blob: 519c89407cf3f2e155042f52bebfe0a68f949714 [file] [log] [blame]
Git Release Notes
Backward Compatibility Notes
* "git branch -l <foo>" used to be a way to ask a reflog to be
created while creating a new branch, but that is no longer the
case. It is a short-hand for "git branch --list <foo>" now.
* "git push" into refs/tags/* hierarchy is rejected without getting
forced, but "git fetch" (misguidedly) used the "fast forwarding"
rule used for the refs/heads/* hierarchy; this has been corrected,
which means some fetches of tags that did not fail with older
version of Git will fail without "--force" with this version.
* "git help -a" now gives verbose output (same as "git help -av").
Those who want the old output may say "git help --no-verbose -a"..
* "git cpn --help", when "cpn" is an alias to, say, "cherry-pick -n",
reported only the alias expansion of "cpn" in earlier versions of
Git. It now runs "git cherry-pick --help" to show the manual page
of the command, while sending the alias expansion to the standard
error stream.
* "git send-email" learned to grab address-looking string on any
trailer whose name ends with "-by". This is a backward-incompatible
change. Adding "--suppress-cc=misc-by" on the command line, or
setting sendemail.suppresscc configuration variable to "misc-by",
can be used to disable this behaviour.
Updates since v2.19
UI, Workflows & Features
* Running "git clone" against a project that contain two files with
pathnames that differ only in cases on a case insensitive
filesystem would result in one of the files lost because the
underlying filesystem is incapable of holding both at the same
time. An attempt is made to detect such a case and warn.
* "git checkout -b newbranch [HEAD]" should not have to do as much as
checking out a commit different from HEAD. An attempt is made to
optimize this special case.
* "git rev-list --stdin </dev/null" used to be an error; it now shows
no output without an error. "git rev-list --stdin --default HEAD"
still falls back to the given default when nothing is given on the
standard input.
* Lift code from GitHub to restrict delta computation so that an
object that exists in one fork is not made into a delta against
another object that does not appear in the same forked repository.
* "git format-patch" learned new "--interdiff" and "--range-diff"
options to explain the difference between this version and the
previous attempt in the cover letter (or after the three-dashes as
a comment).
* "git mailinfo" used in "git am" learned to make a best-effort
recovery of a patch corrupted by MUA that sends text/plain with
format=flawed option.
(merge 3aa4d81f88 rs/mailinfo-format-flowed later to maint).
* The rules used by "git push" and "git fetch" to determine if a ref
can or cannot be updated were inconsistent; specifically, fetching
to update existing tags were allowed even though tags are supposed
to be unmoving anchoring points. "git fetch" was taught to forbid
updates to existing tags without the "--force" option.
* "git multi-pack-index" learned to detect corruption in the .midx
file it uses, and this feature has been integrated into "git fsck".
* Generation of (experimental) commit-graph files have so far been
fairly silent, even though it takes noticeable amount of time in a
meaningfully large repository. The users will now see progress
* The minimum version of Windows supported by Windows port of Git is
now set to Vista.
* The completion script (in contrib/) learned to complete a handful of
options "git stash list" command takes.
* The completion script (in contrib/) learned that "git fetch
--multiple" only takes remote names as arguments and no refspecs.
* "git status" learns to show progress bar when refreshing the index
takes a long time.
(merge ae9af12287 nd/status-refresh-progress later to maint).
* "git help -a" and "git help -av" give different pieces of
information, and generally the "verbose" version is more friendly
to the new users. "git help -a" by default now uses the more
verbose output (with "--no-verbose", you can go back to the
original). Also "git help -av" now lists aliases and external
commands, which it did not used to.
* Unlike "grep", "git grep" by default recurses to the whole tree.
The command learned "git grep --recursive" option, so that "git
grep --no-recursive" can serve as a synonym to setting the
max-depth to 0.
* When pushing into a repository that borrows its objects from an
alternate object store, "git receive-pack" that responds to the
push request on the other side lists the tips of refs in the
alternate to reduce the amount of objects transferred. This
sometimes is detrimental when the number of refs in the alternate
is absurdly large, in which case the bandwidth saved in potentially
fewer objects transferred is wasted in excessively large ref
advertisement. The alternate refs that are advertised are now
configurable with a pair of configuration variables.
* "git cmd --help" when "cmd" is aliased used to only say "cmd is
aliased to ...". Now it shows that to the standard error stream
and runs "git $cmd --help" where $cmd is the first word of the
alias expansion.
* The documentation of "git gc" has been updated to mention that it
is no longer limited to "pruning away crufts" but also updates
ancillary files like commit-graph as a part of repository
* "git p4 unshelve" improvements.
