While the installed scripts have their Perl shebang set to PERL_PATH,
it is nevertheless useful to be able to run the uninstalled script for
manual tests while developing. This change makes the shebang more
portable by having the perl command looked from PATH instead of from a
fixed location.

Signed-off-by: Maxim Cournoyer <maxim@guixotic.coop>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Invalid ports were previously silently dropped; now a warning message
is produced.

Signed-off-by: Maxim Cournoyer <maxim@guixotic.coop>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

To improve support for symbolic port names in netrc files, this
changes does the following:

 - Treat symbolic port names as ports, not protocols in git-credential-netrc
 - Validate the SMTP server port provided to send-email
 - Convert the above symbolic port names to their numerical values.

Before this change, it was not possible to have a SMTP server port set
to "smtps" in a netrc file (e.g. Emacs' ~/.authinfo.gpg), as it would
be registered as a protocol and break the match for a "smtp" protocol
host, as queried for by git-send-email.

Signed-off-by: Maxim Cournoyer <maxim@guixotic.coop>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

These options were introduced in 4e51389 (builtin/config:
introduce "get" subcommand, 2024-05-06) but not documented here.
Use the description from the source code.

Document this option and the negated form according to the current
convention.[1]

`--show-names` is also the default when `--get-regexp` is given.  But
don’t mention it here since all the deprecated modes are quarantined in
the “Deprecated Modes” section.

[1]: https://lore.kernel.org/git/xmqqcyct1mtq.fsf@gitster.g/

Acked-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Kristoffer Haugsbakk <code@khaugsbakk.name>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

This option was introduced in a series of commits from fe3ccc7 (Merge
branch 'ps/config-subcommands', 2024-05-15).  But two styles were used
for the value provided to the option:

1. Synopsis: `--value=<value>`
2. Deprecated Modes: `--value=<pattern>`

(2) is also used in the synopsis on the command.

Use (2) consistently throughout since it’s a pattern in the general
case (`value` sounds more generic).

Acked-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Kristoffer Haugsbakk <code@khaugsbakk.name>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

These options were introduced in a series of commits from
fe3ccc7 (Merge branch 'ps/config-subcommands', 2024-05-15).[1]
But they were not documented here.

Document this option and the negated form according to the current
convention.[2]

[1]: `--value` is a replacement for the `value-pattern`
    positional argument
[2]: https://lore.kernel.org/git/xmqqcyct1mtq.fsf@gitster.g/

Acked-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Kristoffer Haugsbakk <code@khaugsbakk.name>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

This option was introduced in a series of commits from fe3ccc7 (Merge
branch 'ps/config-subcommands', 2024-05-15) and deprecated
`value-pattern`.  But `value-pattern` is still used throughout the doc.

The deprecated modes have been quarantined in the “Deprecated Modes”
section.  So let’s only use `--value=<pattern>` in the rest of the doc.

Acked-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Kristoffer Haugsbakk <code@khaugsbakk.name>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

4e51389 (builtin/config: introduce "get" subcommand, 2024-05-06)
introduced `get` and `--url` but didn’t add `--url` to the synopsis.

Acked-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Kristoffer Haugsbakk <code@khaugsbakk.name>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

When clang-format was introduced to the Git project in
6134de6 (clang-format: outline the git project's coding style,
2017-08-14), the 'ColumnLimit' was set to 80. This is inline with our
recommendation in 'Documentation/CodingGuidelines', which states:

  We try to keep to at most 80 characters per line.

However while this is recommended limit, this is not the enforced
limit. In some cases in we do overflow this limit to prioritize
readability. Setting the 'ColumnLimit' also means that shorter lines are
concatenated to simply as the result would still be below 80 characters,
which is undesirable.

In the past, we tried to adjust the penalties around line wrapping, once
in 42efde4 (clang-format: adjust line break penalties, 2017-09-29)
and another time in 5e9fa0f (clang-format: re-adjust line break
penalties, 2024-10-18). While these settings help tweak the line break
penalties to be more in-line with the requirements of the Git project,
using 'clang-format' still produces a lot of false positives.

