years agoread-cache: load cache entries on worker threads
Ben Peart [Wed, 10 Oct 2018 15:59:38 +0000 (11:59 -0400)]
read-cache: load cache entries on worker threads

This patch helps address the CPU cost of loading the index by utilizing
the Index Entry Offset Table (IEOT) to divide loading and conversion of
the cache entries across multiple threads in parallel.

I used p0002-read-cache.sh to generate some performance data:

Test w/100,000 files reduced the time by 32.24%
Test w/1,000,000 files reduced the time by -4.77%

Note that on the 1,000,000 files case, multi-threading the cache entry parsing
does not yield a performance win.  This is because the cost to parse the
index extensions in this repo, far outweigh the cost of loading the cache
entries.

The high cost of parsing the index extensions is driven by the cache tree
and the untracked cache extensions. As this is currently the longest pole,
any reduction in this time will reduce the overall index load times so is
worth further investigation in another patch series.

Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoieot: add Index Entry Offset Table (IEOT) extension
Ben Peart [Wed, 10 Oct 2018 15:59:37 +0000 (11:59 -0400)]
ieot: add Index Entry Offset Table (IEOT) extension

This patch enables addressing the CPU cost of loading the index by adding
additional data to the index that will allow us to efficiently multi-
thread the loading and conversion of cache entries.

It accomplishes this by adding an (optional) index extension that is a
table of offsets to blocks of cache entries in the index file.  To make
this work for V4 indexes, when writing the cache entries, it periodically
"resets" the prefix-compression by encoding the current entry as if the
path name for the previous entry is completely different and saves the
offset of that entry in the IEOT.  Basically, with V4 indexes, it
generates offsets into blocks of prefix-compressed entries.

Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoread-cache: load cache extensions on a worker thread
Ben Peart [Wed, 10 Oct 2018 15:59:36 +0000 (11:59 -0400)]
read-cache: load cache extensions on a worker thread

This patch helps address the CPU cost of loading the index by loading
the cache extensions on a worker thread in parallel with loading the cache
entries.

In some cases, loading the extensions takes longer than loading the
cache entries so this patch utilizes the new EOIE to start the thread to
load the extensions before loading all the cache entries in parallel.

This is possible because the current extensions don't access the cache
entries in the index_state structure so are OK that they don't all exist
yet.

The CACHE_EXT_TREE, CACHE_EXT_RESOLVE_UNDO, and CACHE_EXT_UNTRACKED
extensions don't even get a pointer to the index so don't have access to the
cache entries.

CACHE_EXT_LINK only uses the index_state to initialize the split index.
CACHE_EXT_FSMONITOR only uses the index_state to save the fsmonitor last
update and dirty flags.

I used p0002-read-cache.sh to generate some performance data:

Test w/100,000 files reduced the time by 0.53%
Test w/1,000,000 files reduced the time by 27.78%

Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoconfig: add new index.threads config setting
Ben Peart [Wed, 10 Oct 2018 15:59:35 +0000 (11:59 -0400)]
config: add new index.threads config setting

Add support for a new index.threads config setting which will be used to
control the threading code in do_read_index().  A value of 0 will tell the
index code to automatically determine the correct number of threads to use.
A value of 1 will make the code single threaded.  A value greater than 1
will set the maximum number of threads to use.

For testing purposes, this setting can be overwritten by setting the
GIT_TEST_INDEX_THREADS=<n> environment variable to a value greater than 0.

Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoeoie: add End of Index Entry (EOIE) extension
Ben Peart [Wed, 10 Oct 2018 15:59:34 +0000 (11:59 -0400)]
eoie: add End of Index Entry (EOIE) extension

The End of Index Entry (EOIE) is used to locate the end of the variable
length index entries and the beginning of the extensions. Code can take
advantage of this to quickly locate the index extensions without having
to parse through all of the index entries.

The EOIE extension is always written out to the index file including to
the shared index when using the split index feature. Because it is always
written out, the SHA checksums in t/t1700-split-index.sh were updated
to reflect its inclusion.

