RelNotes/2.31.0.adoc

Git 2.31 Release Notes
======================

Updates since v2.30
-------------------

Backward incompatible and other important changes

 * The "pack-redundant" command, which has been left stale with almost
   unusable performance issues, now warns loudly when it gets used, as
   we no longer want to recommend its use (instead just "repack -d"
   instead).

 * The development community has adopted Contributor Covenant v2.0 to
   update from v1.4 that we have been using.

 * The support for deprecated PCRE1 library has been dropped.

 * Fixes for CVE-2021-21300 in Git 2.30.2 (and earlier) is included.


UI, Workflows & Features

 * The "--format=%(trailers)" mechanism gets enhanced to make it
   easier to design output for machine consumption.

 * When a user does not tell "git pull" to use rebase or merge, the
   command gives a loud message telling a user to choose between
   rebase or merge but creates a merge anyway, forcing users who would
   want to rebase to redo the operation.  Fix an early part of this
   problem by tightening the condition to give the message---there is
   no reason to stop or force the user to choose between rebase or
   merge if the history fast-forwards.

 * The configuration variable 'core.abbrev' can be set to 'no' to
   force no abbreviation regardless of the hash algorithm.

 * "git rev-parse" can be explicitly told to give output as absolute
   or relative path with the `--path-format=(absolute|relative)` option.

 * Bash completion (in contrib/) update to make it easier for
   end-users to add completion for their custom "git" subcommands.

 * "git maintenance" learned to drive scheduled maintenance on
   platforms whose native scheduling methods are not 'cron'.

 * After expiring a reflog and making a single commit, the reflog for
   the branch would record a single entry that knows both @{0} and
   @{1}, but we failed to answer "what commit were we on?", i.e. @{1}

 * "git bundle" learns "--stdin" option to read its refs from the
   standard input.  Also, it now does not lose refs whey they point
   at the same object.

 * "git log" learned a new "--diff-merges=<how>" option.

 * "git ls-files" can and does show multiple entries when the index is
   unmerged, which is a source for confusion unless -s/-u option is in
   use.  A new option --deduplicate has been introduced.

 * `git worktree list` now annotates worktrees as prunable, shows
   locked and prunable attributes in --porcelain mode, and gained
   a --verbose option.

 * "git clone" tries to locally check out the branch pointed at by
   HEAD of the remote repository after it is done, but the protocol
   did not convey the information necessary to do so when copying an
   empty repository.  The protocol v2 learned how to do so.

 * There are other ways than ".." for a single token to denote a
   "commit range", namely "<rev>^!" and "<rev>^-<n>", but "git
   range-diff" did not understand them.

 * The "git range-diff" command learned "--(left|right)-only" option
   to show only one side of the compared range.

 * "git mergetool" feeds three versions (base, local and remote) of
   a conflicted path unmodified.  The command learned to optionally
   prepare these files with unconflicted parts already resolved.

 * The .mailmap is documented to be read only from the root level of a
   working tree, but a stray file in a bare repository also was read
   by accident, which has been corrected.

 * "git maintenance" tool learned a new "pack-refs" maintenance task.

 * The error message given when a configuration variable that is
   expected to have a boolean value has been improved.

 * Signed commits and tags now allow verification of objects, whose
   two object names (one in SHA-1, the other in SHA-256) are both
   signed.

 * "git rev-list" command learned "--disk-usage" option.

 * "git {diff,log} --{skip,rotate}-to=<path>" allows the user to
   discard diff output for early paths or move them to the end of the
   output.

 * "git difftool" learned "--skip-to=<path>" option to restart an
   interrupted session from an arbitrary path.

 * "git grep" has been tweaked to be limited to the sparse checkout
   paths.

 * "git rebase --[no-]fork-point" gained a configuration variable
   rebase.forkPoint so that users do not have to keep specifying a
   non-default setting.


Performance, Internal Implementation, Development Support etc.

 * A 3-year old test that was not testing anything useful has been
   corrected.

 * Retire more names with "sha1" in it.

 * The topological walk codepath is covered by new trace2 stats.

 * Update the Code-of-conduct to version 2.0 from the upstream (we've
   been using version 1.4).

 * "git mktag" validates its input using its own rules before writing
   a tag object---it has been updated to share the logic with "git
   fsck".

 * Two new ways to feed configuration variable-value pairs via
   environment variables have been introduced, and the way
   GIT_CONFIG_PARAMETERS encodes variable/value pairs has been tweaked
   to make it more robust.

 * Tests have been updated so that they do not to get affected by the
   name of the default branch "git init" creates.