So to make 'clang-format' more usable, set the 'ColumnLimit' to 0. This
means that line-wrapping is no-longer a concern of the formatter and
something that the user needs to take care of. The previous commit also
added a more flexible guideline to the '.editorconfig' setting a
'max_line_length' of 120 characters. This should provide some guidance
to users.

In the future, it would be nice to re-instate this limit with adequate
penalties which would follow our guidelines, but currently, it makes
more sense to have a working formatter which we can rely on and which
doesn't create too many false positives.

Signed-off-by: Karthik Nayak <karthik.188@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

In 1b8f306 (ci/style-check: add `RemoveBracesLLVM` in CI job,
2024-07-23) we added 'RemoveBracesLLVM' to the CI job of running the
clang formatter.

This rule checks and warns against using braces on simple
single-statement bodies of statements. Since we haven't had any issues
regarding this rule, we can now move it into the main clang-format
config and remove it from being CI exclusive.

Signed-off-by: Karthik Nayak <karthik.188@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

The Makefile has a 'style' rule to run 'git clang-format'. While Meson
intrinsically supports a 'clang-format' target, which can be run when
using the ninja backend by running 'ninja clang-format', this runs the
formatting on all existing files.

Our Meson build doesn't yet support a way to run 'git clang-format',
which runs the formatter between the working directory and commit
provided. Add a new 'style' target to Meson to mimic the target in the
Makefile.

Signed-off-by: Karthik Nayak <karthik.188@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

FreeBSD 6 introduced memmem(), but the implementation diverged
from what was standard everywhere else (including our "compat"
fallback).

FreeBSD 10.4 (went EOL in 2018) corrected the functionality bugs
but kept a suboptimal implementation until FreeBSD 11.4 (the last
version of FreeBSD 11, that went EOL in September 2021).

Let's draw the line to require FreeBSD 12 or newer, which allows us
to drop the special casing of FreeBSD 4.x and rely on the platform
implementation of memmem() unconditionally for all versions that are
still being supported.

Suggested-by: Brad Smith <brad@comstyle.com>
Helped-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

A previous commit removed the last user of it, and it is no
longer useful with the codebase moving towards C99, which
specifies its definition.

Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Add missing colon to fix formatting.

Signed-off-by: Brett A C Sheffield <bacs@librecast.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

The "reftable" format has come a long way and has matured nicely since
it has been merged into git via 57db2a0 (refs: introduce reftable
backend, 2024-02-07). It fixes longstanding issues that cannot be fixed
with the "files" format in a backwards-compatible way and performs
significantly better in many use cases.

Announce that we will switch to the "reftable" format in Git 3.0 for
newly created repositories and wire up the change, hidden behind the
WITH_BREAKING_CHANGES preprocessor define.

This switch is dependent on support in the larger Git ecosystem. Most
importantly, libraries like JGit, libgit2 and Gitoxide should support
the reftable backend so that we don't break all applications and tools
built on top of those libraries.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

With the preceding commit we have announced the switch to the "reftable"
format in Git 3.0 for newly created repositories. The format is being
battle tested by GitLab and a couple of other developers, and except for
a small handful of issues exposed early after it has been merged it has
been rock solid. Regardless of that though the test user base is still
comparatively small, which increases the risk that we miss critical
bugs.

Address this by enabling the reftable format when experimental features
are enabled. This should increase the test user base by some margin and
thus give us more input before making the format the default.

Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

When the user gives us a diff filter like --diff-filter=D, we need to do
a tree diff even if we're not planning to show the diff result itself,
in order to decide whether to show the commit at all. So there's an
explicit check of revs->diffopt.filter in setup_revisions(), and we set
revs->diff if any bits are set.

Originally that "filter" field covered both positive capital-letter
filters (like "D") and also negative lowercase filters (like "d"), so it
was sufficient for both cases. But later, 75408ca (diff-filter: be
more careful when looking for negative bits, 2022-01-28) split the
negative bits out into a "filter_not" field.

We eventually fold those into "filter", but not until diff_setup_done()
is called, which happens after our explicit check. As a result, a purely
negative filter like:

  git log --diff-filter=d

failed to turn on diffs at all. But rather than fail to filter by diff,
because the filter variable is eventually set, we mistakenly show no
commits at all, thinking that the empty diffs were cases where nothing
passed through the filter.