* The logic to select the default user name and e-mail on Windows has
been improved.
(merge 501afcb8b0 js/mingw-default-ident later to maint).
* The "rev-list --filter" feature learned to exclude all trees via
"tree:0" filter.
* "git send-email" learned to grab address-looking string on any
trailer whose name ends with "-by"; --suppress-cc=misc-by on the
command line, or setting sendemail.suppresscc configuration
variable to "misc-by", can be used to disable this behaviour.
* Developer builds now uses -Wunused-function compilation option.
* One of our CI tests to run with "unusual/experimental/random"
settings now also uses commit-graph and midx.
* "git mergetool" learned to take the "--[no-]gui" option, just like
"git difftool" does.
* "git rebase -i" learned a new insn, 'break', that the user can
insert in the to-do list. Upon hitting it, the command returns
control back to the user.
* New "--pretty=format:" placeholders %GF and %GP that show the GPG
key fingerprints have been invented.
* On platforms with recent cURL library, http.sslBackend configuration
variable can be used to choose a different SSL backend at runtime.
The Windows port uses this mechanism to switch between OpenSSL and
Secure Channel while talking over the HTTPS protocol.
* "git send-email" learned to disable SMTP authentication via the
"--smtp-auth=none" option, even when the smtp username is given
(which turns the authentication on by default).
* A fourth class of configuration files (in addition to the
traditional "system wide", "per user in the $HOME directory" and
"per repository in the $GIT_DIR/config") has been introduced so
that different worktrees that share the same repository (hence the
same $GIT_DIR/config file) can use different customization.
* A pattern with '**' that does not have a slash on either side used
to be an invalid one, but the code now treats such double-asterisks
the same way as two normal asterisks that happen to be adjacent to
each other.
(merge e5bbe09e88 nd/wildmatch-double-asterisk later to maint).
Performance, Internal Implementation, Development Support etc.
* When there are too many packfiles in a repository (which is not
recommended), looking up an object in these would require
consulting many pack .idx files; a new mechanism to have a single
file that consolidates all of these .idx files is introduced.
* "git submodule update" is getting rewritten piece-by-piece into C.
* The code for computing history reachability has been shuffled,
obtained a bunch of new tests to cover them, and then being
* The unpack_trees() API used in checking out a branch and merging
walks one or more trees along with the index. When the cache-tree
in the index tells us that we are walking a tree whose flattened
contents is known (i.e. matches a span in the index), as linearly
scanning a span in the index is much more efficient than having to
open tree objects recursively and listing their entries, the walk
can be optimized, which has been done.
* When creating a thin pack, which allows objects to be made into a
delta against another object that is not in the resulting pack but
is known to be present on the receiving end, the code learned to
take advantage of the reachability bitmap; this allows the server
to send a delta against a base beyond the "boundary" commit.
* spatch transformation to replace boolean uses of !hashcmp() to
newly introduced oideq() is added, and applied, to regain
performance lost due to support of multiple hash algorithms.
* Fix a bug in which the same path could be registered under multiple
worktree entries if the path was missing (for instance, was removed
manually). Also, as a convenience, expand the number of cases in
which --force is applicable.
* Split Documentation/config.txt for easier maintenance.
(merge 6014363f0b nd/config-split later to maint).
* Test helper binaries clean-up.
(merge c9a1f4161f nd/test-tool later to maint).
* Various tests have been updated to make it easier to swap the
hash function used for object identification.
(merge ae0c89d41b bc/hash-independent-tests later to maint).
* Update fsck.skipList implementation and documentation.
(merge 371a655074 ab/fsck-skiplist later to maint).
* An alias that expands to another alias has so far been forbidden,
but now it is allowed to create such an alias.
* Various test scripts have been updated for style and also correct
handling of exit status of various commands.
* "gc --auto" ended up calling exit(-1) upon error, which has been
corrected to use exit(1). Also the error reporting behaviour when
daemonized has been updated to exit with zero status when stopping
due to a previously discovered error (which implies there is no
point running gc to improve the situation); we used to exit with
failure in such a case.
* Various codepaths in the core-ish part learned to work on an
arbitrary in-core index structure, not necessarily the default
instance "the_index".
(merge b3c7eef9b0 nd/the-index later to maint).
* Code clean-up in the internal machinery used by "git status" and
"git commit --dry-run".
(merge 73ba5d78b4 ss/wt-status-committable later to maint).
* Some environment variables that control the runtime options of Git
used during tests are getting renamed for consistency.
(merge 4231d1ba99 bp/rename-test-env-var later to maint).
* A new extension to the index file has been introduced, which allows
the index file to be read in parallel for performance.
* The oidset API was built on top of the oidmap API which in turn is
on the hashmap API. Replace the implementation to build on top of
the khash API and gain performance.
