Add a `--revert <branch>` mode to git replay that undoes the changes
introduced by the specified commits. Like --onto and --advance, --revert
is a standalone mode: it takes a branch argument and updates that branch
with the newly created revert commits.
At GitLab, we need this in Gitaly for reverting commits directly on bare
repositories without requiring a working tree checkout.
The approach is the same as sequencer.c's do_pick_commit() -- cherry-pick
and revert are just the same three-way merge with swapped arguments:
- Cherry-pick: merge(ancestor=parent, ours=current, theirs=commit)
- Revert: merge(ancestor=commit, ours=current, theirs=parent)
We swap the base and pickme trees passed to merge_incore_nonrecursive()
to reverse the diff direction.
Reverts are processed newest-first (matching git revert behavior) to
reduce conflicts by peeling off changes from the top. Each revert
builds on the result of the previous one via the last_commit fallback
in the main replay loop, rather than relying on the parent-mapping
used for cherry-pick.
Revert commit messages follow the usual git revert conventions: prefixed
with "Revert" (or "Reapply" when reverting a revert), and including
"This reverts commit <hash>.". The author is set to the current user
rather than preserving the original author, matching git revert behavior.
Helped-by: Christian Couder <christian.couder@gmail.com>
Helped-by: Patrick Steinhardt <ps@pks.im>
Helped-by: Elijah Newren <newren@gmail.com>
Helped-by: Phillip Wood <phillip.wood123@gmail.com>
Helped-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Helped-by: Junio C Hamano <gitster@pobox.com>
Helped-by: Toon Claes <toon@iotcl.com>
Signed-off-by: Siddharth Asthana <siddharthasthana31@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The logic for formatting revert commit messages (handling "Revert" and
"Reapply" cases, appending "This reverts commit <ref>.", and handling
merge-parent references) currently lives inline in do_pick_commit().
The upcoming replay --revert mode needs to reuse this logic.
Extract all of this into a new sequencer_format_revert_message()
function. The function takes a repository, the subject line, commit,
parent, a use_commit_reference flag, and the output strbuf. It handles
both regular reverts ("Revert "<subject>"") and revert-of-revert cases
("Reapply "<subject>""), and uses refer_to_commit() internally to
format the commit reference.
Update refer_to_commit() to take a struct repository parameter instead
of relying on the_repository, and a bool instead of reading from
replay_opts directly. This makes it usable from the new shared function
without pulling in sequencer-specific state.
Signed-off-by: Siddharth Asthana <siddharthasthana31@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Teach "git hook run" and "git hook list" to reject hook event names
that are not recognized by Git. This helps catch typos such as
"prereceive" when "pre-receive" was intended, since in 99% of the
cases users want known (already-existing) hook names.
The list of known hooks is derived from the generated hook-list.h
(built from Documentation/githooks.adoc). This is why the Makefile
is updated, so builtin/hook.c depends on hook-list.h. In meson the
header is already a dependency for all builtins, no change required.
The "--allow-unknown-hook-name" flag can be used to bypass this check.
Suggested-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Disabled hooks were filtered out of the cache entirely, making them
invisible to "git hook list". Keep them in the cache with a new
"disabled" flag which is propagated to the respective struct hook.
"git hook list" now shows disabled hooks as tab-separated columns,
with the status as a prefix before the name (like scope with
--show-scope). With --show-scope it looks like:
$ git hook list --show-scope pre-commit
global linter
local disabled no-leaks
hook from hookdir
A disabled hook without a command issues a warning instead of the
fatal "hook.X.command must be configured" error. We could also throw
an error, however it seemd a bit excessive to me in this case.
Suggested-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Users running "git hook list" can see which hooks are configured but
have no way to tell at which config scope (local, global, system...)
each hook was defined.
Store the scope from ctx->kvi->scope in the single-pass config callback,
then carry it through the cache to the hook structs, so we can expose it
to users via the "git hook list --show-scope" flag, which mirrors the
existing git config --show-scope convention.
Without the flag the output is unchanged.
The scope is printed as a tab-separated prefix (like "git config --show-scope"),
making it unambiguously machine-parseable even when the friendly name
contains spaces.
Example usage:
$ git hook list --show-scope pre-commit
global linter
local no-leaks
hook from hookdir
Traditional hooks from the hookdir are unaffected by --show-scope since
the config scope concept does not apply to them.
Suggested-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Replace the bare `char *command` util pointer stored in each string_list
item with a heap-allocated `struct hook_config_cache_entry` that carries
that command string.
This is just a refactoring with no behavior changes, to give the cache
entry room to grow, so it can carry the additional hook metadata we'll
be adding in the following commits.
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There is a documented expectation that configured hooks are
run before the hook from the hookdir. Add a test for it.
While at it, I noticed that `git hook list -h` runs twice
in the `git hook usage` test, so remove one invocation.
