Updates since v2.21
-------------------
+Backward compatibility note
+
+ * The filter specification "--filter=sparse:path=<path>" used to
+ create a lazy/partial clone has been removed. Using a blob that is
+ part of the project as sparse specification is still supported with
+ the "--filter=sparse:oid=<blob>" option.
+
UI, Workflows & Features
* "git checkout --no-overlay" can be used to trigger a new mode of
* "git submodule" learns "set-branch" subcommand that allows the
submodule.*.branch settings to be modified.
+ * "git merge-recursive" backend recently learned a new heuristics to
+ infer file movement based on how other files in the same directory
+ moved. As this is inherently less robust heuristics than the one
+ based on the content similarity of the file itself (rather than
+ based on what its neighbours are doing), it sometimes gives an
+ outcome unexpected by the end users. This has been toned down to
+ leave the renamed paths in higher/conflicted stages in the index so
+ that the user can examine and confirm the result.
+
+ * "git tag" learned to give an advice suggesting it might be a
+ mistake when creating an annotated or signed tag that points at
+ another tag.
+
+ * The "git pack-objects" command learned to report the number of
+ objects it packed via the trace2 mechanism.
+
+ * The list of conflicted paths shown in the editor while concluding a
+ conflicted merge was shown above the scissors line when the
+ clean-up mode is set to "scissors", even though it was commented
+ out just like the list of updated paths and other information to
+ help the user explain the merge better.
+
+ * The trace2 tracing facility learned to auto-generate a filename
+ when told to log to a directory.
+
+ * "git clone" learned a new --server-option option when talking over
+ the protocol version 2.
+
+ * The connectivity bitmaps are created by default in bare
+ repositories now; also the pathname hash-cache is created by
+ default to avoid making crappy deltas when repacking.
+
+ * "git branch new A...B" and "git checkout -b new A...B" have been
+ taught that in their contexts, the notation A...B means "the merge
+ base between these two commits", just like "git checkout A...B"
+ detaches HEAD at that commit.
+
+ * Update "git difftool" and "git mergetool" so that the combinations
+ of {diff,merge}.{tool,guitool} configuration variables serve as
+ fallback settings of each other in a sensible order.
+
+ * The "--dir-diff" mode of "git difftool" is not useful in "--no-index"
+ mode; they are now explicitly marked as mutually incompatible.
+
Performance, Internal Implementation, Development Support etc.
achieve better performance by batching the request for these
promised blobs.
+ * During an initial "git clone --depth=..." partial clone, it is
+ pointless to spend cycles for a large portion of the connectivity
+ check that enumerates and skips promisor objects (which by
+ definition is all objects fetched from the other side). This has
+ been optimized out.
+
+ * Mechanically and systematically drop "extern" from function
+ declaration.
+
+ * The script to aggregate perf result unconditionally depended on
+ libjson-perl even though it did not have to, which has been
+ corrected.
+
+ * The internal implementation of "git rebase -i" has been updated to
+ avoid forking a separate "rebase--interactive" process.
+
+ * Allow DEP and ASLR for Windows build to for security hardening.
+
+ * Performance test framework has been broken and measured the version
+ of Git that happens to be on $PATH, not the specified one to
+ measure, for a while, which has been corrected.
+
+ * Optionally "make coccicheck" can feed multiple source files to
+ spatch, gaining performance while spending more memory.
+
+ * Attempt to use an abbreviated option in "git clone --recurs" is
+ responded by a request to disambiguate between --recursive and
+ --recurse-submodules, which is bad because these two are synonyms.
+ The parse-options API has been extended to define such synonyms
+ more easily and not produce an unnecessary failure.
+
+ * A pair of private functions in http.c that had names similar to
+ fread/fwrite did not return the number of elements, which was found
+ to be confusing.
+
+ * Update collision-detecting SHA-1 code to build properly on HP-UX.
