Git 2.25 Release Notes
======================
Updates since v2.24
-------------------
Backward compatibility notes
UI, Workflows & Features
* A tutorial on object enumeration has been added.
* The branch description ("git branch --edit-description") has been
used to fill the body of the cover letters by the format-patch
command; this has been enhanced so that the subject can also be
filled.
* "git rebase --preserve-merges" has been marked as deprecated; this
release stops advertising it in the "git rebase -h" output.
* The code to generate multi-pack index learned to show (or not to
show) progress indicators.
* "git apply --3way" learned to honor merge.conflictStyle
configuration variable, like merges would.
* The custom format for "git log --format=<format>" learned the l/L
placeholder that is similar to e/E that fills in the e-mail
address, but only the local part on the left side of '@'.
* Documentation pages for "git shortlog" now list commit limiting
options explicitly.
* The patterns to detect function boundary for Elixir language has
been added.
* The completion script (in contrib/) learned that the "--onto"
option of "git rebase" can take its argument as the value of the
option.
* The userdiff machinery has been taught that "async def" is another
way to begin a "function" in Python.
* "git range-diff" learned to take the "--notes=<ref>" and the
"--no-notes" options to control the commit notes included in the
log message that gets compared.
* "git rev-parse --show-toplevel" run outside of any working tree did
not error out, which has been corrected.
* A few commands learned to take the pathspec from the standard input
or a named file, instead of taking it as the command line
arguments, with the "--pathspec-from-file" option.
* "git submodule" learned a subcommand "set-url".
* "git log" family learned "--pretty=reference" that gives the name
of a commit in the format that is often used to refer to it in log
messages.
* The interaction between "git clone --recurse-submodules" and
alternate object store was ill-designed. The documentation and
code have been taught to make more clear recommendations when the
users see failures.
* Management of sparsely checked-out working tree has gained a
dedicated "sparse-checkout" command.
* Miscellaneous small UX improvements on "git-p4".
* "git sparse-checkout list" subcommand learned to give its output in
a more concise form when the "cone" mode is in effect.
Performance, Internal Implementation, Development Support etc.
* Debugging support for lazy cloning has been a bit improved.
* Move the definition of a set of bitmask constants from 0ctal
literal to (1U<<count) notation.
* Test updates to prepare for SHA-2 transition continues.
* Crufty code and logic accumulated over time around the object
parsing and low-level object access used in "git fsck" have been
cleaned up.
* The implementation of "git log --graph" got refactored and then its
output got simplified.
* Follow recent push to move API docs from Documentation/ to header
files and update config.h
* "git bundle" has been taught to use the parse options API. "git
bundle verify" learned "--quiet" and "git bundle create" learned
options to control the progress output.
* Handling of commit objects that use non UTF-8 encoding during
"rebase -i" has been improved.
* The beginning of rewriting "git add -i" in C.
* A label used in the todo list that are generated by "git rebase
--rebase-merges" is used as a part of a refname; the logic to come
up with the label has been tightened to avoid names that cannot be
used as such.
* The logic to avoid duplicate label names generated by "git rebase
--rebase-merges" forgot that the machinery itself uses "onto" as a
label name, which must be avoided by auto-generated labels, which
has been corrected.
* We have had compatibility fallback macro definitions for "PRIuMAX",
"PRIu32", etc. but did not for "PRIdMAX", while the code used the
last one apparently without any hiccup reported recently. The
fallback macro definitions for these <inttypes.h> macros that must
appear in C99 systems have been removed.
* Recently we have declared that GIT_TEST_* variables take the
usual boolean values (it used to be that some used "non-empty
means true" and taking GIT_TEST_VAR=YesPlease as true); make
sure we notice and fail when non-bool strings are given to
these variables.
* Users of oneway_merge() (like "reset --hard") learned to take
advantage of fsmonitor to avoid unnecessary lstat(2) calls.
* Performance tweak on "git push" into a repository with many refs
that point at objects we have never heard of.
* PerfTest fix to avoid stale result mixed up with the latest round
of test results.
* Hide lower-level verify_signed-buffer() API as a pure helper to
implement the public check_signature() function, in order to
encourage new callers to use the correct and more strict
validation.
* Unnecessary reading of state variables back from the disk during
sequencer operation has been reduced.
* The code has been made to avoid gmtime() and localtime() and prefer
their reentrant counterparts.
* In a repository with many packfiles, the cost of the procedure that
avoids registering the same packfile twice was unnecessarily high
by using an inefficient search algorithm, which has been corrected.
* Redo "git name-rev" to avoid recursive calls.
* FreeBSD CI support via Cirrus-CI has been added.
Fixes since v2.24
-----------------
* "rebase -i" ceased to run post-commit hook by mistake in an earlier
update, which has been corrected.
* "git notes copy $original" ought to copy the notes attached to the
original object to HEAD, but a mistaken tightening to command line
parameter validation made earlier disabled that feature by mistake.
* When all files from some subdirectory were renamed to the root
directory, the directory rename heuristics would fail to detect that
as a rename/merge of the subdirectory to the root directory, which has
been corrected.
* Code clean-up and a bugfix in the logic used to tell worktree local
and repository global refs apart.
* "git stash save" in a working tree that is sparsely checked out
mistakenly removed paths that are outside the area of interest.
* "git rev-parse --git-path HEAD.lock" did not give the right path
when run in a secondary worktree.
* "git merge --no-commit" needs "--no-ff" if you do not want to move
HEAD, which has been corrected in the manual page for "git bisect".
* "git worktree add" internally calls "reset --hard" that should not
descend into submodules, even when submodule.recurse configuration
is set, but it was affected. This has been corrected.
* Messages from die() etc. can be mixed up from multiple processes
without even line buffering on Windows, which has been worked
around.
* HTTP transport had possible allocator/deallocator mismatch, which
has been corrected.
* The watchman integration for fsmonitor was racy, which has been
corrected to be more conservative.
* Fetching from multiple remotes into the same repository in parallel
had a bad interaction with the recent change to (optionally) update
the commit-graph after a fetch job finishes, as these parallel
fetches compete with each other. Which has been corrected.
* Recent update to "git stash pop" made the command empty the index
when run with the "--quiet" option, which has been corrected.
* "git fetch" codepath had a big "do not lazily fetch missing objects
when I ask if something exists" switch. This has been corrected by
marking the "does this thing exist?" calls with "if not please do not
lazily fetch it" flag.
* Test update to avoid wasted cycles.
* Error handling after "git push" finishes sending the packdata and
waits for the response to the remote side has been improved.
* Some codepaths in "gitweb" that forgot to escape URLs generated
based on end-user input have been corrected.
* CI jobs for macOS has been made less chatty when updating perforce
package used during testing.
* "git unpack-objects" used to show progress based only on the number
of received and unpacked objects, which stalled when it has to
handle an unusually large object. It now shows the throughput as
well.
* The sequencer machinery compared the HEAD and the state it is
attempting to commit to decide if the result would be a no-op
commit, even when amending a commit, which was incorrect, and
has been corrected.
* The code to parse GPG output used to assume incorrectly that the
finterprint for the primary key would always be present for a valid
signature, which has been corrected.
* "git submodule status" and "git submodule status --cached" show
different things, but the documentation did not cover them
correctly, which has been corrected.
* "git reset --patch $object" without any pathspec should allow a
tree object to be given, but incorrectly required a committish,
which has been corrected.
* "git submodule status" that is run from a subdirectory of the
superproject did not work well, which has been corrected.
* The revision walking machinery uses resources like per-object flag
bits that need to be reset before a new iteration of walking
begins, but the resources related to topological walk were not
cleared correctly, which has been corrected.
* TravisCI update.
* While running "revert" or "cherry-pick --edit" for multiple
commits, a recent regression incorrectly detected "nothing to
commit, working tree clean", instead of replaying the commits,
which has been corrected.
* Work around a issue where a FD that is left open when spawning a
child process and is kept open in the child can interfere with the
operation in the parent process on Windows.
* One kind of progress messages were always given during commit-graph
generation, instead of following the "if it takes more than two
seconds, show progress" pattern, which has been corrected.
* "git rebase" did not work well when format.useAutoBase
configuration variable is set, which has been corrected.
* The "diff" machinery learned not to lose added/removed blank lines
in the context when --ignore-blank-lines and --function-context are
used at the same time.
* The test on "fast-import" used to get stuck when "fast-import" died
in the middle.
* "git format-patch" can take a set of configured format.notes values
to specify which notes refs to use in the log message part of the
output. The behaviour of this was not consistent with multiple
--notes command line options, which has been corrected.
* "git p4" used to ignore lfs.storage configuration variable, which
has been corrected.
* Assorted fixes to the directory traversal API.
* Forbid pathnames that the platform's filesystem cannot represent on
MinGW.
* "git rebase --signoff" stopped working when the command was written
in C, which has been corrected.
* An earlier update to Git for Windows declared that a tree object is
invalid if it has a path component with backslash in it, which was
overly strict, which has been corrected. The only protection the
Windows users need is to prevent such path (or any path that their
filesystem cannot check out) from entering the index.
* The code to write split commit-graph file(s) upon fetching computed
bogus value for the parameter used in splitting the resulting
files, which has been corrected.
* Other code cleanup, docfix, build fix, etc.
At least with NetBSD (9.99.17) sh(1) the `streaming support for --stdin'
test of t0008-ignores.sh hangs. Force to use bash for tests and add it as a
test dependency. (This needs to be investigate further.)
The entire test suite is now runs but at least on NetBSD 9.99.17
t0301-credential-cache.sh test 29 `use user socket if user directory
exists' and test 30 `use user socket if user directory is a symlink to
a directory' fail.
Changes:
2.24.1
======
This release merges up the fixes that appear in v2.14.6, v2.15.4,
v2.17.3, v2.20.2 and in v2.21.1, addressing the security issues
CVE-2019-1348, CVE-2019-1349, CVE-2019-1350, CVE-2019-1351,
CVE-2019-1352, CVE-2019-1353, CVE-2019-1354, CVE-2019-1387, and
CVE-2019-19604.
* CVE-2019-1348:
The --export-marks option of git fast-import is exposed also via
the in-stream command feature export-marks=... and it allows
overwriting arbitrary paths.
* CVE-2019-1349:
When submodules are cloned recursively, under certain circumstances
Git could be fooled into using the same Git directory twice. We now
require the directory to be empty.
* CVE-2019-1350:
Incorrect quoting of command-line arguments allowed remote code
execution during a recursive clone in conjunction with SSH URLs.
* CVE-2019-1351:
While the only permitted drive letters for physical drives on
Windows are letters of the US-English alphabet, this restriction
does not apply to virtual drives assigned via subst <letter>:
<path>. Git mistook such paths for relative paths, allowing writing
outside of the worktree while cloning.
* CVE-2019-1352:
Git was unaware of NTFS Alternate Data Streams, allowing files
inside the .git/ directory to be overwritten during a clone.
* CVE-2019-1353:
When running Git in the Windows Subsystem for Linux (also known as
"WSL") while accessing a working directory on a regular Windows
drive, none of the NTFS protections were active.
* CVE-2019-1354:
Filenames on Linux/Unix can contain backslashes. On Windows,
backslashes are directory separators. Git did not use to refuse to
write out tracked files with such filenames.
* CVE-2019-1387:
Recursive clones are currently affected by a vulnerability that is
caused by too-lax validation of submodule names, allowing very
targeted attacks via remote code execution in recursive clones.
Credit for finding these vulnerabilities goes to Microsoft Security
Response Center, in particular to Nicolas Joly. The `fast-import`
fixes were provided by Jeff King, the other fixes by Johannes
Schindelin with help from Garima Singh.
* CVE-2019-19604:
The change to disallow `submodule.<name>.update=!command` entries in
`.gitmodules` which was introduced v2.15.4 (and for which v2.17.3
added explicit fsck checks) fixes the vulnerability in v2.20.x where
a recursive clone followed by a submodule update could execute code
contained within the repository without the user explicitly having
asked for that.
Credit for finding this vulnerability goes to Joern Schneeweisz,
credit for the fixes goes to Jonathan Nieder.
Git 2.24 Release Notes
======================
Updates since v2.23
-------------------
Backward compatibility note
* "filter-branch" is showing its age and alternatives are available.
From this release, we started to discourage its use and hint
people about filter-repo.
UI, Workflows & Features
* We now have an active interim maintainer for the Git-Gui part of
the system. Praise and thank Pratyush Yadav for volunteering.
* The command line parser learned "--end-of-options" notation; the
standard convention for scripters to have hardcoded set of options
first on the command line, and force the command to treat end-user
input as non-options, has been to use "--" as the delimiter, but
that would not work for commands that use "--" as a delimiter
between revs and pathspec.
* A mechanism to affect the default setting for a (related) group of
configuration variables is introduced.
* "git fetch" learned "--set-upstream" option to help those who first
clone from their private fork they intend to push to, add the true
upstream via "git remote add" and then "git fetch" from it.
* Device-tree files learned their own userdiff patterns.
(merge 3c81760bc6 sb/userdiff-dts later to maint).
* "git rebase --rebase-merges" learned to drive different merge
strategies and pass strategy specific options to them.