It is written as an optional extension to ensure compatibility with other
git implementations that do not yet support it.  It is always written out
to ensure it is available as often as possible to speed up index operations.

Because it must be able to be loaded before the variable length cache
entries and other index extensions, this extension must be written last.
The signature for this extension is { 'E', 'O', 'I', 'E' }.

The extension consists of:

- 32-bit offset to the end of the index entries

- 160-bit SHA-1 over the extension types and their sizes (but not
their contents).  E.g. if we have "TREE" extension that is N-bytes
long, "REUC" extension that is M-bytes long, followed by "EOIE",
then the hash would be:

SHA-1("TREE" + <binary representation of N> +
    "REUC" + <binary representation of M>)

Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoread-cache: clean up casting and byte decoding
Ben Peart [Wed, 10 Oct 2018 15:59:33 +0000 (11:59 -0400)]
read-cache: clean up casting and byte decoding

This patch does a clean up pass to minimize the casting required to work
with the memory mapped index (mmap).

It also makes the decoding of network byte order more consistent by using
get_be32() where possible.

Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoThird batch for 2.20
Junio C Hamano [Wed, 10 Oct 2018 03:38:03 +0000 (12:38 +0900)]
Third batch for 2.20

Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoMerge branch 'ab/fsck-skiplist'
Junio C Hamano [Wed, 10 Oct 2018 03:37:16 +0000 (12:37 +0900)]
Merge branch 'ab/fsck-skiplist'

Update fsck.skipList implementation and documentation.

* ab/fsck-skiplist:
  fsck: support comments & empty lines in skipList
  fsck: use oidset instead of oid_array for skipList
  fsck: use strbuf_getline() to read skiplist file
  fsck: add a performance test for skipList
  fsck: add a performance test
  fsck: document that skipList input must be unabbreviated
  fsck: document and test commented & empty line skipList input
  fsck: document and test sorted skipList input
  fsck tests: add a test for no skipList input
  fsck tests: setup of bogus commit object

2 years agoMerge branch 'ds/multi-pack-verify'
Junio C Hamano [Wed, 10 Oct 2018 03:37:16 +0000 (12:37 +0900)]
Merge branch 'ds/multi-pack-verify'

"git multi-pack-index" learned to detect corruption in the .midx
file it uses, and this feature has been integrated into "git fsck".

* ds/multi-pack-verify:
  fsck: verify multi-pack-index
  multi-pack-index: report progress during 'verify'
  multi-pack-index: verify object offsets
  multi-pack-index: fix 32-bit vs 64-bit size check
  multi-pack-index: verify oid lookup order
  multi-pack-index: verify oid fanout order
  multi-pack-index: verify missing pack
  multi-pack-index: verify packname order
  multi-pack-index: verify corrupt chunk lookup table
  multi-pack-index: verify bad header
  multi-pack-index: add 'verify' verb

2 years agoMerge branch 'bc/hash-independent-tests'
Junio C Hamano [Wed, 10 Oct 2018 03:37:16 +0000 (12:37 +0900)]
Merge branch 'bc/hash-independent-tests'

Various tests have been updated to make it easier to swap the
hash function used for object identification.

* bc/hash-independent-tests:
  t5318: use test_oid for HASH_LEN
  t1407: make hash size independent
  t1406: make hash-size independent
  t1405: make hash size independent
  t1400: switch hard-coded object ID to variable
  t1006: make hash size independent
  t0064: make hash size independent
  t0002: abstract away SHA-1 specific constants
  t0000: update tests for SHA-256
  t0000: use hash translation table
  t: add test functions to translate hash-related values

2 years agoMerge branch 'nd/test-tool'
Junio C Hamano [Wed, 10 Oct 2018 03:37:16 +0000 (12:37 +0900)]
Merge branch 'nd/test-tool'

Test helper binaries clean-up.

* nd/test-tool:
  Makefile: add a hint about TEST_BUILTINS_OBJS
  t/helper: merge test-dump-fsmonitor into test-tool
  t/helper: merge test-parse-options into test-tool
  t/helper: merge test-pkt-line into test-tool
  t/helper: merge test-dump-untracked-cache into test-tool
  t/helper: keep test-tool command list sorted

2 years agoMerge branch 'nd/config-split'
Junio C Hamano [Wed, 10 Oct 2018 03:37:15 +0000 (12:37 +0900)]
Merge branch 'nd/config-split'

Split Documentation/config.txt for easier maintenance.