+
Fixes since v2.21
-----------------
* On platforms where "git fetch" is killed with SIGPIPE (e.g. OSX),
the upload-pack that runs on the other end that hangs up after
detecting an error could cause "git fetch" to die with a signal,
- which led to a flakey test. "git fetch" now ignores SIGPIPE during
+ which led to a flaky test. "git fetch" now ignores SIGPIPE during
the network portion of its operation (this is not a problem as we
check the return status from our write(2)s).
(merge 143588949c jk/no-sigpipe-during-network-transport later to maint).
(merge b5a0bd694c nd/read-tree-reset-doc later to maint).
* Code clean-up around a much-less-important-than-it-used-to-be
- update_server_info() funtion.
+ update_server_info() function.
(merge b3223761c8 jk/server-info-rabbit-hole later to maint).
* The message given when "git commit -a <paths>" errors out has been
eager and considered nonsense strings as if they can be legitimate
beginning of *-by: trailer. This has been tightened.
- * Build with gettext breaks on recent macOS w/ Homebrew when
- /usr/local/bin is not on PATH, which has been corrected.
+ * Builds with gettext broke on recent macOS w/ Homebrew, which
+ seems to have stopped including from /usr/local/include; this
+ has been corrected.
(merge 92a1377a2a js/macos-gettext-build later to maint).
+ * Running "git add" on a repository created inside the current
+ repository is an explicit indication that the user wants to add it
+ as a submodule, but when the HEAD of the inner repository is on an
+ unborn branch, it cannot be added as a submodule. Worse, the files
+ in its working tree can be added as if they are a part of the outer
+ repository, which is not what the user wants. These problems are
+ being addressed.
+ (merge f937bc2f86 km/empty-repo-is-still-a-repo later to maint).
+
+ * "git cherry-pick" run with the "-x" or the "--signoff" option used
+ to (and more importantly, ought to) clean up the commit log message
+ with the --cleanup=space option by default, but this has been
+ broken since late 2017. This has been fixed.
+
+ * When given a tag that points at a commit-ish, "git replace --graft"
+ failed to peel the tag before writing a replace ref, which did not
+ make sense because the old graft mechanism the feature wants to
+ mimic only allowed to replace one commit object with another.
+ This has been fixed.
+ (merge ee521ec4cb cc/replace-graft-peel-tags later to maint).
+
+ * Code tightening against a "wrong" object appearing where an object
+ of a different type is expected, instead of blindly assuming that
+ the connection between objects are correctly made.
+ (merge 97dd512af7 tb/unexpected later to maint).
+
+ * An earlier update for MinGW and Cygwin accidentally broke MSVC build,
+ which has been fixed.
+ (merge 22c3634c0f ss/msvc-path-utils-fix later to maint).
+
+ * %(push:track) token used in the --format option to "git
+ for-each-ref" and friends was not showing the right branch, which
+ has been fixed.
+ (merge c646d0934e dr/ref-filter-push-track-fix later to maint).
+
+ * "make check-docs", "git help -a", etc. did not account for cases
+ where a particular build may deliberately omit some subcommands,
+ which has been corrected.
+
+ * The logic to tell if a Git repository has a working tree protects
+ "git branch -D" from removing the branch that is currently checked
+ out by mistake. The implementation of this logic was broken for
+ repositories with unusual name, which unfortunately is the norm for
+ submodules these days. This has been fixed.
+ (merge f3534c98e4 jt/submodule-repo-is-with-worktree later to maint).
+
+ * AIX shared the same build issues with other BSDs around fileno(fp),
+ which has been corrected.
+ (merge ee662bf5c6 cc/aix-has-fileno-as-a-macro later to maint).
+
+ * The autoconf generated configure script failed to use the right
+ gettext() implementations from -libintl by ignoring useless stub
+ implementations shipped in some C library, which has been
+ corrected.
+ (merge b71e56a683 vk/autoconf-gettext later to maint).
