Junio C Hamano [Mon, 9 Mar 2026 22:15:05 +0000 (15:15 -0700)]
SubmittingPatches: spell out "replace fully to pretend to be perfect"
It unfortunately is a recurring theme that new developers tend to
pile more "fixup" patches on top of the already reviewed patches,
making the topic longer and keeping the history of all wrong turns,
which interests nobody in the larger picture. Even picking a narrow
search in the list archive for "pretend to be a perfect " substring,
we find these:
The SubmittingPatches guide does talk about going incremental once a
topic hits the 'next' branch, but it does not say much about how a
new iteration of the topic should be prepared before that happens,
and it does not mention that the developers are encouraged to seize
the opportunity to pretend to be perfect with a full replacement set
of patches.
Add a new paragraph to stress this point in the section that
describes the life-cycle of a patch series.
"git log --graph --stat" did not count the display width of colored
graph part of its own output correctly, which has been corrected.
* lp/diff-stat-utf8-display-width-fix:
t4052: test for diffstat width when prefix contains ANSI escape codes
diff: handle ANSI escape codes in prefix when calculating diffstat width
Junio C Hamano [Mon, 9 Mar 2026 21:36:55 +0000 (14:36 -0700)]
Merge branch 'ar/run-command-hook-take-2'
Use the hook API to replace ad-hoc invocation of hook scripts via
the run_command() API.
* ar/run-command-hook-take-2:
builtin/receive-pack: avoid spinning no-op sideband async threads
receive-pack: convert receive hooks to hook API
receive-pack: convert update hooks to new API
run-command: poll child input in addition to output
hook: add jobs option
reference-transaction: use hook API instead of run-command
transport: convert pre-push to hook API
hook: allow separate std[out|err] streams
hook: convert 'post-rewrite' hook in sequencer.c to hook API
hook: provide stdin via callback
run-command: add stdin callback for parallelization
run-command: add helper for pp child states
t1800: add hook output stream tests
Junio C Hamano [Mon, 9 Mar 2026 20:07:50 +0000 (13:07 -0700)]
Merge branch 'ar/config-hooks' into ar/config-hook-cleanups
* ar/config-hooks: (21 commits)
builtin/receive-pack: avoid spinning no-op sideband async threads
hook: add -z option to "git hook list"
hook: allow out-of-repo 'git hook' invocations
hook: allow event = "" to overwrite previous values
hook: allow disabling config hooks
hook: include hooks from the config
hook: add "git hook list" command
hook: run a list of hooks to prepare for multihook support
hook: add internal state alloc/free callbacks
receive-pack: convert receive hooks to hook API
receive-pack: convert update hooks to new API
run-command: poll child input in addition to output
hook: add jobs option
reference-transaction: use hook API instead of run-command
transport: convert pre-push to hook API
hook: allow separate std[out|err] streams
hook: convert 'post-rewrite' hook in sequencer.c to hook API
hook: provide stdin via callback
run-command: add stdin callback for parallelization
run-command: add helper for pp child states
...
Tian Yuchen [Mon, 9 Mar 2026 06:51:40 +0000 (14:51 +0800)]
patch-ids: document intentional const-casting in patch_id_neq()
The hashmap API requires the comparison function to take const pointers.
However, patch_id_neq() uses lazy evaluation to compute patch IDs on
demand. As established in b3dfeebb (rebase: avoid computing unnecessary
patch IDs, 2016-07-29), this avoids unnecessary work since not all
objects in the hashmap will eventually be compared.
Remove the ten-year-old "NEEDSWORK" comment and formally document
this intentional design trade-off.
Signed-off-by: Tian Yuchen <cat@malon.dev> Signed-off-by: Junio C Hamano <gitster@pobox.com>
René Scharfe [Sun, 8 Mar 2026 09:57:02 +0000 (10:57 +0100)]
history: initialize rev_info in cmd_history_reword()
git history reword expects a single valid revision argument and errors
out if it doesn't get it. In that case the struct rev_info passed to
release_revisions() for cleanup is still uninitialized, which can result
in attempts to free(3) random pointers. Avoid that by initializing the
structure.
Signed-off-by: René Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Collin Funk [Mon, 9 Mar 2026 02:55:11 +0000 (19:55 -0700)]
bloom: remove a misleading const qualifier
When building with glibc-2.43 there is the following warning:
bloom.c: In function ‘get_or_compute_bloom_filter’:
bloom.c:515:52: warning: initialization discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
515 | char *last_slash = strrchr(path, '/');
| ^~~~~~~
In this case, we always write through "path" through the "last_slash"
pointer. Therefore, the const qualifier on "path" is misleading and we
can just remove it.
Signed-off-by: Collin Funk <collin.funk1@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
t3310: avoid hiding failures from rev-parse in command substitutions
Running `git` commands inside command substitutions like
test "$(git rev-parse A)" = "$(git rev-parse B)"
can hide failures from the `git` invocations and provide little
diagnostic information when `test` fails.