* A new "pre-merge-commit" hook has been introduced.
* Command line completion updates for "git -c var.name=val" have been
added.
* The lazy clone machinery has been taught that there can be more
than one promisor remote and consult them in order when downloading
missing objects on demand.
* The list-objects-filter API (used to create a sparse/lazy clone)
learned to take a combined filter specification.
* The documentation and tests for "git format-patch" have been
cleaned up.
* On Windows, the root level of UNC share is now allowed to be used
just like any other directory.
* The command line completion support (in contrib/) learned about the
"--skip" option of "git revert" and "git cherry-pick".
* "git rebase --keep-base <upstream>" tries to find the original base
of the topic being rebased and rebase on top of that same base,
which is useful when running the "git rebase -i" (and its limited
variant "git rebase -x").
The command also has learned to fast-forward in more cases where it
can instead of replaying to recreate identical commits.
* A configuration variable tells "git fetch" to write the commit
graph after finishing.
* "git add -i" has been taught to show the total number of hunks and
the hunks that has been processed so far when showing prompts.
* "git fetch --jobs=<n>" allowed <n> parallel jobs when fetching
submodules, but this did not apply to "git fetch --multiple" that
fetches from multiple remote repositories. It now does.
* The installation instruction for zsh completion script (in
contrib/) has been a bit improved.
Performance, Internal Implementation, Development Support etc.
* The code to write commit-graph over given commit object names has
been made a bit more robust.
* The first line of verbose output from each test piece now carries
the test name and number to help scanning with eyeballs.
* Further clean-up of the initialization code.
* xmalloc() used to have a mechanism to ditch memory and address
space resources as the last resort upon seeing an allocation
failure from the underlying malloc(), which made the code complex
and thread-unsafe with dubious benefit, as major memory resource
users already do limit their uses with various other mechanisms.
It has been simplified away.
* Unnecessary full-tree diff in "git log -L" machinery has been
optimized away.
* The http transport lacked some optimization the native transports
learned to avoid unnecessary ref advertisement, which has been
corrected.
* Preparation for SHA-256 upgrade continues in the test department.
(merge 0c37c41d13 bc/hash-independent-tests-part-5 later to maint).
* The memory ownership model of the "git fast-import" got
straightened out.
* Output from trace2 subsystem is formatted more prettily now.
* The internal code originally invented for ".gitignore" processing
got reshuffled and renamed to make it less tied to "excluding" and
stress more that it is about "matching", as it has been reused for
things like sparse checkout specification that want to check if a
path is "included".
* "git stash" learned to write refreshed index back to disk.
* Coccinelle checks are done on more source files than before now.
* The cache-tree code has been taught to be less aggressive in
attempting to see if a tree object it computed already exists in
the repository.
* The code to parse and use the commit-graph file has been made more
robust against corrupted input.
* The hg-to-git script (in contrib/) has been updated to work with
Python 3.
* Update the way build artifacts in t/helper/ directory are ignored.
* Preparation for SHA-256 upgrade continues.
* "git log --graph" for an octopus merge is sometimes colored
incorrectly, which is demonstrated and documented but not yet
fixed.
* The trace2 output, when sending them to files in a designated
directory, can populate the directory with too many files; a
mechanism is introduced to set the maximum number of files and
discard further logs when the maximum is reached.
* We have adopted a Code-of-conduct document.
(merge 3f9ef874a7 jk/coc later to maint).
Fixes since v2.23
-----------------
* "git grep --recurse-submodules" that looks at the working tree
files looked at the contents in the index in submodules, instead of
files in the working tree.
(merge 6a289d45c0 mt/grep-submodules-working-tree later to maint).
* Codepaths to walk tree objects have been audited for integer
overflows and hardened.
(merge 5aa02f9868 jk/tree-walk-overflow later to maint).
* "git pack-refs" can lose refs that are created while running, which
is getting corrected.
(merge a613d4f817 sc/pack-refs-deletion-racefix later to maint).
* "git checkout" and "git restore" to re-populate the index from a
tree-ish (typically HEAD) did not work correctly for a path that
was removed and then added again with the intent-to-add bit, when
the corresponding working tree file was empty. This has been
corrected.
* Compilation fix.
(merge 70597e8386 rs/nedalloc-fixlets later to maint).
* "git gui" learned to call the clean-up procedure before exiting.
(merge 0d88f3d2c5 py/git-gui-do-quit later to maint).
* We promoted the "indent heuristics" that decides where to split
diff hunks from experimental to the default a few years ago, but
some stale documentation still marked it as experimental, which has
been corrected.
(merge 64e5e1fba1 sg/diff-indent-heuristic-non-experimental later to maint).
* Fix a mismerge that happened in 2.22 timeframe.
(merge acb7da05ac en/checkout-mismerge-fix later to maint).
* "git archive" recorded incorrect length in extended pax header in
some corner cases, which has been corrected.
(merge 71d41ff651 rs/pax-extended-header-length-fix later to maint).
* On-demand object fetching in lazy clone incorrectly tried to fetch
commits from submodule projects, while still working in the
superproject, which has been corrected.
(merge a63694f523 jt/diff-lazy-fetch-submodule-fix later to maint).
* Prepare get_short_oid() codepath to be thread-safe.
(merge 7cfcb16b0e rs/sort-oid-array-thread-safe later to maint).
* "for-each-ref" and friends that show refs did not protect themselves
against ancient tags that did not record tagger names when asked to
show "%(taggername)", which have been corrected.
(merge 8b3f33ef11 mp/for-each-ref-missing-name-or-email later to maint).
* The "git am" based backend of "git rebase" ignored the result of
updating ".gitattributes" done in one step when replaying
subsequent steps.
(merge 2c65d90f75 bc/reread-attributes-during-rebase later to maint).
* Tell cURL library to use the same malloc() implementation, with the
xmalloc() wrapper, as the rest of the system, for consistency.
(merge 93b980e58f cb/curl-use-xmalloc later to maint).
* Build fix to adjust .gitignore to unignore a path that we started to track.
(merge aac6ff7b5b js/visual-studio later to maint).
* A few implementation fixes in the notes API.
(merge 60fe477a0b mh/notes-duplicate-entries later to maint).
* Fix an earlier regression to "git push --all" which should have
been forbidden when the target remote repository is set to be a
mirror.
(merge 8e4c8af058 tg/push-all-in-mirror-forbidden later to maint).
* Fix an earlier regression in the test suite, which mistakenly
stopped running HTTPD tests.
(merge 3960290675 sg/git-test-boolean later to maint).
* "git rebase --autostash <upstream> <branch>", when <branch> is
different from the current branch, incorrectly moved the tip of the
current branch, which has been corrected.
(merge bf1e28e0ad bw/rebase-autostash-keep-current-branch later to maint).
* Update support for Asciidoctor documentation toolchain.
(merge 83b0b8953e ma/asciidoctor-refmiscinfo later to maint).
* Start using DocBook 5 (instead of DocBook 4.5) as Asciidoctor 2.0
no longer works with the older one.
(merge f6461b82b9 bc/doc-use-docbook-5 later to maint).
* The markup used in user-manual has been updated to work better with
asciidoctor.
(merge c4d2f6143a ma/user-manual-markup-update later to maint).
* Make sure the grep machinery does not abort when seeing a payload
that is not UTF-8 even when JIT is not in use with PCRE1.
(merge ad7c543e3b cb/skip-utf8-check-with-pcre1 later to maint).
* The name of the blob object that stores the filter specification
for sparse cloning/fetching was interpreted in a wrong place in the
code, causing Git to abort.
* "git log --decorate-refs-exclude=<pattern>" was incorrectly
overruled when the "--simplify-by-decoration" option is used, which
has been corrected.
(merge 0cc7380d88 rs/simplify-by-deco-with-deco-refs-exclude later to maint).
* The "upload-pack" (the counterpart of "git fetch") needs to disable
commit-graph when responding to a shallow clone/fetch request, but
the way this was done made Git panic, which has been corrected.
* The object traversal machinery has been optimized not to load tree
objects when we are only interested in commit history.
(merge 72ed80c784 jk/list-objects-optim-wo-trees later to maint).
* The object name parser for "Nth parent" syntax has been made more
robust against integer overflows.
(merge 59fa5f5a25 rs/nth-parent-parse later to maint).
* The code used in following tags in "git fetch" has been optimized.
(merge b7e2d8bca5 ms/fetch-follow-tag-optim later to maint).
* Regression fix for progress output.
(merge 2bb74b53a4 sg/progress-fix later to maint).
* A bug in merge-recursive code that triggers when a branch with a
symbolic link is merged with a branch that replaces it with a
directory has been fixed.
(merge 83e3ad3b12 jt/merge-recursive-symlink-is-not-a-dir-in-way later to maint).
* The rename detection logic sorts a list of rename source candidates
by similarity to pick the best candidate, which means that a tie
between sources with the same similarity is broken by the original
location in the original candidate list (which is sorted by path).
Force the sorting by similarity done with a stable sort, which is
not promised by system supplied qsort(3), to ensure consistent
results across platforms.
(merge 2049b8dc65 js/diff-rename-force-stable-sort later to maint).
* The code to skip "UTF" and "UTF-" prefix, when computing an advice
message, did not work correctly when the prefix was "UTF", which
has been fixed.
(merge b181676ce9 rs/convert-fix-utf-without-dash later to maint).
* The author names taken from SVN repositories may have extra leading
or trailing whitespaces, which are now munged away.
(merge 4ddd4bddb1 tk/git-svn-trim-author-name later to maint).
* "git rebase -i" showed a wrong HEAD while "reword" open the editor.
(merge b0a3186140 pw/rebase-i-show-HEAD-to-reword later to maint).
* A few simplification and bugfixes to PCRE interface.
(merge c581e4a749 ab/pcre-jit-fixes later to maint).
* PCRE fixes.
(merge ff61681b46 cb/pcre1-cleanup later to maint).
* "git range-diff" segfaulted when diff.noprefix configuration was
used, as it blindly expected the patch it internally generates to
have the standard a/ and b/ prefixes. The command now forces the
internal patch to be built without any prefix, not to be affected
by any end-user configuration.
(merge 937b76ed49 js/range-diff-noprefix later to maint).
* "git stash apply" in a subdirectory of a secondary worktree failed
to access the worktree correctly, which has been corrected.
(merge dfd557c978 js/stash-apply-in-secondary-worktree later to maint).
* The merge-recursive machinery is one of the most complex parts of
the system that accumulated cruft over time. This large series
cleans up the implementation quite a bit.
(merge b657047719 en/merge-recursive-cleanup later to maint).
* Pretty-printed command line formatter (used in e.g. reporting the
command being run by the tracing API) had a bug that lost an
argument that is an empty string, which has been corrected.
(merge ce2d7ed2fd gs/sq-quote-buf-pretty later to maint).
* "git range-diff" failed to handle mode-only change, which has been
corrected.
(merge 2b6a9b13ca tg/range-diff-output-update later to maint).
* Dev support update.
(merge 4f3c1dc5d6 dl/allow-running-cocci-verbosely later to maint).
* "git format-patch -o <outdir>" did an equivalent of "mkdir <outdir>"
not "mkdir -p <outdir>", which was corrected.
* "git stash save" lost local changes to submodules, which has been
corrected.
(merge 556895d0c8 jj/stash-reset-only-toplevel later to maint).
* The atomic push over smart HTTP transport did not work, which has
been corrected.
(merge 6f1194246a bc/smart-http-atomic-push later to maint).
* Other code cleanup, docfix, build fix, etc.
Git 2.23 Release Notes
======================
Updates since v2.22
-------------------
Backward compatibility note
* The "--base" option of "format-patch" computed the patch-ids for
prerequisite patches in an unstable way, which has been updated to
compute in a way that is compatible with "git patch-id --stable".
* The "git log" command by default behaves as if the --mailmap option
was given.
UI, Workflows & Features
* The "git fast-export/import" pair has been taught to handle commits
with log messages in encoding other than UTF-8 better.
* In recent versions of Git, per-worktree refs are exposed in
refs/worktrees/<wtname>/ hierarchy, which means that worktree names
must be a valid refname component. The code now sanitizes the names
given to worktrees, to make sure these refs are well-formed.
* "git merge" learned "--quit" option that cleans up the in-progress
merge while leaving the working tree and the index still in a mess.
* "git format-patch" learns a configuration to set the default for
its --notes=<ref> option.
* The code to show args with potential typo that cannot be
interpreted as a commit-ish has been improved.
* "git clone --recurse-submodules" learned to set up the submodules
to ignore commit object names recorded in the superproject gitlink
and instead use the commits that happen to be at the tip of the
remote-tracking branches from the get-go, by passing the new
"--remote-submodules" option.
* The pattern "git diff/grep" use to extract funcname and words
boundary for Matlab has been extend to cover Octave, which is more
or less equivalent.
* "git help git" was hard to discover (well, at least for some
people).
* The pattern "git diff/grep" use to extract funcname and words
boundary for Rust has been added.
* "git status" can be told a non-standard default value for the
"--[no-]ahead-behind" option with a new configuration variable
status.aheadBehind.