Suggested-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Both `name` and `friendly-name` is being used. Standardize on
`friendly-name` for consistency since name is rather generic,
even when used in the hooks namespace.
Suggested-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Replace the custom function with string_list_clear_func() which
is a more common pattern for clearing a string_list.
To be able to do this, rework hook_clear() into hook_free(), so
it can be passed to string_list_clear_func().
A slight complication is the need to keep a copy of the internal
cb data free() pointer, however I think it's worth it since the
API becomes cleaner, e.g. no more calls with NULL function args
like hook_list_clear(hooks, NULL).
In other words, the callers don't need to keep track of hook
internal state to determine when cleanup is necessary or not
(pass NULL) because each `struct hook` now owns its data_free
callback.
Suggested-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Trigger a bug when an unknown hook type is encountered while
setting up hook execution.
Also issue a bug if a configured hook is enabled without a cmd.
Mostly useful for defensive coding.
Suggested-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Rename the hook callback function types to use the hook prefix.
This is a style fix with no logic changes.
Suggested-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix some minor style nits pointed out by Patrick, Junio and Eric:
* Use CALLOC_ARRAY instead of xcalloc.
* Init struct members during declaration.
* Simplify if condition boolean logic.
* Missing curly braces in if/else stmts.
* Unnecessary header includes.
* Capitalization and full-stop in error/warn messages.
* Curly brace on separate line when defining struct.
* Comment spelling: free'd -> freed.
* Sort the included headers.
* Blank line fixes to improve readability.
These contain no logic changes, the code behaves the same as before.
Suggested-by: Eric Sunshine <sunshine@sunshineco.com>
Suggested-by: Junio C Hamano <gitster@pobox.com>
Suggested-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The run_receive_hook() stack-allocated `struct receive_hook_feed_state`
is a template with initial values for child states allocated on the heap
for each hook process, by calling receive_hook_feed_state_alloc() when
spinning up each hook child.
All these values are already initialized to zero, however I forgot to
properly initialize the strbuf, which I left NULL.
This is more of a code cleanup because in practice it has no effect,
the states used by the children are always initialized, however it's
good to fix in case someone ends up accidentally dereferencing the NULL
pointer in the future.
Reported-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move the convenience wrapper from hook to string-list since
it's a more suitable place. Add a doc comment to the header.
Also add a free_util arg to make the function more generic
and make the API similar to other functions in string-list.h.
Update the existing call-sites.
Suggested-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The logic to count objects has been cleaned up.
* ps/object-counting:
odb: introduce generic object counting
odb/source: introduce generic object counting
object-file: generalize counting objects
object-file: extract logic to approximate object count
packfile: extract logic to count number of objects
odb: stop including "odb/source.h"
Further work on incremental repacking using MIDX/bitmap
* tb/incremental-midx-part-3.2:
midx: enable reachability bitmaps during MIDX compaction
midx: implement MIDX compaction
t/helper/test-read-midx.c: plug memory leak when selecting layer
midx-write.c: factor fanout layering from `compute_sorted_entries()`
midx-write.c: enumerate `pack_int_id` values directly
midx-write.c: extract `fill_pack_from_midx()`
midx-write.c: introduce `midx_pack_perm()` helper
midx: do not require packs to be sorted in lexicographic order
midx-write.c: introduce `struct write_midx_opts`
midx-write.c: don't use `pack_perm` when assigning `bitmap_pos`
t/t5319-multi-pack-index.sh: fix copy-and-paste error in t5319.39
git-multi-pack-index(1): align SYNOPSIS with 'git multi-pack-index -h'
git-multi-pack-index(1): remove non-existent incompatibility
builtin/multi-pack-index.c: make '--progress' a common option
midx: introduce `midx_get_checksum_hex()`
midx: rename `get_midx_checksum()` to `midx_get_checksum_hash()`
midx: mark `get_midx_checksum()` arguments as const
Use subcommand-specific usage arrays for "git repo info" and
"git repo structure" so that each command shows only its own
synopsis in help output.
Add tests to cover the subcommand help behavior.
Signed-off-by: Mahi Kassa <mahlet.takassa@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Factor the "git repo info" and "git repo structure" usage
strings into shared macros so they can be reused in multiple
usage arrays.
This is a preparatory refactoring for subsequent changes to
subcommand-specific help output.
Signed-off-by: Mahi Kassa <mahlet.takassa@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The test added by 71f4960b91 (t0061: fix test for argv[0] with spaces
(MINGW only), 2019-10-01) checks that we can use a .bat file with spaces
as GIT_SSH.
This is a good test in the sense that it's how the original bug was
detected. And as the commit message there describes, there are some
elements of the bug that are likely to come up with GIT_SSH and not
elsewhere: namely that in addition to the .bat file having spaces, we
must pass an argument with spaces (which happens naturally with ssh,
since we pass the upload-pack shell command for the other side to run).