Use `test_cmp` when comparing against a stored expected value so
mismatches show both expected and actual output. Use `test_cmp_rev`
when comparing two revisions. These helpers produce clearer failure
output, making it easier to understand what went wrong.
Suggested-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Francesco Paparatto <francescopaparatto@gmail.com> Reviewed-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Omri Sarig [Sat, 7 Mar 2026 17:08:01 +0000 (17:08 +0000)]
doc: make it easier to find custom command information
Git supports creating additional commands through aliases, and through
placement of executables with a "git-" prefix in the PATH.
This information was not easy enough to find - users will look for this
information around the command description, but the documentation
exists in other locations.
Update the "GIT COMMANDS" section to reference the relevant sections,
making it easier for to find this information.
Signed-off-by: Omri Sarig <omri.sarig13@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jeff King [Fri, 6 Mar 2026 16:25:13 +0000 (11:25 -0500)]
meson: turn on NO_MMAP when building with LSan
The previous commit taught the Makefile to turn on NO_MMAP in this
instance. We should do the same with meson for consistency. We already
do this for ASan builds, so we can just tweak one conditional.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jeff King [Thu, 5 Mar 2026 23:13:05 +0000 (18:13 -0500)]
Makefile: turn on NO_MMAP when building with LSan
The past few commits fixed some cases where we leak memory allocated by
mmap(). Building with SANITIZE=leak doesn't detect these because it
covers only heap buffers allocated by malloc().
But if we build with NO_MMAP, our compat mmap() implementation will
allocate a heap buffer and pread() into it. And thus Lsan will detect
these leaks for free.
Using NO_MMAP is less performant, of course, since we have to use extra
memory and read in the whole file, rather than faulting in pages from
disk. But LSan builds are already slow, and this doesn't make them
measurably worse. Getting extra coverage for our leak-checking is worth
it.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jeff King [Sat, 7 Mar 2026 02:24:59 +0000 (21:24 -0500)]
object-file: fix mmap() leak in odb_source_loose_read_object_stream()
We mmap() a loose object file, storing the result in the local variable
"mapped", which is eventually assigned into our stream struct as
"st.mapped". If we hit an error, we jump to an error label which does:
munmap(st.mapped, st.mapsize);
to clean up. But this is wrong; we don't assign st.mapped until the end
of the function, after all of the "goto error" jumps. So this munmap()
is never cleaning up anything (st.mapped is always NULL, because we
initialize the struct with calloc).
Instead, we should feed the local variable to munmap().
This leak is due to 595296e124 (streaming: allocate stream inside the
backend-specific logic, 2025-11-23), which introduced the local
variable. Before that, we assigned the mmap result directly into
st.mapped. It was probably switched there so that we do not have to
allocate/free the struct when the map operation fails (e.g., because we
don't have the loose object). Before that commit, the struct was passed
in from the caller, so there was no allocation at all.
You can see the leak in the test suite by building with:
make SANITIZE=leak NO_MMAP=1 CC=clang
and running t1060. We need NO_MMAP so that the mmap() is backed by an
actual malloc(), which allows LSan to detect it. And the leak seems not
to be detected when compiling with gcc, probably due to some internal
compiler decisions about how the stack memory is written.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Mirko Faina [Fri, 6 Mar 2026 23:34:43 +0000 (00:34 +0100)]
format-patch: add commitListFormat config
Using "--cover-letter" we can tell format-patch to generate a cover
letter, in this cover letter there's a list of commits included in the
patch series and the format is specified by the "--cover-letter-format"
option. Would be useful if this format could be configured from the
config file instead of always needing to pass it from the command line.
Teach format-patch how to read the format spec for the cover letter from
the config files. The variable it should look for is called
format.commitListFormat.
Possible values:
- commitListFormat is set but no string is passed: it will default to
"[%(count)/%(total)] %s"
- if a string is passed: will use it as a format spec. Note that this
is either "shortlog" or a format spec prefixed by "log:"
e.g."log:%s (%an)"
- if commitListFormat is not set: it will default to the shortlog
format.
Signed-off-by: Mirko Faina <mroik@delayed.space> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Mirko Faina [Fri, 6 Mar 2026 23:34:42 +0000 (00:34 +0100)]
format-patch: add ability to use alt cover format
Often when sending patch series there's a need to clarify to the
reviewer what's the purpose of said series, since it might be difficult
to understand it from reading the commits messages one by one.
"git format-patch" provides the useful "--cover-letter" flag to declare
if we want it to generate a template for us to use. By default it will
generate a "git shortlog" of the changes, which developers find less
useful than they'd like, mainly because the shortlog groups commits by
author, and gives no obvious chronological order.
Give format-patch the ability to specify an alternative format spec
through the "--cover-letter-format" option. This option either takes
"shortlog", which is the current format, or a format spec prefixed with
"log:".
Mirko Faina [Fri, 6 Mar 2026 23:34:41 +0000 (00:34 +0100)]
format-patch: move cover letter summary generation
As of now format-patch allows generation of a template cover letter for
patch series through "--cover-letter".
Move shortlog summary code generation to its own function. This is done
in preparation to other patches where we enable the user to format the
commit list using thier own format string.