 * "git fetch" learns to treat ref updates atomically in all-or-none
   fashion, just like "git push" does, with the new "--atomic" option.

 * The peel_ref() API has been replaced with peel_iterated_oid().

 * The .use_shell flag in struct child_process that is passed to
   run_command() API has been clarified with a bit more documentation.

 * Document, clean-up and optimize the code around the cache-tree
   extension in the index.

 * The ls-refs protocol operation has been optimized to narrow the
   sub-hierarchy of refs/ it walks to produce response.

 * When removing many branches and tags, the code used to do so one
   ref at a time.  There is another API it can use to delete multiple
   refs, and it makes quite a lot of performance difference when the
   refs are packed.

 * The "pack-objects" command needs to iterate over all the tags when
   automatic tag following is enabled, but it actually iterated over
   all refs and then discarded everything outside "refs/tags/"
   hierarchy, which was quite wasteful.

 * A perf script was made more portable.

 * Our setting of GitHub CI test jobs were a bit too eager to give up
   once there is even one failure found.  Tweak the knob to allow
   other jobs keep running even when we see a failure, so that we can
   find more failures in a single run.

 * We've carried compatibility codepaths for compilers without
   variadic macros for quite some time, but the world may be ready for
   them to be removed.  Force compilation failure on exotic platforms
   where variadic macros are not available to find out who screams in
   such a way that we can easily revert if it turns out that the world
   is not yet ready.

 * Code clean-up to ensure our use of hashtables using object names as
   keys use the "struct object_id" objects, not the raw hash values.

 * Lose the debugging aid that may have been useful in the past, but
   no longer is, in the "grep" codepaths.

 * Some pretty-format specifiers do not need the data in commit object
   (e.g. "%H"), but we were over-eager to load and parse it, which has
   been made even lazier.

 * Get rid of "GETTEXT_POISON" support altogether, which may or may
   not be controversial.

 * Introduce an on-disk file to record revindex for packdata, which
   traditionally was always created on the fly and only in-core.

 * The commit-graph learned to use corrected commit dates instead of
   the generation number to help topological revision traversal.

 * Piecemeal of rewrite of "git bisect" in C continues.

 * When a pager spawned by us exited, the trace log did not record its
   exit status correctly, which has been corrected.

 * Removal of GIT_TEST_GETTEXT_POISON continues.

 * The code to implement "git merge-base --independent" was poorly
   done and was kept from the very beginning of the feature.

 * Preliminary changes to fsmonitor integration.

 * Performance improvements for rename detection.

 * The common code to deal with "chunked file format" that is shared
   by the multi-pack-index and commit-graph files have been factored
   out, to help codepaths for both filetypes to become more robust.

 * The approach to "fsck" the incoming objects in "index-pack" is
   attractive for performance reasons (we have them already in core,
   inflated and ready to be inspected), but fundamentally cannot be
   applied fully when we receive more than one pack stream, as a tree
   object in one pack may refer to a blob object in another pack as
   ".gitmodules", when we want to inspect blobs that are used as
   ".gitmodules" file, for example.  Teach "index-pack" to emit
   objects that must be inspected later and check them in the calling
   "fetch-pack" process.

 * The logic to handle "trailer" related placeholders in the
   "--format=" mechanisms in the "log" family and "for-each-ref"
   family is getting unified.

 * Raise the buffer size used when writing the index file out from
   (obviously too small) 8kB to (clearly sufficiently large) 128kB.

 * It is reported that open() on some platforms (e.g. macOS Big Sur)
   can return EINTR even though our timers are set up with SA_RESTART.
   A workaround has been implemented and enabled for macOS to rerun
   open() transparently from the caller when this happens.


Fixes since v2.30
-----------------

 * Diagnose command line error of "git rebase" early.

 * Clean up option descriptions in "git cmd --help".

 * "git stash" did not work well in a sparsely checked out working
   tree.

 * Some tests expect that "ls -l" output has either '-' or 'x' for
   group executable bit, but setgid bit can be inherited from parent
   directory and make these fields 'S' or 's' instead, causing test
   failures.

 * "git for-each-repo --config=<var> <cmd>" should not run <cmd> for
   any repository when the configuration variable <var> is not defined
   even once.

 * Fix 2.29 regression where "git mergetool --tool-help" fails to list
   all the available tools.

 * Fix for procedure to building CI test environment for mac.

 * The implementation of "git branch --sort" wrt the detached HEAD
   display has always been hacky, which has been cleaned up.

 * Newline characters in the host and path part of git:// URL are
   now forbidden.