The smallest fix here is to just have our check look for any bits in
either "filter" or "filter_not". I suspect it would also be OK to
reorder the function a bit to call diff_setup_done() earlier, but that
risks violating some other subtle ordering dependency. So I went with
the simple and safe solution here.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

There are three main modes of operation for apply: applying only to the
worktree, applying to the worktree and index (--index), and applying
only to the index (--cached).

The --intent-to-add flag modifies the first of these modes, applying
only to the worktree, in a way which touches the index, because intents
to add are special index entries. However, since its introduction
in cff5dc0 (apply: add --intent-to-add, 2018-05-26), it has not
worked correctly in any but the most trivial (empty repository)
cases, because the index is never read in (in apply, this is done in
read_apply_cache()) before writing to it.

This causes the operation to clobber the old, correct index with a
new empty-tree index before writing intent-to-add entries to this
empty index; the final result is that the index now records every
existing file in the repository as deleted, which is incorrect.

This error can be corrected by first reading the index. The
update_index flag is correctly set if ita_only is true, because this
flag causes the index to be updated. However, if we merely gate the
call to read_apply_cache() behind update_index, then it will not be
read when state->apply is false, even if it must be checked due to
being in --index or --cached mode. Therefore, we instead read the
index if it will be either checked or updated, because reading the
index is a prerequisite to either.

Reported-by: Ryan Hodges <rhodges@cisco.com>
Original-patch-by: Johannes Altmanninger <aclopte@gmail.com>
Signed-off-by: Raymond E. Pasco <ray@ameretat.dev>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

In the "apply only to files" mode (i.e., neither --index nor --cached
mode), the index should not be touched except to record intents to
add when --intent-to-add is on. Because having --intent-to-add on sets
update_index, to indicate that we may touch the index, we can't rely
only on that flag in create_file() (which is called to write both new
files and updated files) to decide whether to write an index entry;
if we did, we would write an index entry for every file being patched
(which would moreover be an intent-to-add entry despite not being a
new file, because we are going to turn on the CE_INTENT_TO_ADD flag
in add_index_entry() if we enter it here and ita_only is true).

To decide whether to touch the index, we need to check the
specific reason the index would be updated, rather than merely
their aggregate in the update_index flag. Because we have already
entered write_out_results() and are performing writes, we know that
state->apply is true. If state->check_index is additionally true, we
are in --index or --cached mode, which updates the index and should
always write, whereas if we are merely in ita_only mode we must only
write if the patch is a new file creation patch.

Signed-off-by: Raymond E. Pasco <ray@ameretat.dev>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Test that applying a new file creation patch with --intent-to-add to
an existing index does not modify the index outside adding the correct
intents-to-add, and that applying a patch with both modifications
and new file creations with --intent-to-add correctly only adds
intents-to-add to the index.

Signed-off-by: Raymond E. Pasco <ray@ameretat.dev>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Avoid using a double negative, and keep in mind that --index and
--cached are distinct modes of operation.

Signed-off-by: Raymond E. Pasco <ray@ameretat.dev>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

In the repo_refresh_and_write_index of read-cache.c, we return -1 to
indicate that writing the index to disk failed.
However, callers do not use this information. Commands such as stash print
  "could not write index"
and then exit, which does not help to discover the exact problem.

We can let repo_hold_locked_index print the error message if the locking
failed.

Signed-off-by: Han Young <hanyang.tony@bytedance.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

There are a couple of "-Wsign-compare" warnings in "string-list.c". Fix
trivial ones that result from a mismatched loop iterator type.

There is a single warning left after these fixes. This warning needs
a bit more care and is thus handled in subsequent commits.

Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

In "add_entry", we accept "insert_at" parameter which must be either -1
(auto) or between 0 and `list->nr` inclusive. Any other value is
invalid. When caller specify any invalid "insert_at" value, we won't
check the range and move the element, which would definitely cause the
trouble.

However, we only use "add_entry" in "string_list_insert" function and we
always pass the "-1" for "insert_at" parameter. So, we never use this
parameter to insert element in a user specified position.

And we should know why there is such code path in the first place. We
used to have another function "string_list_insert_at_index()", which
uses the extra "insert_at" parameter. And in f8c4ab6 (string_list:
remove string_list_insert_at_index() from its API, 2014-11-24), we
remove this function but we don't clean all the code path.