Signed-off-by: Mirko Faina <mroik@delayed.space> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Mirko Faina [Fri, 6 Mar 2026 23:34:40 +0000 (00:34 +0100)]
pretty.c: add %(count) and %(total) placeholders
In many commands we can customize the output through the "--format" or
the "--pretty" options. This patch adds two new placeholders used mainly
when there's a range of commits that we want to show.
Currently these two placeholders are not usable as they're coupled with
the rev_info->nr and rev_info->total fields, fields that are used only
by the format-patch numbered email subjects.
Teach repo_format_commit_message() the %(count) and %(total)
placeholders.
Signed-off-by: Mirko Faina <mroik@delayed.space> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Justin Tobler [Thu, 5 Mar 2026 19:38:36 +0000 (13:38 -0600)]
Documentation: extend guidance for submitting patches
Before submitting patches on the mailing list, it is often a good idea
to check for previous related discussions or if similar work is already
in progress. This enables better coordination amongst contributors and
could avoid duplicating work.
Additionally, it is often recommended to give reviewers some time to
reply to a patch series before sending new versions. This helps collect
broader feedback and reduces unnecessary churn from rapid rerolls.
Document this guidance in "Documentation/SubmittingPatches" accordingly.
Signed-off-by: Justin Tobler <jltobler@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jeff King [Thu, 5 Mar 2026 23:12:29 +0000 (18:12 -0500)]
pack-revindex: avoid double-loading .rev files
The usual entry point for loading the pack revindex is the
load_pack_revindex() function. It returns immediately if the packed_git
has a non-NULL revindex or revindex data field (representing an
in-memory or mmap'd .rev file, respectively), since the data is already
loaded.
But in 5a6072f631 (fsck: validate .rev file header, 2023-04-17) the fsck
code path switched to calling load_pack_revindex_from_disk() directly,
since it wants to check the on-disk data (if there is any). But that
function does _not_ check to see if the data has already been loaded; it
just maps the file, overwriting the revindex_map pointer (and pointing
revindex_data inside that map). And in that case we've leaked the mmap()
pointed to by revindex_map (if it was non-NULL).
This usually doesn't happen, since fsck wouldn't need to load the
revindex for any reason before we get to these checks. But there are
some cases where it does. For example, is_promisor_object() runs
odb_for_each_object() with the PACK_ORDER flag, which uses the revindex.
This happens a few times in our test suite, but SANITIZE=leak doesn't
detect it because we are leaking an mmap(), not a heap-allocated buffer
from malloc(). However, if you build with NO_MMAP, then our compat mmap
will read into a heap buffer instead, and LSan will complain. This
causes failures in t5601, t0410, t5702, and t5616.
We can fix it by checking for existing revindex_data when loading from
disk. This is redundant when we're called from load_pack_revindex(), but
it's a cheap check. The alternative is to teach check_pack_rev_indexes()
in fsck to skip the load, but that seems messier; it doesn't otherwise
know about internals like revindex_map and revindex_data.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jeff King [Thu, 5 Mar 2026 23:09:56 +0000 (18:09 -0500)]
check_connected(): fix leak of pack-index mmap
Since c6807a40dc (clone: open a shortcut for connectivity check,
2013-05-26), we may open a one-off packed_git struct to check what's in
the pack we just received. At the end of the function we throw away the
struct (rather than linking it into the repository struct as usual).
We used to leak the struct until dd4143e7bf (connected.c: free the
"struct packed_git", 2022-11-08), which calls free(). But that's not
sufficient; inside the struct we'll have mmap'd the pack idx data from
disk, which needs an munmap() call.
Building with SANITIZE=leak doesn't detect this, because we are leaking
our own mmap(), and it only finds heap allocations from malloc(). But if
we use our compat mmap implementation like this:
make NO_MMAP=MapsBecomeMallocs SANITIZE=leak
then LSan will notice the leak, because now it's a regular heap buffer
allocated by malloc().
We can fix it by calling close_pack(), which will free any associated
memory. Note that we need to check for NULL ourselves; unlike free(), it
is not safe to pass a NULL pointer to close_pack().
Signed-off-by: Jeff King <peff@peff.net> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jeff King [Thu, 5 Mar 2026 23:08:54 +0000 (18:08 -0500)]
check_connected(): delay opening new_pack
In check_connected(), if the transport tells us we got a single packfile
that has already been verified as self-contained and connected, then we
can skip checking connectivity for any tips that are mentioned in that
pack. This goes back to c6807a40dc (clone: open a shortcut for
connectivity check, 2013-05-26).
We don't need to open that pack until we are about to start sending oids
to our child rev-list process, since that's when we check whether they
are in the self-contained pack. Let's push the opening of that pack
further down in the function. That saves us from having to clean it up
when we leave the function early (and by the time have opened the
rev-list process, we never leave the function early, since we have to
clean up the child process).
Signed-off-by: Jeff King <peff@peff.net> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Li Chen [Fri, 6 Mar 2026 14:53:32 +0000 (14:53 +0000)]
rebase: support --trailer
Add a new --trailer=<trailer> option to git rebase to append trailer
lines to each rewritten commit message (merge backend only).