* "git fetch" and "git pull" reports when a fetch results in
non-fast-forward updates to let the user notice unusual situation.
The commands learned "--no-show-forced-updates" option to disable
this safety feature.
* Two new commands "git switch" and "git restore" are introduced to
split "checking out a branch to work on advancing its history" and
"checking out paths out of the index and/or a tree-ish to work on
advancing the current history" out of the single "git checkout"
command.
* "git branch --list" learned to always output the detached HEAD as
the first item (when the HEAD is detached, of course), regardless
of the locale.
* The conditional inclusion mechanism learned to base the choice on
the branch the HEAD currently is on.
* "git rev-list --objects" learned the "--no-object-names" option to
squelch the path to the object that is used as a grouping hint for
pack-objects.
* A new tag.gpgSign configuration variable turns "git tag -a" into
"git tag -s".
* "git multi-pack-index" learned expire and repack subcommands.
* "git blame" learned to "ignore" commits in the history, whose
effects (as well as their presence) get ignored.
* "git cherry-pick/revert" learned a new "--skip" action.
* The tips of refs from the alternate object store can be used as
starting point for reachability computation now.
* Extra blank lines in "git status" output have been reduced.
* The commits in a repository can be described by multiple
commit-graph files now, which allows the commit-graph files to be
updated incrementally.
* "git range-diff" output has been tweaked for easier identification
of which part of what file the patch shown is about.
Performance, Internal Implementation, Development Support etc.
* Update supporting parts of "git rebase" to remove code that should
no longer be used.
* Developer support to emulate unsatisfied prerequisites in tests to
ensure that the remainder of the tests still succeeds when tests
with prerequisites are skipped.
* "git update-server-info" learned not to rewrite the file with the
same contents.
* The way of specifying the path to find dynamic libraries at runtime
has been simplified. The old default to pass -R/path/to/dir has been
replaced with the new default to pass -Wl,-rpath,/path/to/dir,
which is the more recent GCC uses. Those who need to build with an
old GCC can still use "CC_LD_DYNPATH=-R"
* Prepare use of reachability index in topological walker that works
on a range (A..B).
* A new tutorial targeting specifically aspiring git-core
developers has been added.
* Auto-detect how to tell HP-UX aCC where to use dynamically linked
libraries from at runtime.
* "git mergetool" and its tests now spawn fewer subprocesses.
* Dev support update to help tracing out tests.
* Support to build with MSVC has been updated.
* "git fetch" that grabs from a group of remotes learned to run the
auto-gc only once at the very end.
* A handful of Windows build patches have been upstreamed.
* The code to read state files used by the sequencer machinery for
"git status" has been made more robust against a corrupt or stale
state files.
* "git for-each-ref" with multiple patterns have been optimized.
* The tree-walk API learned to pass an in-core repository
instance throughout more codepaths.
* When one step in multi step cherry-pick or revert is reset or
committed, the command line prompt script failed to notice the
current status, which has been improved.
* Many GIT_TEST_* environment variables control various aspects of
how our tests are run, but a few followed "non-empty is true, empty
or unset is false" while others followed the usual "there are a few
ways to spell true, like yes, on, etc., and also ways to spell
false, like no, off, etc." convention.
* Adjust the dir-iterator API and apply it to the local clone
optimization codepath.
* We have been trying out a few language features outside c89; the
coding guidelines document did not talk about them and instead had
a blanket ban against them.
* A test helper has been introduced to optimize preparation of test
repositories with many simple commits, and a handful of test
scripts have been updated to use it.
Fixes since v2.22
-----------------
* A relative pathname given to "git init --template=<path> <repo>"
ought to be relative to the directory "git init" gets invoked in,
but it instead was made relative to the repository, which has been
corrected.
* "git worktree add" used to fail when another worktree connected to
the same repository was corrupt, which has been corrected.
* The ownership rule for the file descriptor to fast-import remote
backend was mixed up, leading to an unrelated file descriptor getting
closed, which has been fixed.
* A "merge -c" instruction during "git rebase --rebase-merges" should
give the user a chance to edit the log message, even when there is
otherwise no need to create a new merge and replace the existing
one (i.e. fast-forward instead), but did not. Which has been
corrected.
* Code cleanup and futureproof.
* More parameter validation.
* "git update-server-info" used to leave stale packfiles in its
output, which has been corrected.
* The server side support for "git fetch" used to show incorrect
value for the HEAD symbolic ref when the namespace feature is in
use, which has been corrected.
* "git am -i --resolved" segfaulted after trying to see a commit as
if it were a tree, which has been corrected.
* "git bundle verify" needs to see if prerequisite objects exist in
the receiving repository, but the command did not check if we are
in a repository upfront, which has been corrected.
* "git merge --squash" is designed to update the working tree and the
index without creating the commit, and this cannot be countermanded
by adding the "--commit" option; the command now refuses to work
when both options are given.
* The data collected by fsmonitor was not properly written back to
the on-disk index file, breaking t7519 tests occasionally, which
has been corrected.
* Update to Unicode 12.1 width table.
* The command line to invoke a "git cat-file" command from inside
"git p4" was not properly quoted to protect a caret and running a
broken command on Windows, which has been corrected.
* "git request-pull" learned to warn when the ref we ask them to pull
from in the local repository and in the published repository are
different.
* When creating a partial clone, the object filtering criteria is
recorded for the origin of the clone, but this incorrectly used a
hardcoded name "origin" to name that remote; it has been corrected
to honor the "--origin <name>" option.
* "git fetch" into a lazy clone forgot to fetch base objects that are
necessary to complete delta in a thin packfile, which has been
corrected.
* The filter_data used in the list-objects-filter (which manages a
lazily sparse clone repository) did not use the dynamic array API
correctly---'nr' is supposed to point at one past the last element
of the array in use. This has been corrected.
* The description about slashes in gitignore patterns (used to
indicate things like "anchored to this level only" and "only
matches directories") has been revamped.
* The URL decoding code has been updated to avoid going past the end
of the string while parsing %-<hex>-<hex> sequence.
* The list of for-each like macros used by clang-format has been
updated.
* "git branch --list" learned to show branches that are checked out
in other worktrees connected to the same repository prefixed with
'+', similar to the way the currently checked out branch is shown
with '*' in front.
(merge 6e9381469e nb/branch-show-other-worktrees-head later to maint).
* Code restructuring during 2.20 period broke fetching tags via
"import" based transports.
* The commit-graph file is now part of the "files that the runtime
may keep open file descriptors on, all of which would need to be
closed when done with the object store", and the file descriptor to
an existing commit-graph file now is closed before "gc" finalizes a
new instance to replace it.
* "git checkout -p" needs to selectively apply a patch in reverse,
which did not work well.
* Code clean-up to avoid signed integer wraparounds during binary search.
* "git interpret-trailers" always treated '#' as the comment
character, regardless of core.commentChar setting, which has been
corrected.
* "git stash show 23" used to work, but no more after getting
rewritten in C; this regression has been corrected.
* "git rebase --abort" used to leave refs/rewritten/ when concluding
"git rebase -r", which has been corrected.
* An incorrect list of options was cached after command line
completion failed (e.g. trying to complete a command that requires
a repository outside one), which has been corrected.
* The code to parse scaled numbers out of configuration files has
been made more robust and also easier to follow.
* The codepath to compute delta islands used to spew progress output
without giving the callers any way to squelch it, which has been
fixed.
* Protocol capabilities that go over wire should never be translated,
but it was incorrectly marked for translation, which has been
corrected. The output of protocol capabilities for debugging has
been tweaked a bit.
* Use "Erase in Line" CSI sequence that is already used in the editor
support to clear cruft in the progress output.
* "git submodule foreach" did not protect command line options passed
to the command to be run in each submodule correctly, when the
"--recursive" option was in use.
* The configuration variable rebase.rescheduleFailedExec should be
effective only while running an interactive rebase and should not
affect anything when running a non-interactive one, which was not
the case. This has been corrected.
* The "git clone" documentation refers to command line options in its
description in the short form; they have been replaced with long
forms to make them more recognisable.
* Generation of pack bitmaps are now disabled when .keep files exist,
as these are mutually exclusive features.
(merge 7328482253 ew/repack-with-bitmaps-by-default later to maint).
* "git rm" to resolve a conflicted path leaked an internal message
"needs merge" before actually removing the path, which was
confusing. This has been corrected.
* "git stash --keep-index" did not work correctly on paths that have
been removed, which has been fixed.
(merge b932f6a5e8 tg/stash-keep-index-with-removed-paths later to maint).
* Window 7 update ;-)
* A codepath that reads from GPG for signed object verification read
past the end of allocated buffer, which has been fixed.
* "git clean" silently skipped a path when it cannot lstat() it; now
it gives a warning.
* "git push --atomic" that goes over the transport-helper (namely,
the smart http transport) failed to prevent refs to be pushed when
it can locally tell that one of the ref update will fail without
having to consult the other end, which has been corrected.
* The internal diff machinery can be made to read out of bounds while
looking for --function-context line in a corner case, which has been
corrected.
(merge b777f3fd61 jk/xdiff-clamp-funcname-context-index later to maint).
* Other code cleanup, docfix, build fix, etc.
(merge fbec05c210 cc/test-oidmap later to maint).
(merge 7a06fb038c jk/no-system-includes-in-dot-c later to maint).
(merge 81ed2b405c cb/xdiff-no-system-includes-in-dot-c later to maint).
(merge d61e6ce1dd sg/fsck-config-in-doc later to maint).
Git 2.21 Release Notes
======================
Backward Compatibility Notes
----------------------------
* Historically, the "-m" (mainline) option can only be used for "git
cherry-pick" and "git revert" when working with a merge commit.
This version of Git no longer warns or errors out when working with
a single-parent commit, as long as the argument to the "-m" option
is 1 (i.e. it has only one parent, and the request is to pick or
revert relative to that first parent). Scripts that relied on the
behaviour may get broken with this change.
Updates since v2.20
-------------------
UI, Workflows & Features
* The "http.version" configuration variable can be used with recent
enough versions of cURL library to force the version of HTTP used
to talk when fetching and pushing.
* Small fixes and features for fast-export and fast-import, mostly on
the fast-export side has been made.
* "git push $there $src:$dst" rejects when $dst is not a fully
qualified refname and it is not clear what the end user meant. The
codepath has been taught to give a clearer error message, and also
guess where the push should go by taking the type of the pushed
object into account (e.g. a tag object would want to go under
refs/tags/).
* "git checkout [<tree-ish>] path..." learned to report the number of
paths that have been checked out of the index or the tree-ish,
which gives it the same degree of noisy-ness as the case in which
the command checks out a branch. "git checkout -m <pathspec>" to
undo conflict resolution gives a similar message.
* "git quiltimport" learned "--keep-non-patch" option.
* "git worktree remove" and "git worktree move" refused to work when
there is a submodule involved. This has been loosened to ignore
uninitialized submodules.
* "git cherry-pick -m1" was forbidden when picking a non-merge
commit, even though there _is_ parent number 1 for such a commit.
This was done to avoid mistakes back when "cherry-pick" was about
picking a single commit, but is no longer useful with "cherry-pick"
that can pick a range of commits. Now the "-m$num" option is
allowed when picking any commit, as long as $num names an existing
parent of the commit.
* Update "git multimail" from the upstream.
* "git p4" update.
* The "--format=<placeholder>" option of for-each-ref, branch and tag
learned to show a few more traits of objects that can be learned by
the object_info API.
* "git rebase -i" learned to re-execute a command given with 'exec'
to run after it failed the last time.
* "git diff --color-moved-ws" updates.
* Custom userformat "log --format" learned %S atom that stands for
the tip the traversal reached the commit from, i.e. --source.
* "git instaweb" learned to drive http.server that comes with
"batteries included" Python installation (both Python2 & 3).
* A new encoding UTF-16LE-BOM has been invented to force encoding to
UTF-16 with BOM in little endian byte order, which cannot be directly
generated by using iconv.
* A new date format "--date=human" that morphs its output depending
on how far the time is from the current time has been introduced.
"--date=auto:human" can be used to use this new format (or any
existing format) when the output is going to the pager or to the
terminal, and otherwise the default format.
Performance, Internal Implementation, Development Support etc.
* Code clean-up with optimization for the codepath that checks
(non-)existence of loose objects.
* More codepaths have become aware of working with in-core repository
instances other than the default "the_repository".
* The "strncat()" function is now among the banned functions.
* Portability updates for the HPE NonStop platform.
* Earlier we added "-Wformat-security" to developer builds, assuming
that "-Wall" (which includes "-Wformat" which in turn is required
to use "-Wformat-security") is always in effect. This is not true
when config.mak.autogen is in use, unfortunately. This has been
fixed by unconditionally adding "-Wall" to developer builds.
* The loose object cache used to optimize existence look-up has been
updated.
* Flaky tests can now be repeatedly run under load with the
"--stress" option.
* Documentation/Makefile is getting prepared for manpage
localization.
* "git fetch-pack" now can talk the version 2 protocol.
* sha-256 hash has been added and plumbed through the code to allow
building Git with the "NewHash".