But using GIT_SSH does complicate matters:
1. We actually run the ssh command _twice_, once to probe the ssh
variant with "-G" in fill_ssh_args(), and then a second time to
actually make the connection. So we have to account for that when
checking the output.
2. Our fake ssh .bat file does not actually run ssh. So we expect the
command to fail, but not before the .bat file has touched the "out"
marker file that tells us it has run.
This works now, but is fragile. In particular, the .bat file by
default will echo commands it runs to stdout. From the perspective
of the parent Git process, this is protocol-breaking garbage, and
upon seeing it will die().
That is OK for now because we don't bother to do any cleanup of the
child process. But there is a patch under discussion, dd3693eb08
(transport-helper, connect: use clean_on_exit to reap children on
abnormal exit, 2026-03-12), which causes us to kill() the .bat
process. This happens before it actually touches the "out" file,
causing the test to fail.
We can simplify this by just using the "test-tool run-command" helper.
That lets us run whatever command we like with the arguments we want.
The argument here has a space, which is enough to trigger the original
bug that 71f4960b91 was testing. I verified that by reverting eb7c786314
(mingw: support spawning programs containing spaces in their names,
2019-07-16), the original fix, and confirming that the test fails (but
succeeds without the revert).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Code paths that loop over another array to push each element into a
strvec have been rewritten to use strvec_pushv() instead.
* rs/use-strvec-pushv:
use strvec_pushv() to add another strvec
Improve the recently introduced `git format-patch
--commit-list-format` (formerly `--cover-letter-format`) option,
including a new "modern" preset and better CLI ergonomics.
* mf/format-patch-commit-list-format:
format-patch: --commit-list-format without prefix
format-patch: add preset for --commit-list-format
format-patch: wrap generate_commit_list_cover()
format.commitListFormat: strip meaning from empty
docs/pretty-formats: add %(count) and %(total)
format-patch: rename --cover-letter-format option
format-patch: refactor generate_commit_list_cover
pretty.c: better die message %(count) and %(total)
Various updates to contrib/diff-highlight, including documentation
updates, test improvements, and color configuration handling.
* jk/diff-highlight-more:
diff-highlight: fetch all config with one process
diff-highlight: allow module callers to pass in color config
diff-highlight: test color config
diff-highlight: use test_decode_color in tests
t: add matching negative attributes to test_decode_color
diff-highlight: check diff-highlight exit status in tests
diff-highlight: drop perl version dependency back to 5.8
diff-highlight: mention build instructions
The HTTP transport learned to react to "429 Too Many Requests".
* vp/http-rate-limit-retries:
http: add support for HTTP 429 rate limit retries
strbuf_attach: fix call sites to pass correct alloc
strbuf: pass correct alloc to strbuf_attach() in strbuf_reencode()
git-replay(1) doesn't allow replaying commits all the way down to the
root commit. Fix that.
Signed-off-by: Toon Claes <toon@iotcl.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a coccinelle rule to break the build when "struct strbuf" gets
passed by value.
* dd/cocci-do-not-pass-strbuf-by-value:
stash: do not pass strbuf by value
coccinelle: detect struct strbuf passed by value
"git diff -U<num>" was too lenient in its command line parsing and
took an empty string as a valid <num>.
* ty/doc-diff-u-wo-number:
diff: document -U without <n> as using default context
Reduce system overhead "git upload-pack" spends on relaying "git
pack-objects" output to the "git fetch" running on the other end of
the connection.
* ps/upload-pack-buffer-more-writes:
builtin/pack-objects: reduce lock contention when writing packfile data
csum-file: drop `hashfd_throughput()`
csum-file: introduce `hashfd_ext()`
sideband: use writev(3p) to send pktlines
wrapper: introduce writev(3p) wrappers
compat/posix: introduce writev(3p) wrapper
upload-pack: reduce lock contention when writing packfile data
upload-pack: prefer flushing data over sending keepalive
upload-pack: adapt keepalives based on buffering
upload-pack: fix debug statement when flushing packfile data
The final clean-up phase of the diff output could turn the result of
histogram diff algorithm suboptimal, which has been corrected.
* yc/histogram-hunk-shift-fix:
xdiff: re-diff shifted change groups when using histogram algorithm
"git history" learned the "split" subcommand.
* ps/history-split:
builtin/history: implement "split" subcommand
builtin/history: split out extended function to create commits
cache-tree: allow writing in-memory index as tree
add-patch: allow disabling editing of hunks
add-patch: add support for in-memory index patching
add-patch: remove dependency on "add-interactive" subsystem
add-patch: split out `struct interactive_options`
add-patch: split out header from "add-interactive.h"
"git fast-import" learned to optionally replace signature on
commits whose signatures get invalidated due to replaying by
signing afresh.