Because the apply backend does not provide a commit-message filter,
reject --trailer when --apply is in effect and require the merge backend
instead.
This option implies --force-rebase so that fast-forwarded commits are
also rewritten. Validate trailer arguments early to avoid starting an
interactive rebase with invalid input.
Add integration tests covering error paths and trailer insertion across
non-interactive and interactive rebases.
Signed-off-by: Li Chen <me@linux.beauty> Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Li Chen [Fri, 6 Mar 2026 14:53:31 +0000 (14:53 +0000)]
commit, tag: parse --trailer with OPT_STRVEC
Now that amend_file_with_trailers() expects raw trailer lines, do not
store argv-style "--trailer=<trailer>" strings in git commit and git
tag.
Parse --trailer using OPT_STRVEC so trailer_args contains only the
trailer value, and drop the temporary prefix stripping in
amend_file_with_trailers().
Signed-off-by: Li Chen <me@linux.beauty> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Li Chen [Fri, 6 Mar 2026 14:53:30 +0000 (14:53 +0000)]
trailer: append trailers without fork/exec
Introduce amend_strbuf_with_trailers() to apply trailer additions to a
message buffer via process_trailers(), avoiding the need to run git
interpret-trailers as a child process.
Update amend_file_with_trailers() to use the in-process helper and
rewrite the target file via tempfile+rename, preserving the previous
in-place semantics. As the trailers are no longer added in a separate
process and trailer_config_init() die()s on missing config values it
is called early on in cmd_commit() and cmd_tag() so that they die()
early before writing the message file. The trailer arguments are now
also sanity checked.
Keep existing callers unchanged by continuing to accept argv-style
--trailer=<trailer> entries and stripping the prefix before feeding the
in-process implementation.
Signed-off-by: Li Chen <me@linux.beauty> Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Li Chen [Fri, 6 Mar 2026 14:53:29 +0000 (14:53 +0000)]
trailer: libify a couple of functions
Move create_in_place_tempfile() and process_trailers() from
builtin/interpret-trailers.c into trailer.c and expose it via trailer.h.
This reverts most of ae0ec2e0e0b (trailer: move interpret_trailers()
to interpret-trailers.c, 2024-03-01) and lets other call sites reuse
the same trailer rewriting logic.
Signed-off-by: Li Chen <me@linux.beauty> Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refactor create_in_place_tempfile() in preparation for moving it
to tralier.c. Change the return type to return a `struct tempfile*`
instead of a `FILE*` so that we can remove the file scope tempfile
variable. Since 076aa2cbda5 (tempfile: auto-allocate tempfiles on
heap, 2017-09-05) it has not been necessary to make tempfile varibales
static so this is safe. Also use error() and return NULL in place of
die() so the caller can exit gracefully and use find_last_dir_sep()
rather than strchr() to find the parent directory.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Junio C Hamano <gitster@pobox.com>
odb/source: make `read_object_info()` function pluggable
Introduce a new callback function in `struct odb_source` to make the
function pluggable.
Note that this function is a bit less straight-forward to convert
compared to the other functions. The reason here is that the logic to
read an object is:
1. We try to read the object. If it exists we return it.
2. If the object does not exist we reprepare the object database
source.
3. We then try reading the object info a second time in case the
reprepare caused it to appear.
The second read is only supposed to happen for the packfile store
though, as reading loose objects is not impacted by repreparing the
object database.
Ideally, we'd just move this whole logic into the ODB source. But that's
not easily possible because we try to avoid the reprepare unless really
required, which is after we have found out that no other ODB source
contains the object, either. So the logic spans across multiple ODB
sources, and consequently we cannot move it into an individual source.
Instead, introduce a new flag `OBJECT_INFO_SECOND_READ` that tells the
backend that we already tried to look up the object once, and that this
time around the ODB source should try to find any new objects that may
have surfaced due to an on-disk change.
With this flag, the "files" backend can trivially skip trying to re-read
the object as a loose object. Furthermore, as we know that we only try
the second read via the packfile store, we can skip repreparing loose
objects and only reprepare the packfile store.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a caller holds a `struct odb_source`, they have no way of telling
what type the source is. This doesn't really cause any problems in the
current status quo as we only have a single type anyway, "files". But
going forward we expect to add more types, and if so it will become
necessary to tell the sources apart.
Introduce a new enum to cover this use case and assert that the given
source actually matches the target source when performing the downcast.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
odb: move reparenting logic into respective subsystems
The primary object database source may be initialized with a relative
path. When the process changes its current working directory we thus
have to update this path and have it point to the same path, but
relative to the new working directory.
This logic is handled in the object database layer. It consists of three
steps:
1. We undo any potential temporary object directory, which are used
for transactions. This is done so that we don't end up modifying
the temporary object database source that got applied for the
transaction.
2. We then iterate through the non-transactional sources and reparent
their respective paths.