Let's simply delete this parameter as we'd better use "strmap" for such
functionality.

Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

When inserting an existing element, "add_entry" would convert "index"
value to "-1-index" to indicate the caller that this element is in the
list already. However, in "string_list_insert", we would simply convert
this to the original positive index without any further action.

In 8fd2cb4 (Extract helper bits from c-merge-recursive work,
2006-07-25), we create "path-list.c" and then introduce above code path.

Let's directly return the index as we don't care about whether the
element is in the list by using "add_entry". In the future, if we want
to let "add_entry" tell the caller, we may add "int *exact_match"
parameter to "add_entry" instead of converting the index to negative to
indicate.

Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

In "add_entry", we call "get_entry_index" function to get the inserted
position. However, as the return type of "get_entry_index" function is
`int`, there is a sign compare warning when comparing the `index` with
the `list-nr` of unsigned type.

"get_entry_index" would always return unsigned index. However, the
current binary search algorithm initializes "left" to be "-1", which
necessitates the use of signed `int` return type.

The reason why we need to assign "left" to be "-1" is that in the
`while` loop, we increment "left" by 1 to determine whether the loop
should end. This design choice, while functional, forces us to use
signed arithmetic throughout the function.

To resolve this sign comparison issue, let's modify the binary search
algorithm with the following approach:

1. Initialize "left" to 0 instead of -1
2. Use `left < right` as the loop termination condition instead of
   `left + 1 < right`
3. When searching the right part, set `left = middle + 1` instead of
   `middle`

Then, we could delete "#define DISABLE_SIGN_COMPARE_WARNING" to enable
sign warnings check for "string-list".

Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

We rely on "test-tool string-list" command to test the functionality of
the "string-list". However, as we have introduced clar test framework,
we'd better move the shell script into C program to improve speed and
readability.

Create a new file "u-string-list.c" under "t/unit-tests", then update
the Makefile and "meson.build" to build the file. And let's first move
"test_split" into unit test and gradually convert the shell script into
C program.

In order to create `string_list` easily by simply specifying strings in
the function call, create "t_vcreate_string_list_dup" function to do
this.

Then port the shell script tests to C program and remove unused
"test-tool" code and tests.

Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

We use "test-tool string-list split_in_place" to test the
"string_list_split_in_place" function. As we have introduced the unit
test, we'd better remove the logic from shell script to C program to
improve test speed and readability.

Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

We use "test-tool string-list filter" to test the "filter_string_list"
function. As we have introduced the unit test, we'd better remove the
logic from shell script to C program to improve test speed and
readability.

Signed-off-by: shejialuo <shejialuo@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

The 'extensions.preciousObjects' setting when set true, prevents
operations that might drop objects from the object storage. This setting
is populated in the global variable
'repository_format_precious_objects'.

Move this global variable to repo scope by adding it to 'struct
repository and also refactor all the occurences accordingly.

This change is part of an ongoing effort to eliminate global variables,
improve modularity and help libify the codebase.

Mentored-by: Christian Couder <christian.couder@gmail.com>
Mentored-by: Ghanshyam Thakkar <shyamthakkar001@gmail.com>
Signed-off-by: Ayush Chandekar <ayu.chandekar@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Refactor builtin/prune.c to remove the dependency on the global
'the_repository'. Replace all the occurrences of 'the_repository' with
repo and thus remove the definition '#define
USE_THE_REPOSITORY_VARIABLE'. Also, add a test to make sure that 'git
prune -h' can be called when the repository is `NULL`.

Mentored-by: Christian Couder <christian.couder@gmail.com>
Mentored-by: Ghanshyam Thakkar <shyamthakkar001@gmail.com>
Signed-off-by: Ayush Chandekar <ayu.chandekar@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Commit 50dec7c ("config.mak.uname: add sysinfo() configuration for
cygwin", 2025-04-17) and later commit 187ce02 ("configure.ac: upgrade
to a compilation check for sysinfo", 2025-05-19) added a 'sysinfo()'
check to the autoconf build.

The FreeBSD system has an optional sysinfo compatibility library, used
to assist in porting software, which causes the build to fail when it
is installed. The reason for the failure is the lack of '-lsysinfo'
during the linking step.