* Debugging help for http transport.
* "git fetch --deepen=<more>" has been corrected to work over v2
protocol.
* The code to walk tree objects has been taught that we may be
working with object names that are not computed with SHA-1.
* The in-core repository instances are passed through more codepaths.
* Update the protocol message specification to allow only the limited
use of scaled quantities. This is to ensure potential compatibility
issues will not get out of hand.
* Micro-optimize the code that prepares commit objects to be walked
by "git rev-list" when the commit-graph is available.
* "git fetch" and "git upload-pack" learned to send all exchanges over
the sideband channel while talking the v2 protocol.
* The codepath to write out commit-graph has been optimized by
following the usual pattern of visiting objects in in-pack order.
* The codepath to show progress meter while writing out commit-graph
file has been improved.
* Cocci rules have been updated to encourage use of strbuf_addbuf().
* "git rebase --merge" has been reimplemented by reusing the internal
machinery used for "git rebase -i".
* More code in "git bisect" has been rewritten in C.
* Instead of going through "git-rebase--am" scriptlet to use the "am"
backend, the built-in version of "git rebase" learned to drive the
"am" backend directly.
* The assumption to work on the single "in-core index" instance has
been reduced from the library-ish part of the codebase.
* The test lint learned to catch non-portable "sed" options.
* "git pack-objects" learned another algorithm to compute the set of
objects to send, that trades the resulting packfile off to save
traversal cost to favor small pushes.
* The travis CI scripts have been corrected to build Git with the
compiler(s) of our choice.
* "git submodule update" learned to abort early when core.worktree
for the submodule is not set correctly to prevent spreading damage.
* Test suite has been adjusted to run on Azure Pipeline.
* Running "Documentation/doc-diff x" from anywhere other than the
top-level of the working tree did not show the usage string
correctly, which has been fixed.
* Use of the sparse tool got easier to customize from the command
line to help developers.
* A new target "coverage-prove" to run the coverage test under
"prove" has been added.
* A flakey "p4" test has been removed.
* The code and tests assume that the system supplied iconv() would
always use BOM in its output when asked to encode to UTF-16 (or
UTF-32), but apparently some implementations output big-endian
without BOM. A compile-time knob has been added to help such
systems (e.g. NonStop) to add BOM to the output to increase
portability.
Fixes since v2.20
-----------------
* Updates for corner cases in merge-recursive.
(merge cc4cb0902c en/merge-path-collision later to maint).
* "git checkout frotz" (without any double-dash) avoids ambiguity by
making sure 'frotz' cannot be interpreted as a revision and as a
path at the same time. This safety has been updated to check also
a unique remote-tracking branch 'frotz' in a remote, when dwimming
to create a local branch 'frotz' out of a remote-tracking branch
'frotz' from a remote.
(merge be4908f103 nd/checkout-dwim-fix later to maint).
* Refspecs configured with "git -c var=val clone" did not propagate
to the resulting repository, which has been corrected.
(merge 7eae4a3ac4 sg/clone-initial-fetch-configuration later to maint).
* A properly configured username/email is required under
user.useConfigOnly in order to create commits; now "git stash"
(even though it creates commit objects to represent stash entries)
command is exempt from the requirement.
(merge 3bc2111fc2 sd/stash-wo-user-name later to maint).
* The http-backend CGI process did not correctly clean up the child
processes it spawns to run upload-pack etc. when it dies itself,
which has been corrected.
(merge 02818a98d7 mk/http-backend-kill-children-before-exit later to maint).
* "git rev-list --exclude-promisor-objects" had to take an object
that does not exist locally (and is lazily available) from the
command line without barfing, but the code dereferenced NULL.
(merge 4cf67869b2 md/list-lazy-objects-fix later to maint).
* The traversal over tree objects has learned to honor
":(attr:label)" pathspec match, which has been implemented only for
enumerating paths on the filesystem.
(merge 5a0b97b34c nd/attr-pathspec-in-tree-walk later to maint).
* BSD port updates.
(merge 4e3ecbd439 cb/openbsd-allows-reading-directory later to maint).
(merge b6bdc2a0f5 cb/t5004-empty-tar-archive-fix later to maint).
(merge 82cbc8cde2 cb/test-lint-cp-a later to maint).
* Lines that begin with a certain keyword that come over the wire, as
well as lines that consist only of one of these keywords, ought to
be painted in color for easier eyeballing, but the latter was
broken ever since the feature was introduced in 2.19, which has
been corrected.
(merge 1f67290450 hn/highlight-sideband-keywords later to maint).
* "git log -G<regex>" looked for a hunk in the "git log -p" patch
output that contained a string that matches the given pattern.
Optimize this code to ignore binary files, which by default will
not show any hunk that would match any pattern (unless textconv or
the --text option is in effect, that is).
(merge e0e7cb8080 tb/log-G-binary later to maint).
* "git submodule update" ought to use a single job unless asked, but
by mistake used multiple jobs, which has been fixed.
(merge e3a9d1aca9 sb/submodule-fetchjobs-default-to-one later to maint).
* "git stripspace" should be usable outside a git repository, but
under the "-s" or "-c" mode, it didn't.
(merge 957da75802 jn/stripspace-wo-repository later to maint).
* Some of the documentation pages formatted incorrectly with
Asciidoctor, which have been fixed.
(merge b62eb1d2f4 ma/asciidoctor later to maint).
* The core.worktree setting in a submodule repository should not be
pointing at a directory when the submodule loses its working tree
(e.g. getting deinit'ed), but the code did not properly maintain
this invariant.
* With zsh, "git cmd path<TAB>" was completed to "git cmd path name"
when the completed path has a special character like SP in it,
without any attempt to keep "path name" a single filename. This
has been fixed to complete it to "git cmd path\ name" just like
Bash completion does.
* The test suite tried to see if it is run under bash, but the check
itself failed under some other implementations of shell (notably
under NetBSD). This has been corrected.
(merge 54ea72f09c sg/test-bash-version-fix later to maint).
* "git gc" and "git repack" did not close the open packfiles that
they found unneeded before removing them, which didn't work on a
platform incapable of removing an open file. This has been
corrected.
(merge 5bdece0d70 js/gc-repack-close-before-remove later to maint).
* The code to drive GIT_EXTERNAL_DIFF command relied on the string
returned from getenv() to be non-volatile, which is not true, that
has been corrected.
(merge 6776a84dae kg/external-diff-save-env later to maint).
* There were many places the code relied on the string returned from
getenv() to be non-volatile, which is not true, that have been
corrected.
(merge 0da0e9268b jk/save-getenv-result later to maint).
* The v2 upload-pack protocol implementation failed to honor
hidden-ref configuration, which has been corrected.
(merge e20b4192a3 jk/proto-v2-hidden-refs-fix later to maint).
* "git fetch --recurse-submodules" may not fetch the necessary commit
that is bound to the superproject, which is getting corrected.
(merge be76c21282 sb/submodule-recursive-fetch-gets-the-tip later to maint).
* "git rebase" internally runs "checkout" to switch between branches,
and the command used to call the post-checkout hook, but the
reimplementation stopped doing so, which is getting fixed.
* "git add -e" got confused when the change it wants to let the user
edit is smaller than the previous change that was left over in a
temporary file.
(merge fa6f225e01 js/add-e-clear-patch-before-stating later to maint).
* "git p4" failed to update a shelved change when there were moved
files, which has been corrected.
(merge 7a10946ab9 ld/git-p4-shelve-update-fix later to maint).
* The codepath to read from the commit-graph file attempted to read
past the end of it when the file's table-of-contents was corrupt.
* The compat/obstack code had casts that -Wcast-function-type
compilation option found questionable.
(merge 764473d257 sg/obstack-cast-function-type-fix later to maint).
* An obvious typo in an assertion error message has been fixed.
(merge 3c27e2e059 cc/test-ref-store-typofix later to maint).
* In Git for Windows, "git clone \\server\share\path" etc. that uses
UNC paths from command line had bad interaction with its shell
emulation.
* "git add --ignore-errors" did not work as advertised and instead
worked as an unintended synonym for "git add --renormalize", which
has been fixed.
(merge e2c2a37545 jk/add-ignore-errors-bit-assignment-fix later to maint).
* On a case-insensitive filesystem, we failed to compare the part of
the path that is above the worktree directory in an absolute
pathname, which has been corrected.
* Asking "git check-attr" about a macro (e.g. "binary") on a specific
path did not work correctly, even though "git check-attr -a" listed
such a macro correctly. This has been corrected.
(merge 7b95849be4 jk/attr-macro-fix later to maint).
* "git pack-objects" incorrectly used uninitialized mutex, which has
been corrected.
(merge edb673cf10 ph/pack-objects-mutex-fix later to maint).
* "git checkout -b <new> [HEAD]" to create a new branch from the
current commit and check it out ought to be a no-op in the index
and the working tree in normal cases, but there are corner cases
that do require updates to the index and the working tree. Running
it immediately after "git clone --no-checkout" is one of these
cases that an earlier optimization kicked in incorrectly, which has
been fixed.
(merge 8424bfd45b bp/checkout-new-branch-optim later to maint).
* "git diff --color-moved --cc --stat -p" did not work well due to
funny interaction between a bug in color-moved and the rest, which
has been fixed.
(merge dac03b5518 jk/diff-cc-stat-fixes later to maint).
* When GIT_SEQUENCE_EDITOR is set, the command was incorrectly
started when modes of "git rebase" that implicitly uses the
machinery for the interactive rebase are run, which has been
corrected.
(merge 891d4a0313 pw/no-editor-in-rebase-i-implicit later to maint).
* The commit-graph facility did not work when in-core objects that
are promoted from unknown type to commit (e.g. a commit that is
accessed via a tag that refers to it) were involved, which has been
corrected.
(merge 4468d4435c sg/object-as-type-commit-graph-fix later to maint).
* "git fetch" output cleanup.
(merge dc40b24df4 nd/fetch-compact-update later to maint).
* "git cat-file --batch" reported a dangling symbolic link by
mistake, when it wanted to report that a given name is ambiguous.
* Documentation around core.crlf has been updated.
(merge c9446f0504 jk/autocrlf-overrides-eol-doc later to maint).
* The documentation of "git commit-tree" said that the command
understands "--gpg-sign" in addition to "-S", but the command line
parser did not know about the longhand, which has been corrected.
* "git rebase -x $cmd" did not reject multi-line command, even though
the command is incapable of handling such a command. It now is
rejected upfront.
(merge c762aada1a pw/rebase-x-sanity-check later to maint).
* Output from "git help" was not correctly aligned, which has been
fixed.
(merge 6195a76da4 nd/help-align-command-desc later to maint).
* The "git submodule summary" subcommand showed shortened commit
object names by mechanically truncating them at 7-hexdigit, which
has been improved to let "rev-parse --short" scale the length of
the abbreviation with the size of the repository.
(merge 0586a438f6 sh/submodule-summary-abbrev-fix later to maint).
* The way the OSX build jobs updates its build environment used the
"--quiet" option to "brew update" command, but it wasn't all that
quiet to be useful. The use of the option has been replaced with
an explicit redirection to the /dev/null (which incidentally would
have worked around a breakage by recent updates to homebrew, which
has fixed itself already).
(merge a1ccaedd62 sg/travis-osx-brew-breakage-workaround later to maint).
* "git --work-tree=$there --git-dir=$here describe --dirty" did not
work correctly as it did not pay attention to the location of the
worktree specified by the user by mistake, which has been
corrected.
(merge c801170b0c ss/describe-dirty-in-the-right-directory later to maint).
* "git fetch" over protocol v2 that needs to make a second connection
to backfill tags did not clear a variable that holds shallow
repository information correctly, leading to an access of freed
piece of memory.
* Some errors from the other side coming over smart HTTP transport
were not noticed, which has been corrected.
* Code cleanup, docfix, build fix, etc.
Git v2.20.1:
This release is primarily to fix brown-paper-bag breakages in the
2.20.0 release.
Fixes since v2.20
* A few newly added tests were not portable and caused minority
platforms to report false breakages, which have been fixed.
* Portability fix for a recent update to parse-options API.
* "git help -a" did not work well when an overly long alias is
defined, which has been corrected.
* A recent update accidentally squelched an error message when the
run_command API failed to run a missing command, which has been
corrected.
Git 2.19 Release Notes
Updates since v2.18
-------------------
UI, Workflows & Features
* "git diff" compares the index and the working tree. For paths
added with intent-to-add bit, the command shows the full contents
of them as added, but the paths themselves were not marked as new
files. They are now shown as new by default.
"git apply" learned the "--intent-to-add" option so that an
otherwise working-tree-only application of a patch will add new
paths to the index marked with the "intent-to-add" bit.
* "git grep" learned the "--column" option that gives not just the
line number but the column number of the hit.
* The "-l" option in "git branch -l" is an unfortunate short-hand for
"--create-reflog", but many users, both old and new, somehow expect
it to be something else, perhaps "--list". This step warns when "-l"
is used as a short-hand for "--create-reflog" and warns about the
future repurposing of the it when it is used.
* The userdiff pattern for .php has been updated.