3. We reapply the temporary object directory, but update its path.
All of this logic is heavily tied to how the object database source
handles paths in the first place. It's an internal implementation
detail, and as sources may not even use an on-disk path at all it is not
a mechanism that applies to all potential sources.
Refactor the code so that the logic to reparent the sources is hosted by
the "files" source and the temporary object directory subsystems,
respectively. This logic is easier to reason about, but it also ensures
that this logic is handled at the correct level.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The "files" backend is implemented as a pointer in the `struct
odb_source`. This contradicts our typical pattern for pluggable backends
like we use it for example in the ref store or for object database
streams, where we typically embed the generic base structure in the
specialized implementation. This pattern has a couple of small benefits:
- We avoid an extra allocation.
- We hide implementation details in the generic structure.
- We can easily downcast from a generic backend to the specialized
structure and vice versa because the offsets are known at compile
time.
- It becomes trivial to identify locations where we depend on backend
specific logic because the cast needs to be explicit.
Refactor our "files" object database source to do the same and embed the
`struct odb_source` in the `struct odb_source_files`.
There are still a bunch of sites in our code base where we do have to
access internals of the "files" backend. The intent is that those will
go away over time, but this will certainly take a while. Meanwhile,
provide a `odb_source_files_downcast()` function that can convert a
generic source into a "files" source.
As we only have a single source the downcast succeeds unconditionally
for now. Eventually though the intent is to make the cast `BUG()` in
case the caller requests to downcast a non-"files" backend to a "files"
backend.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce a new "files" object database source. This source encapsulates
access to both loose object files and the packfile store, similar to how
the "files" backend for refs encapsulates access to loose refs and the
packed-refs file.
Note that for now the "files" source is still a direct member of a
`struct odb_source`. This architecture will be reversed in the next
commit so that the files source contains a `struct odb_source`.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
odb: split `struct odb_source` into separate header
Subsequent commits will expand the `struct odb_source` to become a
generic interface for accessing an object database source. As part of
these refactorings we'll add a set of function pointers that will
significantly expand the structure overall.
Prepare for this by splitting out the `struct odb_source` into a
separate header. This keeps the high-level object database interface
detached from the low-level object database sources.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move the setup logic into a 'test_expect_success' block.
This ensures that the code is properly tracked by the test harness.
Additionally, we use the 'test_when_finished' helper at the start of
the block to ensure that the 'import' directory is removed even if the
test fails.
This is cleaner than the previous manual 'rm -rf import' approach.
Signed-off-by: Siddharth Shrimali <r.siddharth.shrimali@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Aditya Garg [Thu, 5 Mar 2026 17:01:15 +0000 (17:01 +0000)]
send-email: pass smtp hostname and port to Authen::SASL
Starting from version 2.2000, Authen::SASL supports passing the SMTP
server hostname and port to the OAUTHBEARER string passed via SMTP AUTH.
Add support for the same in git-send-email.
It's safe to add the new parameters unconditionally as older versions of
Authen::SASL will simply ignore them without any error. Something
similar is already being done for the authname parameter, which is not
supported by every authentication mechanism. This can be understood as
declaring a variable but not using at all.
meson: detect broken iconv that requires ICONV_RESTART_RESET
In d0cec08d70 (utf8.c: prepare workaround for iconv under macOS 14/15,
2026-01-12) we have introduced a new workaround for a broken version of
libiconv on macOS. This workaround has for now only been wired up for
our Makefile, so using Meson with such a broken version will fail.
We can rather easily detect the broken behaviour. Some encodings have
different modes that can be switched to via an escape sequence. In the
case of ISO-2022-JP this can be done via "<Esc>$B" and "<Esc>(J" to
switch between ASCII and JIS modes. The bug now triggers when one does
multiple calls to iconv(3p) to convert a string piece by piece, where
the first call enters JIS mode. The second call forgets about the fact
that it is still in JIS mode, and consequently it will incorrectly treat
the input as ASCII, and thus the produced output is of course garbage.
Wire up a test that exercises this in Meson and, if it fails, set the
`ICONV_RESTART_RESET` define.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Seyi Kufoiji [Thu, 5 Mar 2026 10:05:26 +0000 (11:05 +0100)]
builtin/rev-list: migrate missing_objects cleanup to oidmap_clear_with_free()
As part of the conversion away from oidmap_clear(), switch the
missing_objects map to use oidmap_clear_with_free().
missing_objects stores struct missing_objects_map_entry instances,
which own an xstrdup()'d path string in addition to the container
struct itself. Previously, rev-list manually freed entry->path
before calling oidmap_clear(&missing_objects, true).
Introduce a dedicated free callback and pass it to
oidmap_clear_with_free(), consolidating entry teardown into a
single place and making cleanup semantics explicit.
Signed-off-by: Seyi Kuforiji <kuforiji98@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Seyi Kufoiji [Thu, 5 Mar 2026 10:05:25 +0000 (11:05 +0100)]
oidmap: make entry cleanup explicit in oidmap_clear
Replace oidmap's use of hashmap_clear_() and layout-dependent freeing
with an explicit iteration and optional free callback. This removes
reliance on struct layout assumptions while keeping the existing API
intact.