* nd/config-split:
  config.txt: move submodule part out to a separate file
  config.txt: move sequence.editor out of "core" part
  config.txt: move sendemail part out to a separate file
  config.txt: move receive part out to a separate file
  config.txt: move push part out to a separate file
  config.txt: move pull part out to a separate file
  config.txt: move gui part out to a separate file
  config.txt: move gitcvs part out to a separate file
  config.txt: move format part out to a separate file
  config.txt: move fetch part out to a separate file
  config.txt: follow camelCase naming

2 years agocache-tree: skip some blob checks in partial clone
Jonathan Tan [Tue, 9 Oct 2018 18:40:37 +0000 (11:40 -0700)]
cache-tree: skip some blob checks in partial clone

In a partial clone, whenever a sparse checkout occurs, the existence of
all blobs in the index is verified, whether they are included or
excluded by the .git/info/sparse-checkout specification. This
significantly degrades performance because a lazy fetch occurs whenever
the existence of a missing blob is checked.

This is because cache_tree_update() checks the existence of all objects
in the index, whether or not CE_SKIP_WORKTREE is set on them. Teach
cache_tree_update() to skip checking CE_SKIP_WORKTREE objects when the
repository is a partial clone. This improves performance for sparse
checkout and also other operations that use cache_tree_update().

Instead of completely removing the check, an argument could be made that
the check should instead be replaced by a check that the blob is
promised, but for performance reasons, I decided not to do this.
If the user needs to verify the repository, it can be done using fsck
(which will notify if a tree points to a missing and non-promised blob,
whether the blob is included or excluded by the sparse-checkout
specification).

Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agoDeclare that the next one will be named 2.20
Junio C Hamano [Wed, 10 Oct 2018 00:20:03 +0000 (09:20 +0900)]
Declare that the next one will be named 2.20

Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotransport.c: introduce core.alternateRefsPrefixes
Taylor Blau [Mon, 8 Oct 2018 18:09:30 +0000 (11:09 -0700)]
transport.c: introduce core.alternateRefsPrefixes

The recently-introduced "core.alternateRefsCommand" allows callers to
specify with high flexibility the tips that they wish to advertise from
alternates. This flexibility comes at the cost of some inconvenience
when the caller only wishes to limit the advertisement to one or more
prefixes.

For example, to advertise only tags, a caller using
'core.alternateRefsCommand' would have to do:

  $ git config core.alternateRefsCommand ' \
      f() { git -C "$1" for-each-ref \
              refs/tags --format="%(objectname)" }; f "$@"'

The above is cumbersome to write, so let's introduce a
"core.alternateRefsPrefixes" to address this common case. Instead, the
caller can run:

  $ git config core.alternateRefsPrefixes 'refs/tags'

Which will behave identically to the longer example using
"core.alternateRefsCommand".

Since the value of "core.alternateRefsPrefixes" is appended to 'git
for-each-ref' and then executed, include a "--" before taking the
configured value to avoid misinterpreting arguments as flags to 'git
for-each-ref'.

In the case that the caller wishes to specify multiple prefixes, they
may separate them by whitespace. If "core.alternateRefsCommand" is set,
it will take precedence over "core.alternateRefsPrefixes".

Signed-off-by: Taylor Blau <me@ttaylorr.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotransport.c: introduce core.alternateRefsCommand
Taylor Blau [Mon, 8 Oct 2018 18:09:28 +0000 (11:09 -0700)]
transport.c: introduce core.alternateRefsCommand

When in a repository containing one or more alternates, Git would
sometimes like to list references from those alternates. For example,
'git receive-pack' lists the "tips" pointed to by references in those
alternates as special ".have" references.

Listing ".have" references is designed to make pushing changes from
upstream to a fork a lightweight operation, by advertising to the pusher
that the fork already has the objects (via its alternate). Thus, the
client can avoid sending them.