* jt/fast-import-sign-again:
fast-import: add mode to sign commits with invalid signatures
gpg-interface: allow sign_buffer() to use default signing key
commit: remove unused forward declaration
Add and apply a semantic patch that simplifies the code by letting
strvec_pushv() append the items of a second strvec instead of pushing
them one by one.
Suggested-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: René Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The topic moves the coccinelle rules from contrib/ to tools/
directory, breaking merges with this topic.
* ps/build-tweaks:
meson: precompile "git-compat-util.h"
meson: compile compatibility sources separately
git-compat-util.h: move warning infra to prepare for PCHs
builds: move build scripts into "tools/"
contrib: move "update-unicode.sh" script into "tools/"
contrib: move "coverage-diff.sh" script into "tools/"
contrib: move "coccinelle/" directory into "tools/"
Introduce new "tools/" directory
The `timestamp_t` type is declared as `uintmax_t` and thus typically has
64 bits of precision. Usually, the full precision of such dates is not
required: it would be comforting to know that Git is still around in
millions of years, but all in all the chance is rather low.
We abuse this fact in the commit-graph: instead of storing the full 64
bits of precision, committer dates only store 34 bits. This is still
plenty of headroom, as it means that we can represent dates until year
2514. Commits which are dated beyond that year will simply get a date
whose remaining bits are masked.
The result of this is somewhat curious: the committer date will be
different depending on whether a commit gets parsed via the commit-graph
or via the object database. This isn't really too much of an issue in
general though, as we don't typically use the date parsed from the
commit-graph in user-facing output.
But with 024b4c9697 (commit: make `repo_parse_commit_no_graph()` more
robust, 2026-02-16) it started to become a problem when writing the
commit-graph itself. This commit changed `repo_parse_commit_no_graph()`
so that we re-parse the commit via the object database in case it was
already parsed beforehand via the commit-graph.
The consequence is that we may now act with two different commit dates
at different stages:
- Initially, we use the 34-bit precision timestamp when writing the
chunk generation data. We thus correctly compute the offsets
relative to the on-disk timestamp here.
- Later, when writing the overflow data, we may end up with the
full-precision timestamp. When the date is larger than 34 bits the
result of this is an underflow when computing the offset.
This causes a mismatch in the number of generation data overflow records
we want to write, and that ultimately causes Git to die.
Introduce a new helper function that computes the generation offset for
a commit while correctly masking the date to 34 bits. This makes the
previously-implicit assumptions about the commit date precision explicit
and thus hopefully less fragile going forward.
Adapt sites that compute the offset to use the function.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a remote helper like git-remote-http is invoked outside of a
repository (for example, by running git ls-remote in a non-git
directory), setup_git_directory_gently() leaves the_hash_algo
uninitialized as NULL.
If the user has a globally configured fetch refspec, remote-curl
attempts to parse it during initialization. Inside parse_refspec(),
it checks whether the LHS of the refspec is an exact OID by evaluating
llen == the_hash_algo->hexsz. Because the_hash_algo is NULL, this
results in a segmentation fault.
In 9e89dcb66a (builtin/ls-remote: fall back to SHA1 outside of a repo,
2024-08-02), we added a workaround to ls-remote to fall back to the
default hash algorithm to prevent exactly this type of crash when
parsing refspec capabilities. However, because remote-curl runs as a
separate process, it does not inherit that fallback and crashes anyway.
Instead of pushing a NULL-guard workaround down into parse_refspec(),
fix this by mirroring the ls-remote workaround directly in
remote-curl.c. If we are operating outside a repository, initialize
the_hash_algo to GIT_HASH_DEFAULT. This keeps the HTTP transport
consistent with non-HTTP transports that execute in-process, preventing
crashes without altering the generic refspec parsing logic.
Reported-by: Jo Liss <joliss@gmail.com>
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: K Jayatheerth <jayatheerthkulkarni2005@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Having to prefix a custom format-string with "log:" when passed from the
CLI can be annoying. It would be great if this prefix wasn't required.
Teach make_cover_letter() to accept custom format-strings without the
"log:" prefix if a placeholder is detected.
Note that both here and in "git log --format" the check is done naively
by just checking for the presence of a '%'.
Signed-off-by: Mirko Faina <mroik@delayed.space>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git format-patch --commit-list-format" enables the user to make their
own format for the commit list in the cover letter. It would be nice to
have a ready to use format to replace shortlog.
Teach make_cover_letter() the "modern" format preset.
This new format is the same as: "log:[%(count)/%(total)] %s".
Signed-off-by: Mirko Faina <mroik@delayed.space>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
While most conventions should not allow for the text lines in commit
messages to get too long, when they do it could make emails harder to
read.
Teach generate_commit_list_cover() to wrap its commit lines if they are
too long.
Signed-off-by: Mirko Faina <mroik@delayed.space>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