Add tests for oidmap_clear_with_free behavior.
test_oidmap__clear_with_free_callback verifies that entries are freed
when a callback is provided, while
test_oidmap__clear_without_free_callback verifies that entries are not
freed when no callback is given. These tests ensure the new clear
implementation behaves correctly and preserves ownership semantics.
Signed-off-by: Seyi Kuforiji <kuforiji98@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
"fsck" iterates over packfiles and its access to pack data caused
the list to be permuted, which caused it to loop forever; the code
to access pack data by "fsck" has been updated to avoid this.
* ps/fsck-stream-from-the-right-object-instance:
pack-check: fix verification of large objects
packfile: expose function to read object stream for an offset
object-file: adapt `stream_object_signature()` to take a stream
t/helper: improve "genrandom" test helper
The core.attributesfile is intended to be set per repository, but
were kept track of by a single global variable in-core, which has
been corrected by moving it to per-repository data structure.
* ob/core-attributesfile-in-repository:
environment: move "branch.autoSetupMerge" into `struct repo_config_values`
environment: stop using core.sparseCheckout globally
environment: stop storing `core.attributesFile` globally
Junio C Hamano [Wed, 4 Mar 2026 18:53:02 +0000 (10:53 -0800)]
Merge branch 'ds/config-list-with-type'
"git config list" is taught to show the values interpreted for
specific type with "--type=<X>" option.
* ds/config-list-with-type:
config: use an enum for type
config: restructure format_config()
config: format colors quietly
color: add color_parse_quietly()
config: format expiry dates quietly
config: format paths gently
config: format bools or strings in helper
config: format bools or ints gently
config: format bools gently
config: format int64s gently
config: make 'git config list --type=<X>' work
config: add 'gently' parameter to format_config()
config: move show_all_config()
Mark the marge-ort codebase to prevent more uses of the_repository
from getting added.
* en/merge-ort-almost-wo-the-repository:
replay: prevent the_repository from coming back
merge-ort: prevent the_repository from coming back
merge-ort: replace the_hash_algo with opt->repo->hash_algo
merge-ort: replace the_repository with opt->repo
merge-ort: pass repository to write_tree()
merge,diff: remove the_repository check before prefetching blobs
Junio C Hamano [Wed, 4 Mar 2026 18:53:01 +0000 (10:53 -0800)]
Merge branch 'lo/repo-leftover-bits'
Clean-up the code around "git repo info" command.
* lo/repo-leftover-bits:
Documentation/git-repo: capitalize format descriptions
Documentation/git-repo: replace 'NUL' with '_NUL_'
t1901: adjust nul format output instead of expected value
t1900: rename t1900-repo to t1900-repo-info
repo: rename struct field to repo_info_field
repo: replace get_value_fn_for_key by get_repo_info_field
repo: rename repo_info_fields to repo_info_field
CodingGuidelines: instruct to name arrays in singular
Junio C Hamano [Wed, 4 Mar 2026 18:53:01 +0000 (10:53 -0800)]
Merge branch 'ps/maintenance-geometric-default'
"git maintenance" starts using the "geometric" strategy by default.
* ps/maintenance-geometric-default:
builtin/maintenance: use "geometric" strategy by default
t7900: prepare for switch of the default strategy
t6500: explicitly use "gc" strategy
t5510: explicitly use "gc" strategy
t5400: explicitly use "gc" strategy
t34xx: don't expire reflogs where it matters
t: disable maintenance where we verify object database structure
t: fix races caused by background maintenance
Junio C Hamano [Wed, 4 Mar 2026 18:52:59 +0000 (10:52 -0800)]
Merge branch 'rr/gitweb-mobile'
"gitweb" has been taught to be mobile friendly.
* rr/gitweb-mobile:
gitweb: let page header grow on mobile for long wrapped project names
gitweb: fix mobile footer overflow by wrapping text and clearing floats
gitweb: fix mobile page overflow across log/commit/blob/diff views
gitweb: prevent project search bar from overflowing on mobile
gitweb: add viewport meta tag for mobile devices
Junio C Hamano [Wed, 4 Mar 2026 18:52:58 +0000 (10:52 -0800)]
Merge branch 'kn/ref-location'
Allow the directory in which reference backends store their data to
be specified.
* kn/ref-location:
refs: add GIT_REFERENCE_BACKEND to specify reference backend
refs: allow reference location in refstorage config
refs: receive and use the reference storage payload
refs: move out stub modification to generic layer
refs: extract out `refs_create_refdir_stubs()`
setup: don't modify repo in `create_reference_database()`
Harald Nordgren [Wed, 4 Mar 2026 12:25:31 +0000 (12:25 +0000)]
status: clarify how status.compareBranches deduplicates
The order of output when multiple branches are specified on the
configuration variable was not clearly spelled out in the
documentation.
Add a paragraph to describe the order and also how the branches are
deduplicated. Update t6040 with additional tests to illustrate how
multiple branches are shown and deduplicated.