However, when the alternate (upstream, in the previous example) has a
pathologically large number of references, the initial advertisement is
too expensive. In fact, it can dominate any such optimization where the
pusher avoids sending certain objects.

Introduce "core.alternateRefsCommand" in order to provide a facility to
limit or filter alternate references. This can be used, for example, to
filter out references the alternate does not wish to send (for space
concerns, or otherwise) during the initial advertisement.

Let the repository that has alternates configure this command to avoid
trusting the alternate to provide us a safe command to run in the shell.
To find the alternate, pass its absolute path as the first argument.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotransport.c: extract 'fill_alternate_refs_command'
Taylor Blau [Mon, 8 Oct 2018 18:09:26 +0000 (11:09 -0700)]
transport.c: extract 'fill_alternate_refs_command'

To list alternate references, 'read_alternate_refs' creates a child
process running 'git for-each-ref' in the alternate's Git directory.

Prepare to run other commands besides 'git for-each-ref' by introducing
and moving the relevant code from 'read_alternate_refs' to
'fill_alternate_refs_command'.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotransport: drop refnames from for_each_alternate_ref
Jeff King [Mon, 8 Oct 2018 18:09:23 +0000 (11:09 -0700)]
transport: drop refnames from for_each_alternate_ref

None of the current callers use the refname parameter we pass to their
callbacks. In theory somebody _could_ do so, but it's actually quite
weird if you think about it: it's a ref in somebody else's repository.
So the name has no meaning locally, and in fact there may be duplicates
if there are multiple alternates.

The users of this interface really only care about seeing some ref tips,
since that promises that the alternate has the full commit graph
reachable from there. So let's keep the information we pass back to the
bare minimum.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agodocs: typo: s/isimilar/similar/
Michael Witten [Sat, 6 Oct 2018 04:20:22 +0000 (04:20 +0000)]
docs: typo: s/isimilar/similar/

Signed-off-by: Michael Witten <mfwitten@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agodocs: graph: remove unnecessary `graph_update()' call
Michael Witten [Sat, 6 Oct 2018 04:20:16 +0000 (04:20 +0000)]
docs: graph: remove unnecessary `graph_update()' call

The sample code calls `get_revision()' followed by `graph_update()',
but the documentation and source code indicate that `get_revision()'
already calls `graph_update()' for you.

Signed-off-by: Michael Witten <mfwitten@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agodocs: typo: s/go/to/
Michael Witten [Sat, 6 Oct 2018 04:20:09 +0000 (04:20 +0000)]
docs: typo: s/go/to/

Signed-off-by: Michael Witten <mfwitten@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agogit-completion.bash: add completion for stash list
Steven Fernandez [Thu, 27 Sep 2018 19:59:00 +0000 (20:59 +0100)]
git-completion.bash: add completion for stash list

Since stash list accepts git-log options, add the following useful
options that make sense in the context of the `git stash list` command:

  --name-status --oneline --patch-with-stat

Signed-off-by: Steven Fernandez <steve@lonetwin.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agofetch: do not list refs if fetching only hashes
Jonathan Tan [Thu, 27 Sep 2018 19:24:07 +0000 (12:24 -0700)]
fetch: do not list refs if fetching only hashes

If only hash literals are given on a "git fetch" command-line, tag
following is not requested, and the fetch is done using protocol v2, a
list of refs is not required from the remote. Therefore, optimize by
invoking transport_get_remote_refs() only if we need the refs.

Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotransport: list refs before fetch if necessary
Jonathan Tan [Thu, 27 Sep 2018 19:24:06 +0000 (12:24 -0700)]
transport: list refs before fetch if necessary

The built-in bundle transport and the transport helper interface do not
work when transport_fetch_refs() is called immediately after transport
creation. This will be needed in a subsequent patch, so fix this.

Evidence: fetch_refs_from_bundle() relies on data->header being
initialized in get_refs_from_bundle(), and fetch() in transport-helper.c
relies on either data->fetch or data->import being set by get_helper(),
but neither transport_helper_init() nor fetch() calls get_helper().