 * "git diff" showed a submodule working tree with untracked cruft as
   "Submodule commit <objectname>-dirty", but a natural expectation is
   that the "-dirty" indicator would align with "git describe --dirty",
   which does not consider having untracked files in the working tree
   as source of dirtiness.  The inconsistency has been fixed.

 * When more than one commit with the same patch ID appears on one
   side, "git log --cherry-pick A...B" did not exclude them all when a
   commit with the same patch ID appears on the other side.  Now it
   does.

 * Documentation for "git fsck" lost stale bits that has become
   incorrect.

 * Doc fix for packfile URI feature.

 * When "git rebase -i" processes "fixup" insn, there is no reason to
   clean up the commit log message, but we did the usual stripspace
   processing.  This has been corrected.
   (merge f7d42ceec5 js/rebase-i-commit-cleanup-fix later to maint).

 * Fix in passing custom args from "git clone" to "upload-pack" on the
   other side.
   (merge ad6b5fefbd jv/upload-pack-filter-spec-quotefix later to maint).

 * The command line completion (in contrib/) completed "git branch -d"
   with branch names, but "git branch -D" offered tagnames in addition,
   which has been corrected.  "git branch -M" had the same problem.
   (merge 27dc071b9a jk/complete-branch-force-delete later to maint).

 * When commands are started from a subdirectory, they may have to
   compare the path to the subdirectory (called prefix and found out
   from $(pwd)) with the tracked paths.  On macOS, $(pwd) and
   readdir() yield decomposed path, while the tracked paths are
   usually normalized to the precomposed form, causing mismatch.  This
   has been fixed by taking the same approach used to normalize the
   command line arguments.
   (merge 5c327502db tb/precompose-prefix-too later to maint).

 * Even though invocations of "die()" were logged to the trace2
   system, "BUG()"s were not, which has been corrected.
   (merge 0a9dde4a04 jt/trace2-BUG later to maint).

 * "git grep --untracked" is meant to be "let's ALSO find in these
   files on the filesystem" when looking for matches in the working
   tree files, and does not make any sense if the primary search is
   done against the index, or the tree objects.  The "--cached" and
   "--untracked" options have been marked as mutually incompatible.
   (merge 0c5d83b248 mt/grep-cached-untracked later to maint).

 * Fix "git fsck --name-objects" which apparently has not been used by
   anybody who is motivated enough to report breakage.
   (merge e89f89361c js/fsck-name-objects-fix later to maint).

 * Avoid individual tests in t5411 from getting affected by each other
   by forcing them to use separate output files during the test.
   (merge 822ee894f6 jx/t5411-unique-filenames later to maint).

 * Test to make sure "git rev-parse one-thing one-thing" gives
   the same thing twice (when one-thing is --since=X).
   (merge a5cdca4520 ew/rev-parse-since-test later to maint).

 * When certain features (e.g. grafts) used in the repository are
   incompatible with the use of the commit-graph, we used to silently
   turned commit-graph off; we now tell the user what we are doing.
   (merge c85eec7fc3 js/commit-graph-warning later to maint).

 * Objects that lost references can be pruned away, even when they
   have notes attached to it (and these notes will become dangling,
   which in turn can be pruned with "git notes prune").  This has been
   clarified in the documentation.
   (merge fa9ab027ba mz/doc-notes-are-not-anchors later to maint).

 * The error codepath around the "--temp/--prefix" feature of "git
   checkout-index" has been improved.
   (merge 3f7ba60350 mt/checkout-index-corner-cases later to maint).

 * The "git maintenance register" command had trouble registering bare
   repositories, which had been corrected.

 * A handful of multi-word configuration variable names in
   documentation that are spelled in all lowercase have been corrected
   to use the more canonical camelCase.
   (merge 7dd0eaa39c dl/doc-config-camelcase later to maint).

 * "git push $there --delete ''" should have been diagnosed as an
   error, but instead turned into a matching push, which has been
   corrected.
   (merge 20e416409f jc/push-delete-nothing later to maint).

 * Test script modernization.
   (merge 488acf15df sv/t7001-modernize later to maint).

 * An under-allocation for the untracked cache data has been corrected.
   (merge 6347d649bc jh/untracked-cache-fix later to maint).

 * Other code cleanup, docfix, build fix, etc.
   (merge e3f5da7e60 sg/t7800-difftool-robustify later to maint).
   (merge 9d336655ba js/doc-proto-v2-response-end later to maint).
   (merge 1b5b8cf072 jc/maint-column-doc-typofix later to maint).
   (merge 3a837b58e3 cw/pack-config-doc later to maint).
   (merge 01168a9d89 ug/doc-commit-approxidate later to maint).
   (merge b865734760 js/params-vs-args later to maint).