Signed-off-by: Harald Nordgren <haraldnordgren@gmail.com>
[jc: made a whole replacement into incremental; wrote log message.] Signed-off-by: Junio C Hamano <gitster@pobox.com>
Tian Yuchen [Wed, 4 Mar 2026 14:15:26 +0000 (22:15 +0800)]
setup: improve error diagnosis for invalid .git files
'read_gitfile_gently()' treats any non-regular file as
'READ_GITFILE_ERR_NOT_A_FILE' and fails to discern between 'ENOENT'
and other stat failures. This flawed error reporting is noted by two
'NEEDSWORK' comments.
Address these comments by introducing two new error codes:
'READ_GITFILE_ERR_MISSING'(which groups the "file missing" scenarios
together) and 'READ_GITFILE_ERR_IS_A_DIR':
1. Update 'read_gitfile_error_die()' to treat 'IS_A_DIR', 'MISSING',
'NOT_A_FILE' and 'STAT_FAILED' as non-fatal no-ops. This accommodates
intentional non-repo scenarios (e.g., GIT_DIR=/dev/null).
2. Explicitly catch 'NOT_A_FILE' and 'STAT_FAILED' during
discovery and call 'die()' if 'die_on_error' is set.
3. Unconditionally pass '&error_code' to 'read_gitfile_gently()'.
4. Only invoke 'is_git_directory()' when we explicitly receive
'READ_GITFILE_ERR_IS_A_DIR', avoiding redundant checks.
Additionally, audit external callers of 'read_gitfile_gently()' in
'submodule.c' and 'worktree.c' to accommodate the refined error codes.
Signed-off-by: Tian Yuchen <a3205153416@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
K Jayatheerth [Wed, 4 Mar 2026 13:05:02 +0000 (18:35 +0530)]
path: remove redundant function calls
repo_settings_get_shared_repository() is invoked multiple times in
calc_shared_perm(). While the function internally caches the value,
repeated calls still add unnecessary noise.
Store the result in a local variable and reuse it instead. This makes
it explicit that the value is expected to remain constant and avoids
repeated calls in the same scope.
Signed-off-by: K Jayatheerth <jayatheerthkulkarni2005@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
K Jayatheerth [Wed, 4 Mar 2026 13:05:01 +0000 (18:35 +0530)]
path: use size_t for dir_prefix length
The strlen() function returns a size_t. Storing this in a standard
signed int is a bad practice that invites overflow vulnerabilities if
paths get absurdly long.
Switch the variable to size_t. This is safe to do because 'len' is
strictly used as an argument to strncmp() (which expects size_t) and
as a positive array index, involving no signed arithmetic that could
rely on negative values.
Signed-off-by: K Jayatheerth <jayatheerthkulkarni2005@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Wolfgang Faust [Wed, 4 Mar 2026 01:30:52 +0000 (17:30 -0800)]
git-gui: grey out comment lines in commit message
Comment lines are stripped by wash_commit_message, but there is no
indication in the UI that they are special and will be removed.
Grey these lines out to indicate that they will be removed.
Signed-off-by: Wolfgang Faust <contrib-git@wolfgangfaust.com> Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Nasser Grainawi [Tue, 3 Mar 2026 23:40:44 +0000 (15:40 -0800)]
submodule: fetch missing objects from default remote
When be76c21282 (fetch: ensure submodule objects fetched, 2018-12-06)
added support for fetching a missing submodule object by id, it
hardcoded the remote name as "origin" and deferred anything more
complicated for a later patch. Implement the NEEDSWORK item to remove
the hardcoded assumption by adding and using a submodule helper subcmd
'get-default-remote'. Fixing this lets 'git fetch --recurse-submodules'
succeed when the fetched commit(s) in the superproject trigger a
submodule fetch, and that submodule's default remote name is not
"origin".
Add non-"origin" remote tests to t5526-fetch-submodules.sh and
t5572-pull-submodule.sh demonstrating this works as expected and add
dedicated tests for get-default-remote.
Signed-off-by: Nasser Grainawi <nasser.grainawi@oss.qualcomm.com> Reviewed-by: Jacob Keller <jacob.keller@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
It is quite a common use case that one wants to split up one commit into
multiple commits by moving parts of the changes of the original commit
out into a separate commit. This is quite an involved operation though:
1. Identify the commit in question that is to be dropped.
2. Perform an interactive rebase on top of that commit's parent.
3. Modify the instruction sheet to "edit" the commit that is to be
split up.
4. Drop the commit via "git reset HEAD~".
5. Stage changes that should go into the first commit and commit it.
6. Stage changes that should go into the second commit and commit it.
7. Finalize the rebase.
This is quite complex, and overall I would claim that most people who
are not experts in Git would struggle with this flow.
Introduce a new "split" subcommand for git-history(1) to make this way
easier. All the user needs to do is to say `git history split $COMMIT`.
From hereon, Git asks the user which parts of the commit shall be moved
out into a separate commit and, once done, asks the user for the commit
message. Git then creates that split-out commit and applies the original
commit on top of it.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
builtin/history: split out extended function to create commits
In the next commit we're about to introduce a new command that splits up
a commit into two. Most of the logic will be shared with rewording
commits, except that we also need to have control over the parents and
the old/new trees.