+
+ * Fix index-pack perf test so that the repeated invocations always
+ run in an empty repository, which emulates the initial clone
+ situation better.
+ (merge 775c71e16d jk/p5302-avoid-collision-check-cost later to maint).
+
+ * A "ls-files" that emulates "find" to enumerate files in the working
+ tree resulted in duplicated Makefile rules that caused the build to
+ issue an unnecessary warning during a trial build after merge
+ conflicts are resolved in working tree *.h files but before the
+ resolved results are added to the index. This has been corrected.
+
+ * "git cherry-pick" (and "revert" that shares the same runtime engine)
+ that deals with multiple commits got confused when the final step
+ gets stopped with a conflict and the user concluded the sequence
+ with "git commit". Attempt to fix it by cleaning up the state
+ files used by these commands in such a situation.
+ (merge 4a72486de9 pw/clean-sequencer-state-upon-final-commit later to maint).
+
+ * On a filesystem like HFS+, the names of the refs stored as filesystem
+ entities may become different from what the end-user expects, just
+ like files in the working tree get "renamed". Work around the
+ mismatch by paying attention to the core.precomposeUnicode
+ configuration.
+ (merge 8e712ef6fc en/unicode-in-refnames later to maint).
+
+ * The code to generate the multi-pack idx file was not prepared to
+ see too many packfiles and ran out of open file descriptor, which
+ has been corrected.
+
+ * To run tests for Git SVN, our scripts for CI used to install the
+ git-svn package (in the hope that it would bring in the right
+ dependencies). This has been updated to install the more direct
+ dependency, namely, libsvn-perl.
+ (merge db864306cf sg/ci-libsvn-perl later to maint).
+
+ * "git cvsexportcommit" running on msys did not expect cvsnt showed
+ "cvs status" output with CRLF line endings.
+
+ * The fsmonitor interface got out of sync after the in-core index
+ file gets discarded, which has been corrected.
+ (merge 398a3b0899 js/fsmonitor-refresh-after-discarding-index later to maint).
+
+ * "git status" did not know that the "label" instruction in the
+ todo-list "rebase -i -r" uses should not be shown as a hex object
+ name.
+
+ * A prerequisite check in the test suite to see if a working jgit is
+ available was made more robust.
+ (merge abd0f28983 tz/test-lib-check-working-jgit later to maint).
+
+ * The codepath to parse :<path> that obtains the object name for an
+ indexed object has been made more robust.
+
* Code cleanup, docfix, build fix, etc.
(merge 11f470aee7 jc/test-yes-doc later to maint).
(merge 90503a240b js/doc-symref-in-proto-v1 later to maint).
(merge f64a21bd82 tz/doc-apostrophe-no-longer-needed later to maint).
(merge dbe7b41019 js/t3301-unbreak-notes-test later to maint).
(merge d8083e4180 km/t3000-retitle later to maint).
+ (merge 9e4cbccbd7 tz/git-svn-doc-markup-fix later to maint).
+ (merge da9ca955a7 jk/ls-files-doc-markup-fix later to maint).
+ (merge 6804ba3a58 cw/diff-highlight later to maint).
+ (merge 1a8787144d nd/submodule-helper-incomplete-line-fix later to maint).
+ (merge d9ef573837 jk/apache-lsan later to maint).
+ (merge c871fbee2b js/t6500-use-windows-pid-on-mingw later to maint).
+ (merge ce4c7bfc90 bl/t4253-exit-code-from-format-patch later to maint).
+ (merge 397a46db78 js/t5580-unc-alternate-test later to maint).
+ (merge d4907720a2 cm/notes-comment-fix later to maint).
+ (merge 9dde06de13 cb/http-push-null-in-message-fix later to maint).
+ (merge 4c785c0edc js/rebase-config-bitfix later to maint).
+ (merge 8e9fe16c87 es/doc-gitsubmodules-markup later to maint).