Several solutions were considered:

  - add a 'linking' check to configure.ac in order to determine the
    need to link a separate library (-lsysinfo). (This would require
    a similar change to meson.build).

  - change the order of the preprocessor conditionals in the total_ram()
    function in 'builtin/gc.c', so that the *BSD sysctl() function
    (in the HAVE_BSD_SYSCTL block) takes priority over the sysinfo()
    function (in the HAVE_SYSINFO block).

  - suppress the setting of HAVE_SYSINFO when HAVE_BSD_SYSCTL has been
    defined (in both configure.ac and meson.build).

The first solution above, while simple, adds unnecessary code (the
sysinfo compat function is likely implemented using sysctl() anyway)
when git is happy to use sysctl() on *BSD systems.

The second solution would only be required by the autoconf and meson
build systems, the Makefile already sets the build variables to the
required values (since they are not 'auto-detected').

Here we opt for the final solution above, since it only requires that
we prioritise the 'auto-detected' build variables in the autoconf and
meson builds.

In order to fix the FreeBSD build, move the sysinfo() check after the
determination of the HAVE_BSD_SYSCTL build variable, suppressing the
setting of HAVE_SYSINFO if HAVE_BSD_SYSCTL is defined. Apply this logic
to both the configure.ac and meson.build file.

[Thanks go to Renato Botelho <garga@FreeBSD.org> for testing this patch
on FreeBSD.]

Tested-by: Renato Botelho <garga@FreeBSD.org>
Signed-off-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Documentation for git-merge incorrectly notes that
tip of the current branch on ascii diagram is C,
while it is actually G (current branch is master,
HEAD on diagram is G).

Additionally diagrams on the page are adjusted
to use spaces instead of tabs, so that they align
regardless of tab size. This is in line with
diagrams on other git documentation pages.

Signed-off-by: Timur Sultanaev <str.write@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

In 6aec8d3 (t: refactor tests depending on Perl to print data,
2025-04-03) we have changed one of the tests in t4150 to use awk(1)
instead of Perl. The test works, but at least gawk(1) prints a warning
now:

    awk: cmd. line:3: warning: escape sequence `\@' treated as plain `@'

Fix this by removing the backslash.

Reported-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

In 6aec8d3 (t: refactor tests depending on Perl to print data,
2025-04-03) we have changed some of the tests in t4150 to use sed(1)
instead of Perl. One of the conversions is broken though:

    sed: -e expression #1, char 41: unterminated `s' command

Curiously enough, the test itself still passes. This is caused by a
sequence of failures:

  1. The output of sed(1) is piped into git-update-ref(1), and because
     sed(1) is the upstream command we don't notice that it fails.

  2. git-update-ref(1) does not receive any input and thus won't create
     any references.

  3. We then repack the repository with the configured pseudo merges
     pattern, but as we didn't create any references the pattern doesn't
     match anything.

  4. We use `test_pseudo_merges()` to compute the list of pseudo-merges
     and write it into a file. This file is empty as there are none.

  5. The loop over the pseudo-merges becomes a no-op.

  6. The final test succeeds as well because the number of lines in an
     empty file is obviously the same as the number of unique lines,
     namely zero.

Fix the issue by adding the terminating '|' to the sed(1) command.
Furthermore, make the test a tiny bit more robust by not using it as
part of a pipe.

Reported-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

The calls to sysctl() assume a 64-bit memory size for the variable
holding the value, but the actual size depends on the key name and
platform, at least for HW_PHYSMEM.

Detect any mismatched reads, and retry with a shorter variable
when needed.

Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Detaching the filename string from the tempfile structure used to cause
delete_tempfile() to fail and the temporary file was not cleaned up.

While it's possible to get rid of the allocation and copy from
xstrdup(), it keeps the code symetric with the other branch since
interpolate_path() also allocates and ssh_signing_key_file is freed
in both cases.

The exisiting test was updated to check if the temporary files are
properly deleted. To prevent TMPDIR from leaking into the other tests, a
new subshell is created, however this prevents test_config from working.
The cleanup of the config changed in the subshell is done by
test_unconfig in a call to test_when_finished outside of it.