Up until the introduction of the partial clone feature, this has not
been a problem, because transport_fetch_refs() is always called after
transport_get_remote_refs(). With the introduction of the partial clone
feature, which involves calling transport_fetch_refs() (to fetch objects
by their OIDs) without transport_get_remote_refs(), this is still not a
problem, but only coincidentally - we do not support partially cloning a
bundle, and as for cloning using a transport-helper-using protocol, it
so happens that before transport_fetch_refs() is called, fetch_refs() in
fetch-object.c calls transport_set_option(), which means that the
aforementioned get_helper() is invoked through set_helper_option() in
transport-helper.c.

This could be fixed by fixing the transports themselves, but it doesn't
seem like a good idea to me to open up previously untested code paths;
also, there may be transport helpers in the wild that assume that "list"
is always called before "fetch". Instead, fix this by having
transport_fetch_refs() call transport_get_remote_refs() to ensure that
the latter is always called at least once, unless the transport
explicitly states that it supports fetching without listing refs.

Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotransport: do not list refs if possible
Jonathan Tan [Thu, 27 Sep 2018 19:24:05 +0000 (12:24 -0700)]
transport: do not list refs if possible

When all refs to be fetched are exact OIDs, it is possible to perform a
fetch without requiring the remote to list refs if protocol v2 is used.
Teach Git to do this.

This currently has an effect only for lazy fetches done from partial
clones. The change necessary to likewise optimize "git fetch <remote>
<sha-1>" will be done in a subsequent patch.

Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotransport: allow skipping of ref listing
Jonathan Tan [Thu, 27 Sep 2018 19:24:04 +0000 (12:24 -0700)]
transport: allow skipping of ref listing

The get_refs_via_connect() function both performs the handshake
(including determining the protocol version) and obtaining the list of
remote refs. However, the fetch protocol v2 supports fetching objects
without the listing of refs, so make it possible for the user to skip
the listing by creating a new handshake() function. This will be used in
a subsequent commit.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotests: order arguments to git-rev-list properly
Matthew DeVore [Fri, 5 Oct 2018 21:54:07 +0000 (14:54 -0700)]
tests: order arguments to git-rev-list properly

It is a common mistake to put positional arguments before flags when
invoking git-rev-list. Order the positional arguments last.

This patch skips git-rev-list invocations which include the --not flag,
since the ordering of flags and positional arguments affects the
behavior. This patch also skips invocations of git-rev-list that occur
in command substitution in which the exit code is discarded, since
fixing those properly will require a more involved cleanup.

Signed-off-by: Matthew DeVore <matvore@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agot9109: don't swallow Git errors upstream of pipes
Matthew DeVore [Fri, 5 Oct 2018 21:54:06 +0000 (14:54 -0700)]
t9109: don't swallow Git errors upstream of pipes

'git ... | foo' will mask any errors or crashes in git, so split up such
pipes in this file.

One testcase uses several separate pipe sequences in a row which are
awkward to split up. Wrap the split-up pipe in a function so the
awkwardness is not repeated. Also change that testcase's surrounding
quotes from double to single to avoid premature string interpolation.

Signed-off-by: Matthew DeVore <matvore@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agotests: don't swallow Git errors upstream of pipes
Matthew DeVore [Fri, 5 Oct 2018 21:54:05 +0000 (14:54 -0700)]
tests: don't swallow Git errors upstream of pipes

Some pipes in tests lose the exit code of git processes, which can mask
unexpected behavior like crashes. Split these pipes up so that git
commands are only at the end of pipes rather than the beginning or
middle.

The violations fixed in this patch were found in the process of fixing
pipe placement in a prior patch.

Signed-off-by: Matthew DeVore <matvore@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years agot/*: fix ordering of expected/observed arguments
Matthew DeVore [Fri, 5 Oct 2018 21:54:04 +0000 (14:54 -0700)]
t/*: fix ordering of expected/observed arguments

Fix various places where the ordering was obviously wrong, meaning it
was easy to find with grep.

Signed-off-by: Matthew DeVore <matvore@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>