* The content-transfer-encoding of the message "git send-email" sends
out by default was 8bit, which can cause trouble when there is an
overlong line to bust RFC 5322/2822 limit. A new option 'auto' to
automatically switch to quoted-printable when there is such a line
in the payload has been introduced and is made the default.
* "git checkout" and "git worktree add" learned to honor
checkout.defaultRemote when auto-vivifying a local branch out of a
remote tracking branch in a repository with multiple remotes that
have tracking branches that share the same names.
(merge 8d7b558bae ab/checkout-default-remote later to maint).
* "git grep" learned the "--only-matching" option.
* "git rebase --rebase-merges" mode now handles octopus merges as
well.
* Add a server-side knob to skip commits in exponential/fibbonacci
stride in an attempt to cover wider swath of history with a smaller
number of iterations, potentially accepting a larger packfile
transfer, instead of going back one commit a time during common
ancestor discovery during the "git fetch" transaction.
(merge 42cc7485a2 jt/fetch-negotiator-skipping later to maint).
* A new configuration variable core.usereplacerefs has been added,
primarily to help server installations that want to ignore the
replace mechanism altogether.
* Teach "git tag -s" etc. a few configuration variables (gpg.format
that can be set to "openpgp" or "x509", and gpg.<format>.program
that is used to specify what program to use to deal with the format)
to allow x.509 certs with CMS via "gpgsm" to be used instead of
openpgp via "gnupg".
* Many more strings are prepared for l10n.
* "git p4 submit" learns to ask its own pre-submit hook if it should
continue with submitting.
* The test performed at the receiving end of "git push" to prevent
bad objects from entering repository can be customized via
receive.fsck.* configuration variables; we now have gained a
counterpart to do the same on the "git fetch" side, with
fetch.fsck.* configuration variables.
* "git pull --rebase=interactive" learned "i" as a short-hand for
"interactive".
* "git instaweb" has been adjusted to run better with newer Apache on
RedHat based distros.
* "git range-diff" is a reimplementation of "git tbdiff" that lets us
compare individual patches in two iterations of a topic.
* The sideband code learned to optionally paint selected keywords at
the beginning of incoming lines on the receiving end.
* "git branch --list" learned to take the default sort order from the
'branch.sort' configuration variable, just like "git tag --list"
pays attention to 'tag.sort'.
* "git worktree" command learned "--quiet" option to make it less
verbose.
Performance, Internal Implementation, Development Support etc.
* The bulk of "git submodule foreach" has been rewritten in C.
* The in-core "commit" object had an all-purpose "void *util" field,
which was tricky to use especially in library-ish part of the
code. All of the existing uses of the field has been migrated to a
more dedicated "commit-slab" mechanism and the field is eliminated.
* A less often used command "git show-index" has been modernized.
(merge fb3010c31f jk/show-index later to maint).
* The conversion to pass "the_repository" and then "a_repository"
throughout the object access API continues.
* Continuing with the idea to programatically enumerate various
pieces of data required for command line completion, teach the
codebase to report the list of configuration variables
subcommands care about to help complete them.
* Separate "rebase -p" codepath out of "rebase -i" implementation to
slim down the latter and make it easier to manage.
* Make refspec parsing codepath more robust.
* Some flaky tests have been fixed.
* Continuing with the idea to programmatically enumerate various
pieces of data required for command line completion, the codebase
has been taught to enumerate options prefixed with "--no-" to
negate them.
* Build and test procedure for netrc credential helper (in contrib/)
has been updated.
* Remove unused function definitions and declarations from ewah
bitmap subsystem.
* Code preparation to make "git p4" closer to be usable with Python 3.
* Tighten the API to make it harder to misuse in-tree .gitmodules
file, even though it shares the same syntax with configuration
files, to read random configuration items from it.
* "git fast-import" has been updated to avoid attempting to create
delta against a zero-byte-long string, which is pointless.
* The codebase has been updated to compile cleanly with -pedantic
option.
(merge 2b647a05d7 bb/pedantic later to maint).
* The character display width table has been updated to match the
latest Unicode standard.
(merge 570951eea2 bb/unicode-11-width later to maint).
* test-lint now looks for broken use of "VAR=VAL shell_func" in test
scripts.
* Conversion from uchar[40] to struct object_id continues.
* Recent "security fix" to pay attention to contents of ".gitmodules"
while accepting "git push" was a bit overly strict than necessary,
which has been adjusted.
* "git fsck" learns to make sure the optional commit-graph file is in
a sane state.
* "git diff --color-moved" feature has further been tweaked.
* Code restructuring and a small fix to transport protocol v2 during
fetching.
* Parsing of -L[<N>][,[<M>]] parameters "git blame" and "git log"
take has been tweaked.
* lookup_commit_reference() and friends have been updated to find
in-core object for a specific in-core repository instance.
* Various glitches in the heuristics of merge-recursive strategy have
been documented in new tests.
* "git fetch" learned a new option "--negotiation-tip" to limit the
set of commits it tells the other end as "have", to reduce wasted
bandwidth and cycles, which would be helpful when the receiving
repository has a lot of refs that have little to do with the
history at the remote it is fetching from.
* For a large tree, the index needs to hold many cache entries
allocated on heap. These cache entries are now allocated out of a
dedicated memory pool to amortize malloc(3) overhead.
* Tests to cover various conflicting cases have been added for
merge-recursive.
* Tests to cover conflict cases that involve submodules have been
added for merge-recursive.
* Look for broken "&&" chains that are hidden in subshell, many of
which have been found and corrected.
* The singleton commit-graph in-core instance is made per in-core
repository instance.
* "make DEVELOPER=1 DEVOPTS=pedantic" allows developers to compile
with -pedantic option, which may catch more problematic program
constructs and potential bugs.
* Preparatory code to later add json output for telemetry data has
been added.
* Update the way we use Coccinelle to find out-of-style code that
need to be modernised.
* It is too easy to misuse system API functions such as strcat();
these selected functions are now forbidden in this codebase and
will cause a compilation failure.
* Add a script (in contrib/) to help users of VSCode work better with
our codebase.
* The Travis CI scripts were taught to ship back the test data from
failed tests.
(merge aea8879a6a sg/travis-retrieve-trash-upon-failure later to maint).
* The parse-options machinery learned to refrain from enclosing
placeholder string inside a "<bra" and "ket>" pair automatically
without PARSE_OPT_LITERAL_ARGHELP. Existing help text for option
arguments that are not formatted correctly have been identified and
fixed.
(merge 5f0df44cd7 rs/parse-opt-lithelp later to maint).
* Noiseword "extern" has been removed from function decls in the
header files.
* A few atoms like %(objecttype) and %(objectsize) in the format
specifier of "for-each-ref --format=<format>" can be filled without
getting the full contents of the object, but just with the object
header. These cases have been optimized by calling
oid_object_info() API (instead of reading and inspecting the data).
* The end result of documentation update has been made to be
inspected more easily to help developers.
* The API to iterate over all objects learned to optionally list
objects in the order they appear in packfiles, which helps locality
of access if the caller accesses these objects while as objects are
enumerated.
* Improve built-in facility to catch broken &&-chain in the tests.
* The more library-ish parts of the codebase learned to work on the
in-core index-state instance that is passed in by their callers,
instead of always working on the singleton "the_index" instance.
* A test prerequisite defined by various test scripts with slightly
different semantics has been consolidated into a single copy and
made into a lazily defined one.
(merge 6ec633059a wc/make-funnynames-shared-lazy-prereq later to maint).
* After a partial clone, repeated fetches from promisor remote would
have accumulated many packfiles marked with .promisor bit without
getting them coalesced into fewer packfiles, hurting performance.
"git repack" now learned to repack them.
* Partially revert the support for multiple hash functions to regain
hash comparison performance; we'd think of a way to do this better
in the next cycle.
* "git help --config" (which is used in command line completion)
missed the configuration variables not described in the main
config.txt file but are described in another file that is included
by it, which has been corrected.
* The test linter code has learned that the end of here-doc mark
"EOF" can be quoted in a double-quote pair, not just in a
single-quote pair.
Git v2.17.1:
Fixes:
* This release contains the same fixes made in the v2.13.7 version of
Git, covering CVE-2018-11233 and 11235, and forward-ported to
v2.14.4, v2.15.2 and v2.16.4 releases. See release notes to
v2.13.7 for details.
* In addition to the above fixes, this release has support on the
server side to reject pushes to repositories that attempt to create
such problematic .gitmodules file etc. as tracked contents, to help
hosting sites protect their customers by preventing malicious
contents from spreading.
Git 2.17:
UI, Workflows & Features
* "diff" family of commands learned "--find-object=<object-id>" option
to limit the findings to changes that involve the named object.
* "git format-patch" learned to give 72-cols to diffstat, which is
consistent with other line length limits the subcommand uses for
its output meant for e-mails.
* The log from "git daemon" can be redirected with a new option; one
relevant use case is to send the log to standard error (instead of
syslog) when running it from inetd.
* "git rebase" learned to take "--allow-empty-message" option.
* "git am" has learned the "--quit" option, in addition to the
existing "--abort" option; having the pair mirrors a few other
commands like "rebase" and "cherry-pick".
* "git worktree add" learned to run the post-checkout hook, just like
"git clone" runs it upon the initial checkout.
* "git tag" learned an explicit "--edit" option that allows the
message given via "-m" and "-F" to be further edited.
* "git fetch --prune-tags" may be used as a handy short-hand for
getting rid of stale tags that are locally held.
* The new "--show-current-patch" option gives an end-user facing way
to get the diff being applied when "git rebase" (and "git am")
stops with a conflict.
* "git add -p" used to offer "/" (look for a matching hunk) as a
choice, even there was only one hunk, which has been corrected.
Also the single-key help is now given only for keys that are
enabled (e.g. help for '/' won't be shown when there is only one
hunk).
* Since Git 1.7.9, "git merge" defaulted to --no-ff (i.e. even when
the side branch being merged is a descendant of the current commit,
create a merge commit instead of fast-forwarding) when merging a
tag object. This was appropriate default for integrators who pull
signed tags from their downstream contributors, but caused an
unnecessary merges when used by downstream contributors who
habitually "catch up" their topic branches with tagged releases
from the upstream. Update "git merge" to default to --no-ff only
when merging a tag object that does *not* sit at its usual place in
refs/tags/ hierarchy, and allow fast-forwarding otherwise, to
mitigate the problem.
* "git status" can spend a lot of cycles to compute the relation
between the current branch and its upstream, which can now be
disabled with "--no-ahead-behind" option.
* "git diff" and friends learned funcname patterns for Go language
source files.
* "git send-email" learned "--reply-to=<address>" option.
* Funcname pattern used for C# now recognizes "async" keyword.
* In a way similar to how "git tag" learned to honor the pager
setting only in the list mode, "git config" learned to ignore the
pager setting when it is used for setting values (i.e. when the
purpose of the operation is not to "show").
Performance, Internal Implementation, Development Support etc.
* More perf tests for threaded grep
* "perf" test output can be sent to codespeed server.
* The build procedure for perl/ part has been greatly simplified by
weaning ourselves off of MakeMaker.
* Perl 5.8 or greater has been required since Git 1.7.4 released in
2010, but we continued to assume some core modules may not exist and
used a conditional "eval { require <<module>> }"; we no longer do
this. Some platforms (Fedora/RedHat/CentOS, for example) ship Perl
without all core modules by default (e.g. Digest::MD5, File::Temp,
File::Spec, Net::Domain, Net::SMTP). Users on such platforms may
need to install these additional modules.
* As a convenience, we install copies of Perl modules we require which
are not part of the core Perl distribution (e.g. Error and
Mail::Address). Users and packagers whose operating system provides
these modules can set NO_PERL_CPAN_FALLBACKS to avoid installing the
bundled modules.
* In preparation for implementing narrow/partial clone, the machinery
for checking object connectivity used by gc and fsck has been
taught that a missing object is OK when it is referenced by a
packfile specially marked as coming from trusted repository that
promises to make them available on-demand and lazily.
* The machinery to clone & fetch, which in turn involves packing and
unpacking objects, has been told how to omit certain objects using
the filtering mechanism introduced by another topic. It now knows
to mark the resulting pack as a promisor pack to tolerate missing
objects, laying foundation for "narrow" clones.
* The first step to getting rid of mru API and using the
doubly-linked list API directly instead.
* Retire mru API as it does not give enough abstraction over
underlying list API to be worth it.
* Rewrite two more "git submodule" subcommands in C.
* The tracing machinery learned to report tweaking of environment
variables as well.
* Update Coccinelle rules to catch and optimize strbuf_addf(&buf, "%s", str)
* Prevent "clang-format" from breaking line after function return type.
* The sequencer infrastructure is shared across "git cherry-pick",
"git rebase -i", etc., and has always spawned "git commit" when it
needs to create a commit. It has been taught to do so internally,
when able, by reusing the codepath "git commit" itself uses, which
gives performance boost for a few tens of percents in some sample
scenarios.
* Push the submodule version of collision-detecting SHA-1 hash
implementation a bit harder on builders.
* Avoid mmapping small files while using packed refs (especially ones
with zero size, which would cause later munmap() to fail).