Helped-by: brian m. carlson <sandals@crustytoothpaste.net>
Helped-by: Patrick Steinhardt <ps@pks.im>
Helped-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: redoste <redoste@redoste.xyz>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Docfix.

* bs/remote-helpers-doc-markup-fix:
  gitremote-helpers.adoc: fix formatting

Use of sysctl() system call to learn the total RAM size used on
BSDs has been corrected.

* cb/total-ram-bsd-fix:
  builtin/gc: correct total_ram calculation with HAVE_BSD_SYSCTL

Drop FreeBSD 4 support and assume we are at least at FreeBSD 6 with
memmem() supported.

* bs/config-mak-freebsd:
  build: retire NO_UINTMAX_T
  config.mak.uname: set NO_MEMMEM only for functional version

Some code paths in the "git prune" used to ignore passed in
repository object and used the_repository singleton instance
instead, which has been corrected.

* ac/prune-wo-the-repository:
  builtin/prune: stop depending on 'the_repository'
  repository: move 'repository_format_precious_objects' to repo scope

A diff-filter with negative-only specification like "git log
--diff-filter=d" did not trigger correctly, which has been fixed.

* jk/all-negative-diff-filter-fix:
  setup_revisions(): turn on diffs for all-negative diff filter

The reftable ref backend has matured enough; Git 3.0 will make it
the default format in a newly created repositories by default.

* ps/use-reftable-as-default-in-3.0:
  setup: use "reftable" format when experimental features are enabled
  BreakingChanges: announce switch to "reftable" format

"make coccicheck" succeeds even when spatch made suggestions, which
has been updated to fail in such a case.

* jc/coccicheck-fails-make-when-it-fails:
  coccicheck: fail "make" when it fails

"netrc" credential helper has been improved to understand textual
service names (like smtp) in addition to the numeric port numbers
(like 25).

* mc/netrc-service-names:
  contrib: better support symbolic port names in git-credential-netrc
  contrib: warn for invalid netrc file ports in git-credential-netrc
  contrib: use a more portable shebang for git-credential-netrc

Documentation updates.

* kh/doc-config-subcommands:
  config: mention --url in the synopsis
  config: use --value instead of value-pattern
  config: document --[no-]value
  config: use --value=<pattern> consistently
  config: document --[no-]show-names

Update ".clang-format" and ".editorconfig" to match our style guide
a bit better.

* kn/clang-format-updates:
  meson: add rule to run 'git clang-format'
  clang-format: add 'RemoveBracesLLVM' to the main config
  clang-format: set 'ColumnLimit' to 0

A failure to open the index file for writing due to conflicting
access did not state what went wrong, which has been corrected.

* hy/read-cache-lock-error-fix:
  read-cache: report lock error when refreshing index

Tempfile removal fix in the codepath to sign commits with SSH keys.

* re/ssh-sign-buffer-fix:
  ssh signing: don't detach the filename strbuf from key_file tempfile

Test fixes.

* ps/perlless-test-fixes:
  t5333: fix missing terminator for sed(1) 's' command
  t4150: fix warning printed by awk due to escaped '\@'

Doc fix.

* ts/merge-orig-head-doc-fix:
  docs: correct ORIG_HEAD example in "git merge" documentation

Build fix for FreeBSD.

* rj/freebsd-sysinfo-build-fix:
  build: fix FreeBSD build when sysinfo compat library installed

Code and test clean-up around string-list API.

* sj/string-list:
  u-string-list: move "remove duplicates" test to "u-string-list.c"
  u-string-list: move "filter string" test to "u-string-list.c"
  u-string-list: move "test_split_in_place" to "u-string-list.c"
  u-string-list: move "test_split" into "u-string-list.c"
  string-list: enable sign compare warnings check
  string-list: return index directly when inserting an existing element
  string-list: remove unused "insert_at" parameter from add_entry
  string-list: fix sign compare warnings for loop iterator

"git apply -N" should start from the current index and register
only new files, but it instead started from an empty index, which
has been corrected.

* rp/apply-intent-to-add-fix:
  apply docs: clarify wording for --intent-to-add
  t4140: test apply --intent-to-add interactions
  apply: only write intents to add for new files
  apply: read in the index in --intent-to-add mode

Signed-off-by: Junio C Hamano <gitster@pobox.com>