Extract a new function `commit_tree_with_edited_message_ext()` to
prepare for this commit.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The function `write_in_core_index_as_tree()` takes a repository and
writes its index into a tree object. What this function cannot do though
is to take an _arbitrary_ in-memory index.
Introduce a new `struct index_state` parameter so that the caller can
pass a different index than the one belonging to the repository. This
will be used in a subsequent commit.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The "add-patch" mode allows the user to edit hunks to apply custom
changes. This is incompatible with a new `git history split` command
that we're about to introduce in a subsequent commit, so we need a way
to disable this mode.
Add a new flag to disable editing hunks.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
add-patch: add support for in-memory index patching
With `run_add_p()` callers have the ability to apply changes from a
specific revision to a repository's index. This infra supports several
different modes, like for example applying changes to the index,
working tree or both.
One feature that is missing though is the ability to apply changes to an
in-memory index different from the repository's index. Add a new
function `run_add_p_index()` to plug this gap.
This new function will be used in a subsequent commit.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
add-patch: remove dependency on "add-interactive" subsystem
With the preceding commit we have split out interactive configuration
that is used by both "git add -p" and "git add -i". But we still
initialize that configuration in the "add -p" subsystem by calling
`init_add_i_state()`, even though we only do so to initialize the
interactive configuration as well as a repository pointer.
Stop doing so and instead store and initialize the interactive
configuration in `struct add_p_state` directly.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The `struct add_p_opt` is reused both by our infra for "git add -p" and
"git add -i". Users of `run_add_i()` for example are expected to pass
`struct add_p_opt`. This is somewhat confusing and raises the question
of which options apply to what part of the stack.
But things are even more confusing than that: while callers are expected
to pass in `struct add_p_opt`, these options ultimately get used to
initialize a `struct add_i_state` that is used by both subsystems. So we
are basically going full circle here.
Refactor the code and split out a new `struct interactive_options` that
hosts common options used by both. These options are then applied to a
`struct interactive_config` that hosts common configuration.
This refactoring doesn't yet fully detangle the two subsystems from one
another, as we still end up calling `init_add_i_state()` in the "git add
-p" subsystem. This will be fixed in a subsequent commit.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
add-patch: split out header from "add-interactive.h"
While we have a "add-patch.c" code file, its declarations are part of
"add-interactive.h". This makes it somewhat harder than necessary to
find relevant code and to identify clear boundaries between the two
subsystems.
Split up concerns and move declarations that relate to "add-patch.c"
into a new "add-patch.h" header.
Signed-off-by: Patrick Steinhardt <ps@pks.im> Signed-off-by: Junio C Hamano <gitster@pobox.com>
t3700: use test_grep helper for better diagnostics
Replace 'grep' and '! grep' invocations with 'test_grep' and
'test_grep !'. This provides better debugging output if tests fail
in the future, as 'test_grep' will automatically print the
contents of the file when a check fails.
While at it, update any remaining instances of 'grep' to 'test_grep'
that were missed in the previous versions to ensure that the entire
file is consistent with modern project style.
Suggested-by: Junio C Hamano <gitster@pobox.com> Signed-off-by: Siddharth Shrimali <r.siddharth.shrimali@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Replace pipelines involving git commands with temporary files (actual)
to ensure that any crashes or unexpected exit codes from the git
commands are properly caught by the test suite. A simple pipeline
like 'git foo | grep bar' ignores the exit code of 'git', which
can hide regressions.
In cases where we were counting lines with 'wc -l' to ensure a
pattern was absent, simplify the logic to use '! grep' to avoid
subshells entirely.
Suggested-by: Junio C Hamano <gitster@pobox.com> Signed-off-by: Siddharth Shrimali <r.siddharth.shrimali@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Johannes Sixt [Mon, 2 Mar 2026 08:14:48 +0000 (09:14 +0100)]
gitk: commit translation files without file information
File information in the translation files is only helpful for the
translators, but is not needed to compile the message catalogs. On top
of that, file information is rather volatile and leads to large patches
that do not carry essential information. For this reason, Git project
has opted to remove the file information from its translation files.
Let's do that in this project, too.
Rewrite the update-po target to generate *.po files that do contain
file information for the benefit of translators. Configure a clean
filter under the name "gettext-no-location", which is the same that
the Git project uses. It is expected that translators have already
configured their repository suitably. Nevertheless, write a reminder
as part of the update-po target.
Junio C Hamano [Tue, 3 Mar 2026 19:08:13 +0000 (11:08 -0800)]
Merge branch 'hy/diff-lazy-fetch-with-break-fix'
A prefetch call can be triggered to access a stale diff_queue entry
after diffcore-break breaks a filepair into two and freed the
original entry that is no longer used, leading to a segfault, which
has been corrected.
* hy/diff-lazy-fetch-with-break-fix:
diffcore-break: avoid segfault with freed entries
projects / thirdparty / git.git / log