* Conversion from uchar[20] to struct object_id continues.
* More tests for wildmatch functions.
* The code to binary search starting from a fan-out table (which is
how the packfile is indexed with object names) has been refactored
into a reusable helper.
* We now avoid using identifiers that clash with C++ keywords. Even
though it is not a goal to compile Git with C++ compilers, changes
like this help use of code analysis tools that targets C++ on our
codebase.
* The executable is now built in 'script' phase in Travis CI integration,
to follow the established practice, rather than during 'before_script'
phase. This allows the CI categorize the failures better ('failed'
is project's fault, 'errored' is build environment's).
* Writing out the index file when the only thing that changed in it
is the untracked cache information is often wasteful, and this has
been optimized out.
* Various pieces of Perl code we have have been cleaned up.
* Internal API clean-up to allow write_locked_index() optionally skip
writing the in-core index when it is not modified.
Git v2.16.2 Release Notes
Fixes since v2.16.1
* An old regression in "git describe --all $annotated_tag^0" has been
fixed.
* "git svn dcommit" did not take into account the fact that a
svn+ssh:// URL with a username@ (typically used for pushing) refers
to the same SVN repository without the username@ and failed when
svn.pushmergeinfo option is set.
* "git merge -Xours/-Xtheirs" learned to use our/their version when
resolving a conflicting updates to a symbolic link.
* "git clone $there $here" is allowed even when here directory exists
as long as it is an empty directory, but the command incorrectly
removed it upon a failure of the operation.
* "git stash -- <pathspec>" incorrectly blew away untracked files in
the directory that matched the pathspec, which has been corrected.
* "git add -p" was taught to ignore local changes to submodules as
they do not interfere with the partial addition of regular changes
anyway.
Git 2.16 Release Notes
Backward compatibility notes and other notable changes.
* Use of an empty string as a pathspec element that is used for
'everything matches' is now an error.
Updates since v2.15
UI, Workflows & Features
* An empty string as a pathspec element that means "everything"
i.e. 'git add ""', is now illegal. We started this by first
deprecating and warning a pathspec that has such an element in
2.11 (Nov 2016).
* A hook script that is set unexecutable is simply ignored. Git
notifies when such a file is ignored, unless the message is
squelched via advice.ignoredHook configuration.
* "git pull" has been taught to accept "--[no-]signoff" option and
pass it down to "git merge".
* The "--push-option=<string>" option to "git push" now defaults to a
list of strings configured via push.pushOption variable.
* "gitweb" checks if a directory is searchable with Perl's "-x"
operator, which can be enhanced by using "filetest 'access'"
pragma, which now we do.
* "git stash save" has been deprecated in favour of "git stash push".
* The set of paths output from "git status --ignored" was tied
closely with its "--untracked=<mode>" option, but now it can be
controlled more flexibly. Most notably, a directory that is
ignored because it is listed to be ignored in the ignore/exclude
mechanism can be handled differently from a directory that ends up
to be ignored only because all files in it are ignored.
* The remote-helper for talking to MediaWiki has been updated to
truncate an overlong pagename so that ".mw" suffix can still be
added.
* The remote-helper for talking to MediaWiki has been updated to
work with mediawiki namespaces.
* The "--format=..." option "git for-each-ref" takes learned to show
the name of the 'remote' repository and the ref at the remote side
that is affected for 'upstream' and 'push' via "%(push:remotename)"
and friends.
* Doc and message updates to teach users "bisect view" is a synonym
for "bisect visualize".
* "git bisect run" that did not specify any command to run used to go
ahead and treated all commits to be tested as 'good'. This has
been corrected by making the command error out.
* The SubmittingPatches document has been converted to produce an
HTML version via AsciiDoc/Asciidoctor.
* We learned to optionally talk to a file system monitor via new
fsmonitor extension to speed up "git status" and other operations
that need to see which paths have been modified. Currently we only
support "watchman". See File System Monitor section of
git-update-index(1) for more detail.
* The "diff" family of commands learned to ignore differences in
carriage return at the end of line.
* Places that know about "sendemail.to", like documentation and shell
completion (in contrib/) have been taught about "sendemail.tocmd",
too.
* "git add --renormalize ." is a new and safer way to record the fact
that you are correcting the end-of-line convention and other
"convert_to_git()" glitches in the in-repository data.
* "git branch" and "git checkout -b" are now forbidden from creating
a branch whose name is "HEAD".
* "git branch --list" learned to show its output through the pager by
default when the output is going to a terminal, which is controlled
by the pager.branch configuration variable. This is similar to a
recent change to "git tag --list".
* "git grep -W", "git diff -W" and their friends learned a heuristic
to extend a pre-context beyond the line that matches the "function
pattern" (aka "diff.*.xfuncname") to include a comment block, if
exists, that immediately precedes it.
* "git config --expiry-date gc.reflogexpire" can read "2.weeks" from
the configuration and report it as a timestamp, just like "--int"
would read "1k" and report 1024, to help consumption by scripts.
* The shell completion (in contrib/) learned that "git pull" can take
the "--autostash" option.
* The tagnames "git log --decorate" uses to annotate the commits can
now be limited to subset of available refs with the two additional
options, --decorate-refs[-exclude]=<pattern>.
* "git grep" compiled with libpcre2 sometimes triggered a segfault,
which is being fixed.
* "git send-email" tries to see if the sendmail program is available
in /usr/lib and /usr/sbin; extend the list of locations to be
checked to also include directories on $PATH.
* "git diff" learned, "--anchored", a variant of the "--patience"
algorithm, to which the user can specify which 'unique' line to be
used as anchoring points.
* The way "git worktree add" determines what branch to create from
where and checkout in the new worktree has been updated a bit.
* Ancient part of codebase still shows dots after an abbreviated
object name just to show that it is not a full object name, but
these ellipses are confusing to people who newly discovered Git
who are used to seeing abbreviated object names and find them
confusing with the range syntax.
* With a configuration variable rebase.abbreviateCommands set,
"git rebase -i" produces the todo list with a single-letter
command names.
* "git worktree add" learned to run the post-checkout hook, just like
"git checkout" does, after the initial checkout.
* "git svn" has been updated to strip CRs in the commit messages, as
recent versions of Subversion rejects them.
* "git imap-send" did not correctly quote the folder name when
making a request to the server, which has been corrected.
* Error messages from "git rebase" have been somewhat cleaned up.
* Git has been taught to support an https:// URL used for http.proxy
when using recent versions of libcurl.
* "git merge" learned to pay attention to merge.verifySignatures
configuration variable and pretend as if '--verify-signatures'
option was given from the command line.
* "git describe" was taught to dig trees deeper to find a
<commit-ish>:<path> that refers to a given blob object.
Fixes since v2.14.1
* Because recent Git for Windows do come with a real msgfmt, the
build procedure for git-gui has been updated to use it instead of a
hand-rolled substitute.
* "%C(color name)" in the pretty print format always produced ANSI
color escape codes, which was an early design mistake. They now
honor the configuration (e.g. "color.ui = never") and also tty-ness
of the output medium.
* The http.{sslkey,sslCert} configuration variables are to be
interpreted as a pathname that honors "~[username]/" prefix, but
weren't, which has been fixed.
* Numerous bugs in walking of reflogs via "log -g" and friends have
been fixed.
* "git commit" when seeing an totally empty message said "you did not
edit the message", which is clearly wrong. The message has been
corrected.
* When a directory is not readable, "gitweb" fails to build the
project list. Work this around by skipping such a directory.
* A recently added test for the "credential-cache" helper revealed
that EOF detection done around the time the connection to the cache
daemon is torn down were flaky. This was fixed by reacting to
ECONNRESET and behaving as if we got an EOF.
* Some versions of GnuPG fail to kill gpg-agent it auto-spawned
and such a left-over agent can interfere with a test. Work it
around by attempting to kill one before starting a new test.
* "git log --tag=no-such-tag" showed log starting from HEAD, which
has been fixed---it now shows nothing.
* The "tag.pager" configuration variable was useless for those who
actually create tag objects, as it interfered with the use of an
editor. A new mechanism has been introduced for commands to enable
pager depending on what operation is being carried out to fix this,
and then "git tag -l" is made to run pager by default.
* "git push --recurse-submodules $there HEAD:$target" was not
propagated down to the submodules, but now it is.
* Commands like "git rebase" accepted the --rerere-autoupdate option
from the command line, but did not always use it. This has been
fixed.
* "git clone --recurse-submodules --quiet" did not pass the quiet
option down to submodules.
* "git am -s" has been taught that some input may end with a trailer
block that is not Signed-off-by: and it should refrain from adding
an extra blank line before adding a new sign-off in such a case.
* "git svn" used with "--localtime" option did not compute the tz
offset for the timestamp in question and instead always used the
current time, which has been corrected.
* Memory leaks in a few error codepaths have been plugged.
* bash 4.4 or newer gave a warning on NUL byte in command
substitution done in "git stash"; this has been squelched.
* "git grep -L" and "git grep --quiet -L" reported different exit
codes; this has been corrected.
* When handshake with a subprocess filter notices that the process
asked for an unknown capability, Git did not report what program
the offending subprocess was running. This has been corrected.
* "git apply" that is used as a better "patch -p1" failed to apply a
taken from a file with CRLF line endings to a file with CRLF line
endings. The root cause was because it misused convert_to_git()
that tried to do "safe-crlf" processing by looking at the index
entry at the same path, which is a nonsense---in that mode, "apply"
is not working on the data in (or derived from) the index at all.
This has been fixed.
* Killing "git merge --edit" before the editor returns control left
the repository in a state with MERGE_MSG but without MERGE_HEAD,
which incorrectly tells the subsequent "git commit" that there was
a squash merge in progress. This has been fixed.
* "git archive" did not work well with pathspecs and the
export-ignore attribute.
* "git cvsserver" no longer is invoked by "git daemon" by default,
as it is old and largely unmaintained.
* Various Perl scripts did not use safe_pipe_capture() instead of
backticks, leaving them susceptible to end-user input. They have
been corrected.
======================
Backward compatibility notes and other notable changes.
* Use of an empty string as a pathspec element that is used for
'everything matches' is still warned and Git asks users to use a
more explicit '.' for that instead. The hope is that existing
users will not mind this change, and eventually the warning can be
turned into a hard error, upgrading the deprecation into removal of
this (mis)feature. That is not scheduled to happen in the upcoming
release (yet).
* Git now avoids blindly falling back to ".git" when the setup
sequence said we are _not_ in Git repository. A corner case that
happens to work right now may be broken by a call to die("BUG").
We've tried hard to locate such cases and fixed them, but there
might still be cases that need to be addressed--bug reports are
greatly appreciated.
* The experiment to improve the hunk-boundary selection of textual
diff output has finished, and the "indent heuristics" has now
become the default.
* Git can now be built with PCRE v2 instead of v1 of the PCRE
library. Replace USE_LIBPCRE=YesPlease with USE_LIBPCRE2=YesPlease
in existing build scripts to build against the new version. As the
upstream PCRE maintainer has abandoned v1 maintenance for all but
the most critical bug fixes, use of v2 is recommended.
Updates since v2.13
-------------------
UI, Workflows & Features
* The colors in which "git status --short --branch" showed the names
of the current branch and its remote-tracking branch are now
configurable.
* "git clone" learned the "--no-tags" option not to fetch all tags
initially, and also set up the tagopt not to follow any tags in
subsequent fetches.
* "git archive --format=zip" learned to use zip64 extension when
necessary to go beyond the 4GB limit.
* "git reset" learned "--recurse-submodules" option.
* "git diff --submodule=diff" now recurses into nested submodules.
* "git repack" learned to accept the --threads=<n> option and pass it
to pack-objects.
* "git send-email" learned to run sendemail-validate hook to inspect
and reject a message before sending it out.
* There is no good reason why "git fetch $there $sha1" should fail
when the $sha1 names an object at the tip of an advertised ref,
even when the other side hasn't enabled allowTipSHA1InWant.
* The "[includeIf "gitdir:$dir"] path=..." mechanism introduced in
2.13.0 would canonicalize the path of the gitdir being matched,
and did not match e.g. "gitdir:~/work/*" against a repo in
"~/work/main" if "~/work" was a symlink to "/mnt/storage/work".
Now we match both the resolved canonical path and what "pwd" would
show. The include will happen if either one matches.
* The "indent" heuristics is now the default in "diff". The
diff.indentHeuristic configuration variable can be set to "false"
for those who do not want it.
* Many commands learned to pay attention to submodule.recurse
configuration.
* The convention for a command line is to follow "git cmdname
--options" with revisions followed by an optional "--"
disambiguator and then finally pathspecs. When "--" is not there,
we make sure early ones are all interpretable as revs (and do not
look like paths) and later ones are the other way around. A
pathspec with "magic" (e.g. ":/p/a/t/h" that matches p/a/t/h from
the top-level of the working tree, no matter what subdirectory you
are working from) are conservatively judged as "not a path", which
required disambiguation more often. The command line parser
learned to say "it's a pathspec" a bit more often when the syntax
looks like so.
* Update "perl-compatible regular expression" support to enable JIT
and also allow linking with the newer PCRE v2 library.
* "filter-branch" learned a pseudo filter "--setup" that can be used
to define common functions/variables that can be used by other
filters.
* Using "git add d/i/r" when d/i/r is the top of the working tree of
a separate repository would create a gitlink in the index, which
would appear as a not-quite-initialized submodule to others. We
learned to give warnings when this happens.
* "git status" learned to optionally give how many stash entries there
are in its output.
* "git status" has long shown essentially the same message as "git
commit"; the message it gives while preparing for the root commit,
i.e. "Initial commit", was hard to understand for some new users.
Now it says "No commits yet" to stress more on the current status
(rather than the commit the user is preparing for, which is more in
line with the focus of "git commit").
* "git send-email" now has --batch-size and --relogin-delay options
which can be used to overcome limitations on SMTP servers that
restrict on how many of e-mails can be sent in a single session.
* An old message shown in the commit log template was removed, as it
has outlived its usefulness.
* "git pull --rebase --recurse-submodules" learns to rebase the
branch in the submodules to an updated base.
* "git log" learned -P as a synonym for --perl-regexp, "git grep"
already had such a synonym.
* "git log" didn't understand --regexp-ignore-case when combined with
--perl-regexp. This has been fixed.
Performance, Internal Implementation, Development Support etc.
* The default packed-git limit value has been raised on larger
platforms to save "git fetch" from a (recoverable) failure while
"gc" is running in parallel.
* Code to update the cache-tree has been tightened so that we won't
accidentally write out any 0{40} entry in the tree object.
* Attempt to allow us notice "fishy" situation where we fail to
remove the temporary directory used during the test.
* Travis CI gained a task to format the documentation with both
AsciiDoc and AsciiDoctor.
* Some platforms have ulong that is smaller than time_t, and our
historical use of ulong for timestamp would mean they cannot
represent some timestamp that the platform allows. Invent a
separate and dedicated timestamp_t (so that we can distingiuish
timestamps and a vanilla ulongs, which along is already a good
move), and then declare uintmax_t is the type to be used as the
timestamp_t.
* We can trigger Windows auto-build tester (credits: Dscho &
Microsoft) from our existing Travis CI tester now.
* Conversion from uchar[20] to struct object_id continues.
* Simplify parse_pathspec() codepath and stop it from looking at the
default in-core index.
* Add perf-test for wildmatch.
* Code from "conversion using external process" codepath has been
extracted to a separate sub-process.[ch] module.
* When "git checkout", "git merge", etc. manipulates the in-core
index, various pieces of information in the index extensions are
discarded from the original state, as it is usually not the case
that they are kept up-to-date and in-sync with the operation on the
main index. The untracked cache extension is copied across these
operations now, which would speed up "git status" (as long as the
cache is properly invalidated).
* The internal implementation of "git grep" has seen some clean-up.
* Update the C style recommendation for notes for translators, as
recent versions of gettext tools can work with our style of
multi-line comments.
* The implementation of "ref" API around the "packed refs" have been
cleaned up, in preparation for further changes.
* The internal logic used in "git blame" has been libified to make it
easier to use by cgit.
* Our code often opens a path to an optional file, to work on its
contents when we can successfully open it. We can ignore a failure
to open if such an optional file does not exist, but we do want to
report a failure in opening for other reasons (e.g. we got an I/O
error, or the file is there, but we lack the permission to open).
The exact errors we need to ignore are ENOENT (obviously) and
ENOTDIR (less obvious). Instead of repeating comparison of errno
with these two constants, introduce a helper function to do so.
* We often try to open a file for reading whose existence is
optional, and silently ignore errors from open/fopen; report such
errors if they are not due to missing files.
* When an existing repository is used for t/perf testing, we first
create bit-for-bit copy of it, which may grab a transient state of
the repository and freeze it into the repository used for testing,
which then may cause Git operations to fail. Single out "the index
being locked" case and forcibly drop the lock from the copy.
* Three instances of the same helper function have been consolidated
to one.
* "fast-import" uses a default pack chain depth that is consistent
with other parts of the system.
* A new test to show the interaction between the pattern [^a-z]
(which matches '/') and a slash in a path has been added. The
pattern should not match the slash with "pathmatch", but should
with "wildmatch".
* The 'diff-highlight' program (in contrib/) has been restructured
for easier reuse by an external project 'diff-so-fancy'.
* A common pattern to free a piece of memory and assign NULL to the
pointer that used to point at it has been replaced with a new
FREE_AND_NULL() macro.
* Traditionally, the default die() routine had a code to prevent it
from getting called multiple times, which interacted badly when a
threaded program used it (one downside is that the real error may
be hidden and instead the only error message given to the user may
end up being "die recursion detected", which is not very useful).
* Introduce a "repository" object to eventually make it easier to
work in multiple repositories (the primary focus is to work with
the superproject and its submodules) in a single process.
* Optimize "what are the object names already taken in an alternate
object database?" query that is used to derive the length of prefix
an object name is uniquely abbreviated to.
* The hashmap API has been updated so that data to customize the
behaviour of the comparison function can be specified at the time a
hashmap is initialized.
* The "collision detecting" SHA-1 implementation shipped with 2.13 is
now integrated into git.git as a submodule (the first submodule to
ship with git.git). Clone git.git with --recurse-submodules to get
it. For now a non-submodule copy of the same code is also shipped
as part of the tree.
* A recent update made it easier to use "-fsanitize=" option while
compiling but supported only one sanitize option. Allow more than
one to be combined, joined with a comma, like "make SANITIZE=foo,bar".
* Use "p4 -G" to make "p4 changes" output more Python-friendly
to parse.
* We started using "%" PRItime, imitating "%" PRIuMAX and friends, as
a way to format the internal timestamp value, but this does not
play well with gettext(1) i18n framework, and causes "make pot"
that is run by the l10n coordinator to create a broken po/git.pot
file. This is a possible workaround for that problem.
* It turns out that Cygwin also needs the fopen() wrapper that
returns failure when a directory is opened for reading.
=========================
Fixes since v2.13
-----------------
* The Web interface to gmane news archive is long gone, even though
the articles are still accessible via NTTP. Replace the links with
ones to public-inbox.org. Because their message identification is
based on the actual message-id, it is likely that it will be easier
to migrate away from it if/when necessary.
* Update tests to pass under GETTEXT_POISON (a mechanism to ensure
that output strings that should not be translated are not
translated by mistake), and tell TravisCI to run them.
* Setting "log.decorate=false" in the configuration file did not take
effect in v2.13, which has been corrected.
* An earlier update to test 7400 needed to be skipped on CYGWIN.
* Git sometimes gives an advice in a rhetorical question that does
not require an answer, which can confuse new users and non native
speakers. Attempt to rephrase them.
* "git read-tree -m" (no tree-ish) gave a nonsense suggestion "use
--empty if you want to clear the index". With "-m", such a request
will still fail anyway, as you'd need to name at least one tree-ish
to be merged.
* The codepath in "git am" that is used when running "git rebase"
leaked memory held for the log message of the commits being rebased.
* "pack-objects" can stream a slice of an existing packfile out when
the pack bitmap can tell that the reachable objects are all needed
in the output, without inspecting individual objects. This
strategy however would not work well when "--local" and other
options are in use, and need to be disabled.
* Clarify documentation for include.path and includeIf.<condition>.path
configuration variables.
* Tag objects, which are not reachable from any ref, that point at
missing objects were mishandled by "git gc" and friends (they
should silently be ignored instead)
* A few http:// links that are redirected to https:// in the
documentation have been updated to https:// links.
* Make sure our tests would pass when the sources are checked out
with "platform native" line ending convention by default on
Windows. Some "text" files out tests use and the test scripts
themselves that are meant to be run with /bin/sh, ought to be
checked out with eol=LF even on Windows.
* Fix memory leaks pointed out by Coverity (and people).
* The receive-pack program now makes sure that the push certificate
records the same set of push options used for pushing.
* "git cherry-pick" and other uses of the sequencer machinery
mishandled a trailer block whose last line is an incomplete line.
This has been fixed so that an additional sign-off etc. are added
after completing the existing incomplete line.
* The shell completion script (in contrib/) learned "git stash" has
a new "push" subcommand.
* Travis CI gained a task to format the documentation with both
AsciiDoc and AsciiDoctor.
* Update the C style recommendation for notes for translators, as
recent versions of gettext tools can work with our style of
multi-line comments.
* "git clone --config var=val" is a way to populate the
per-repository configuration file of the new repository, but it did
not work well when val is an empty string. This has been fixed.
* A few codepaths in "checkout" and "am" working on an unborn branch
tried to access an uninitialized piece of memory.
* "git for-each-ref --format=..." with %(HEAD) in the format used to
resolve the HEAD symref as many times as it had processed refs,
which was wasteful, and "git branch" shared the same problem.
* "git interpret-trailers", when used as GIT_EDITOR for "git commit
-v", looked for and appended to a trailer block at the very end,
i.e. at the end of the "diff" output. The command has been
corrected to pay attention to the cut-mark line "commit -v" adds to
the buffer---the real trailer block should appear just before it.
* A test allowed both "git push" and "git receive-pack" on the other
end write their traces into the same file. This is OK on platforms
that allows atomically appending to a file opened with O_APPEND,
but on other platforms led to a mangled output, causing
intermittent test failures. This has been fixed by disabling
traces from "receive-pack" in the test.
* "foo\bar\baz" in "git fetch foo\bar\baz", even though there is no
slashes in it, cannot be a nickname for a remote on Windows, as
that is likely to be a pathname on a local filesystem.
* The "collision detecting" SHA-1 implementation shipped with 2.13
was quite broken on some big-endian platforms and/or platforms that
do not like unaligned fetches. Update to the upstream code which
has already fixed these issues.
* "git am -h" triggered a BUG().
* The interaction of "url.*.insteadOf" and custom URL scheme's
whitelisting is now documented better.
======================
Backward compatibility notes.
* Use of an empty string as a pathspec element that is used for
'everything matches' is still warned and Git asks users to use a
more explicit '.' for that instead. The hope is that existing
users will not mind this change, and eventually the warning can be
turned into a hard error, upgrading the deprecation into removal of
this (mis)feature. That is not scheduled to happen in the upcoming
release (yet).
* The historical argument order "git merge <msg> HEAD <commit>..."
has been deprecated for quite some time, and is now removed.
* The default location "~/.git-credential-cache/socket" for the
socket used to communicate with the credential-cache daemon has
been moved to "~/.cache/git/credential/socket".
* Git now avoids blindly falling back to ".git" when the setup
sequence said we are _not_ in Git repository. A corner case that
happens to work right now may be broken by a call to die("BUG").
We've tried hard to locate such cases and fixed them, but there
might still be cases that need to be addressed--bug reports are
greatly appreciated.
Updates since v2.12
-------------------
UI, Workflows & Features
* "git describe" and "git name-rev" have been taught to take more
than one refname patterns to restrict the set of refs to base their
naming output on, and also learned to take negative patterns to
name refs not to be used for naming via their "--exclude" option.
* Deletion of a branch "foo/bar" could remove .git/refs/heads/foo
once there no longer is any other branch whose name begins with
"foo/", but we didn't do so so far. Now we do.
* When "git merge" detects a path that is renamed in one history
while the other history deleted (or modified) it, it now reports
both paths to help the user understand what is going on in the two
histories being merged.
* The <url> part in "http.<url>.<variable>" configuration variable
can now be spelled with '*' that serves as wildcard.
E.g. "http.https://*.example.com.proxy" can be used to specify the
proxy used for https://a.example.com, https://b.example.com, etc.,
i.e. any host in the example.com domain.
* "git tag" did not leave useful message when adding a new entry to
reflog; this was left unnoticed for a long time because refs/tags/*
doesn't keep reflog by default.
* The "negative" pathspec feature was somewhat more cumbersome to use
than necessary in that its short-hand used "!" which needed to be
escaped from shells, and it required "exclude from what?" specified.
* The command line options for ssh invocation needs to be tweaked for
some implementations of SSH (e.g. PuTTY plink wants "-P <port>"
while OpenSSH wants "-p <port>" to specify port to connect to), and
the variant was guessed when GIT_SSH environment variable is used
to specify it. The logic to guess now applies to the command
specified by the newer GIT_SSH_COMMAND and also core.sshcommand
configuration variable, and comes with an escape hatch for users to
deal with misdetected cases.
More...
=========================
Fixes since v2.11
-----------------
* The default Travis-CI configuration specifies newer P4 and GitLFS.
* The character width table has been updated to match Unicode 9.0
* Update the isatty() emulation for Windows by updating the previous
hack that depended on internals of (older) MSVC runtime.
* "git rev-parse --symbolic" failed with a more recent notation like
"HEAD^-1" and "HEAD^!".
* An empty directory in a working tree that can simply be nuked used
to interfere while merging or cherry-picking a change to create a
submodule directory there, which has been fixed..
* The code in "git push" to compute if any commit being pushed in the
superproject binds a commit in a submodule that hasn't been pushed
out was overly inefficient, making it unusable even for a small
project that does not have any submodule but have a reasonable
number of refs.
* "git push --dry-run --recurse-submodule=on-demand" wasn't
"--dry-run" in the submodules.
* The output from "git worktree list" was made in readdir() order,
and was unstable.
* mergetool.<tool>.trustExitCode configuration variable did not apply
to built-in tools, but now it does.
* "git p4" LFS support was broken when LFS stores an empty blob.
* Fix a corner case in merge-recursive regression that crept in
during 2.10 development cycle.
* Update the error messages from the dumb-http client when it fails
to obtain loose objects; we used to give sensible error message
only upon 404 but we now forbid unexpected redirects that needs to
be reported with something sensible.
* When diff.renames configuration is on (and with Git 2.9 and later,
it is enabled by default, which made it worse), "git stash"
misbehaved if a file is removed and another file with a very
similar content is added.
* "git diff --no-index" did not take "--no-abbrev" option.
* "git difftool --dir-diff" had a minor regression when started from
a subdirectory, which has been fixed.
* "git commit --allow-empty --only" (no pathspec) with dirty index
ought to be an acceptable way to create a new commit that does not
change any paths, but it was forbidden, perhaps because nobody
needed it so far.
* A pathname that begins with "//" or "\\" on Windows is special but
path normalization logic was unaware of it.
* "git pull --rebase", when there is no new commits on our side since
we forked from the upstream, should be able to fast-forward without
invoking "git rebase", but it didn't.
* The way to specify hotkeys to "xxdiff" that is used by "git
mergetool" has been modernized to match recent versions of xxdiff.
* Unlike "git am --abort", "git cherry-pick --abort" moved HEAD back
to where cherry-pick started while picking multiple changes, when
the cherry-pick stopped to ask for help from the user, and the user
did "git reset --hard" to a different commit in order to re-attempt
the operation.
* Code cleanup in shallow boundary computation.
* A recent update to receive-pack to make it easier to drop garbage
objects made it clear that GIT_ALTERNATE_OBJECT_DIRECTORIES cannot
have a pathname with a colon in it (no surprise!), and this in turn
made it impossible to push into a repository at such a path. This
has been fixed by introducing a quoting mechanism used when
appending such a path to the colon-separated list.
* The function usage_msg_opt() has been updated to say "fatal:"
before the custom message programs give, when they want to die
with a message about wrong command line options followed by the
standard usage string.
* "git index-pack --stdin" needs an access to an existing repository,
but "git index-pack file.pack" to generate an .idx file that
corresponds to a packfile does not.
* Fix for NDEBUG builds.
* A lazy "git push" without refspec did not internally use a fully
specified refspec to perform 'current', 'simple', or 'upstream'
push, causing unnecessary "ambiguous ref" errors.
* "git p4" misbehaved when swapping a directory and a symbolic link.
* Even though an fix was attempted in Git 2.9.3 days, but running
"git difftool --dir-diff" from a subdirectory never worked. This
has been fixed.
* "git p4" that tracks multile p4 paths imported a single changelist
that touches files in these multiple paths as one commit, followed
by many empty commits. This has been fixed.
* A potential but unlikely buffer overflow in Windows port has been
fixed.
* When the http server gives an incomplete response to a smart-http
rpc call, it could lead to client waiting for a full response that
will never come. Teach the client side to notice this condition
and abort the transfer.
* Some platforms no longer understand "latin-1" that is still seen in
the wild in e-mail headers; replace them with "iso-8859-1" that is
more widely known when conversion fails from/to it.
* Update the procedure to generate "tags" for developer support.
* Update the definition of the MacOSX test environment used by
TravisCI.
* A few git-svn updates.
* Compression setting for producing packfiles were spread across
three codepaths, one of which did not honor any configuration.
Unify these so that all of them honor core.compression and
pack.compression variables the same way.
* "git fast-import" sometimes mishandled while rebalancing notes
tree, which has been fixed.
* Recent update to the default abbreviation length that auto-scales
lacked documentation update, which has been corrected.
* Leakage of lockfiles in the config subsystem has been fixed.
* It is natural that "git gc --auto" may not attempt to pack
everything into a single pack, and there is no point in warning
when the user has configured the system to use the pack bitmap,
leading to disabling further "gc".
* "git archive" did not read the standard configuration files, and
failed to notice a file that is marked as binary via the userdiff
driver configuration.
* "git blame --porcelain" misidentified the "previous" <commit, path>
pair (aka "source") when contents came from two or more files.
* "git rebase -i" with a recent update started showing an incorrect
count when squashing more than 10 commits.
* "git <cmd> @{push}" on a detached HEAD used to segfault; it has
been corrected to error out with a message.
* Tighten a test to avoid mistaking an extended ERE regexp engine as
a PRE regexp engine.
* Typing ^C to pager, which usually does not kill it, killed Git and
took the pager down as a collateral damage in certain process-tree
structure. This has been fixed.
=========================
Fixes since v2.10.1
-------------------
* The code that parses the format parameter of for-each-ref command
has seen a micro-optimization.
* The "graph" API used in "git log --graph" miscounted the number of
output columns consumed so far when drawing a padding line, which
has been fixed; this did not affect any existing code as nobody
tried to write anything after the padding on such a line, though.
* Almost everybody uses DEFAULT_ABBREV to refer to the default
setting for the abbreviation, but "git blame" peeked into
underlying variable bypassing the macro for no good reason.
* Doc update to clarify what "log -3 --reverse" does.
* An author name, that spelled a backslash-quoted double quote in the
human readable part "My \"double quoted\" name", was not unquoted
correctly while applying a patch from a piece of e-mail.
* The original command line syntax for "git merge", which was "git
merge <msg> HEAD <parent>...", has been deprecated for quite some
time, and "git gui" was the last in-tree user of the syntax. This
is finally fixed, so that we can move forward with the deprecation.
* Codepaths that read from an on-disk loose object were too loose in
validating what they are reading is a proper object file and
sometimes read past the data they read from the disk, which has
been corrected. H/t to Gustavo Grieco for reporting.
* "git worktree", even though it used the default_abbrev setting that
ought to be affected by core.abbrev configuration variable, ignored
the variable setting. The command has been taught to read the
default set of configuration variables to correct this.
* A low-level function verify_packfile() was meant to show errors
that were detected without dying itself, but under some conditions
it didn't and died instead, which has been fixed.
* When "git fetch" tries to find where the history of the repository
it runs in has diverged from what the other side has, it has a
mechanism to avoid digging too deep into irrelevant side branches.
This however did not work well over the "smart-http" transport due
to a design bug, which has been fixed.
* When we started cURL to talk to imap server when a new enough
version of cURL library is available, we forgot to explicitly add
imap(s):// before the destination. To some folks, that didn't work
and the library tried to make HTTP(s) requests instead.
* The ./configure script generated from configure.ac was taught how
to detect support of SSL by libcurl better.
* http.emptyauth configuration is a way to allow an empty username to
pass when attempting to authenticate using mechanisms like
Kerberos. We took an unspecified (NULL) username and sent ":"
(i.e. no username, no password) to CURLOPT_USERPWD, but did not do
the same when the username is explicitly set to an empty string.
* "git clone" of a local repository can be done at the filesystem
level, but the codepath did not check errors while copying and
adjusting the file that lists alternate object stores.
* Documentation for "git commit" was updated to clarify that "commit
-p <paths>" adds to the current contents of the index to come up
with what to commit.
* A stray symbolic link in $GIT_DIR/refs/ directory could make name
resolution loop forever, which has been corrected.
* The "submodule.<name>.path" stored in .gitmodules is never copied
to .git/config and such a key in .git/config has no meaning, but
the documentation described it and submodule.<name>.url next to
each other as if both belong to .git/config. This has been fixed.
* Recent git allows submodule.<name>.branch to use a special token
"." instead of the branch name; the documentation has been updated
to describe it.
* In a worktree connected to a repository elsewhere, created via "git
worktree", "git checkout" attempts to protect users from confusion
by refusing to check out a branch that is already checked out in
another worktree. However, this also prevented checking out a
branch, which is designated as the primary branch of a bare
reopsitory, in a worktree that is connected to the bare
repository. The check has been corrected to allow it.
* "git rebase" immediately after "git clone" failed to find the fork
point from the upstream.
* When fetching from a remote that has many tags that are irrelevant
to branches we are following, we used to waste way too many cycles
when checking if the object pointed at by a tag (that we are not
going to fetch!) exists in our repository too carefully.
* The Travis CI configuration we ship ran the tests with --verbose
option but this risks non-TAP output that happens to be "ok" to be
misinterpreted as TAP signalling a test that passed. This resulted
in unnecessary failure. This has been corrected by introducing a
new mode to run our tests in the test harness to send the verbose
output separately to the log file.
* Some AsciiDoc formatter mishandles a displayed illustration with
tabs in it. Adjust a few of them in merge-base documentation to
work around them.
Also contains minor documentation updates and code clean-ups.
Fixes since v2.9.1
------------------
* A fix merged to v2.9.1 had a few tests that are not meant to be
run on platforms without 64-bit long, which caused unnecessary
test failures on them because we didn't detect the platform and
skip them. These tests are now skipped on platforms that they
are not applicable to.
Fixes since v2.9
----------------
* When "git daemon" is run without --[init-]timeout specified, a
connection from a client that silently goes offline can hang around
for a long time, wasting resources. The socket-level KEEPALIVE has
been enabled to allow the OS to notice such failed connections.
* The commands in `git log` family take %C(auto) in a custom format
string. This unconditionally turned the color on, ignoring
--no-color or with --color=auto when the output is not connected to
a tty; this was corrected to make the format truly behave as
"auto".
* "git rev-list --count" whose walk-length is limited with "-n"
option did not work well with the counting optimized to look at the
bitmap index.
* "git show -W" (extend hunks to cover the entire function, delimited
by lines that match the "funcname" pattern) used to show the entire
file when a change added an entire function at the end of the file,
which has been fixed.
* The documentation set has been updated so that literal commands,
configuration variables and environment variables are consistently
typeset in fixed-width font and bold in manpages.
* "git svn propset" subcommand that was added in 2.3 days is
documented now.
* The documentation tries to consistently spell "GPG"; when
referring to the specific program name, "gpg" is used.
* "git reflog" stopped upon seeing an entry that denotes a branch
creation event (aka "unborn"), which made it appear as if the
reflog was truncated.
* The git-prompt scriptlet (in contrib/) was not friendly with those
who uses "set -u", which has been fixed.
* A codepath that used alloca(3) to place an unbounded amount of data
on the stack has been updated to avoid doing so.
* "git update-index --add --chmod=+x file" may be usable as an escape
hatch, but not a friendly thing to force for people who do need to
use it regularly. "git add --chmod=+x file" can be used instead.
* Build improvements for gnome-keyring (in contrib/)
* "git status" used to say "working directory" when it meant "working
tree".
* Comments about misbehaving FreeBSD shells have been clarified with
the version number (9.x and before are broken, newer ones are OK).
* "git cherry-pick A" worked on an unborn branch, but "git
cherry-pick A..B" didn't.
* "git add -i/-p" learned to honor diff.compactionHeuristic
experimental knob, so that the user can work on the same hunk split
as "git diff" output.
* "log --graph --format=" learned that "%>|(N)" specifies the width
relative to the terminal's left edge, not relative to the area to
draw text that is to the right of the ancestry-graph section. It
also now accepts negative N that means the column limit is relative
to the right border.
* The ownership rule for the piece of memory that hold references to
be fetched in "git fetch" was screwy, which has been cleaned up.
* "git bisect" makes an internal call to "git diff-tree" when
bisection finds the culprit, but this call did not initialize the
data structure to pass to the diff-tree API correctly.
* Formats of the various data (and how to validate them) where we use
GPG signature have been documented.
* Fix an unintended regression in v2.9 that breaks "clone --depth"
that recurses down to submodules by forcing the submodules to also
be cloned shallowly, which many server instances that host upstream
of the submodules are not prepared for.
* Fix unnecessarily waste in the idiomatic use of ': ${VAR=default}'
to set the default value, without enclosing it in double quotes.
* Some platform-specific code had non-ANSI strict declarations of C
functions that do not take any parameters, which has been
corrected.
* The internal code used to show local timezone offset is not
prepared to handle timestamps beyond year 2100, and gave a
bogus offset value to the caller. Use a more benign looking
+0000 instead and let "git log" going in such a case, instead
of aborting.
* One among four invocations of readlink(1) in our test suite has
been rewritten so that the test can run on systems without the
command (others are in valgrind test framework and t9802).
* t/perf needs /usr/bin/time with GNU extension; the invocation of it
is updated to "gtime" on Darwin.
* A bug, which caused "git p4" while running under verbose mode to
report paths that are omitted due to branch prefix incorrectly, has
been fixed; the command said "Ignoring file outside of prefix" for
paths that are _inside_.
* The top level documentation "git help git" still pointed at the
documentation set hosted at now-defunct google-code repository.
Update it to point to https://git.github.io/htmldocs/git.html
instead.
Also contains minor documentation updates and code clean-ups.
adam
jperkin
ryoon
leot
wiz
gdt
nros
mef
soda
martin
khorben