* Over some transports, fetching objects with an exact commit object
name can be done without first seeing the ref advertisements. The
code has been optimized to exploit this.
* In a partial clone that will lazily be hydrated from the
originating repository, we generally want to avoid "does this
object exist (locally)?" on objects that we deliberately omitted
when we created the clone. The cache-tree codepath (which is used
to write a tree object out of the index) however insisted that the
object exists, even for paths that are outside of the partial
checkout area. The code has been updated to avoid such a check.
* To help developers, an EditorConfig file that attempts to follow
the project convention has been added.
(merge b548d698a0 bc/editorconfig later to maint).
* The result of coverage test can be combined with "git blame" to
check the test coverage of code introduced recently with a new
'coverage-diff' tool (in contrib/).
(merge 783faedd65 ds/coverage-diff later to maint).
* An experiment to fuzz test a few areas, hopefully we can gain more
coverage to various areas.
* More codepaths are moving away from hardcoded hash sizes.
* The way the Windows port figures out the current directory has been
* The way DLLs are loaded on the Windows port has been improved.
* Some tests have been reorganized and renamed; "ls t/" now gives a
better overview of what is tested for these scripts than before.
* "git rebase" and "git rebase -i" have been reimplemented in C.
* Windows port learned to use nano-second resolution file timestamps.
* The overly large Documentation/config.txt file have been split into
million little pieces. This potentially allows each individual piece
included into the manual page of the command it affects more easily.
* Replace three string-list instances used as look-up tables in "git
fetch" with hashmaps.
* Unify code to read the author-script used in "git am" and the
commands that use the sequencer machinery, e.g. "git rebase -i".
* In preparation to the day when we can deprecate and remove the
"rebase -p", make sure we can skip and later remove tests for
* The history traversal used to implement the tag-following has been
optimized by introducing a new helper.
* The helper function to refresh the cached stat information in the
in-core index has learned to perform the lstat() part of the
operation in parallel on multi-core platforms.
* The code to traverse objects for reachability, used to decide what
objects are unreferenced and expendable, have been taught to also
consider per-worktree refs of other worktrees as starting points to
prevent data loss.
* "git add" needs to internally run "diff-files" equivalent, and the
codepath learned the same optimization as "diff-files" has to run
lstat(2) in parallel to find which paths have been updated in the
working tree.
* The procedure to install dependencies before testing at Travis CI
is getting revamped for both simplicity and flexibility, taking
advantage of the recent move to the vm-based environment.
* The support for format-patch (and send-email) by the command-line
completion script (in contrib/) has been simplified a bit.
Fixes since v2.19
* "git interpret-trailers" and its underlying machinery had a buggy
code that attempted to ignore patch text after commit log message,
which triggered in various codepaths that will always get the log
message alone and never get such an input.
(merge 66e83d9b41 jk/trailer-fixes later to maint).
* Malformed or crafted data in packstream can make our code attempt
to read or write past the allocated buffer and abort, instead of
reporting an error, which has been fixed.
* "git rebase -i" did not clear the state files correctly when a run
of "squash/fixup" is aborted and then the user manually amended the
commit instead, which has been corrected.
(merge 10d2f35436 js/rebase-i-autosquash-fix later to maint).
* When fsmonitor is in use, after operation on submodules updates
.gitmodules, we lost track of the fact that we did so and relied on
stale fsmonitor data.
(merge 43f1180814 bp/mv-submodules-with-fsmonitor later to maint).
* Fix for a long-standing bug that leaves the index file corrupt when
it shrinks during a partial commit.
(merge 6c003d6ffb jk/reopen-tempfile-truncate later to maint).
* Further fix for O_APPEND emulation on Windows
(merge eeaf7ddac7 js/mingw-o-append later to maint).
* A corner case bugfix in "git rerere" code.
(merge ad2bf0d9b4 en/rerere-multi-stage-1-fix later to maint).
* "git add ':(attr:foo)'" is not supported and is supposed to be
rejected while the command line arguments are parsed, but we fail
to reject such a command line upfront.
(merge 84d938b732 nd/attr-pathspec-fix later to maint).
* Recent update broke the reachability algorithm when refs (e.g.
tags) that point at objects that are not commit were involved,
which has been fixed.
* "git rebase" etc. in Git 2.19 fails to abort when given an empty
commit log message as result of editing, which has been corrected.
(merge a3ec9eaf38 en/sequencer-empty-edit-result-aborts later to maint).
* The code to backfill objects in lazily cloned repository did not
work correctly, which has been corrected.
(merge e68302011c jt/lazy-object-fetch-fix later to maint).
* Update error messages given by "git remote" and make them consistent.
(merge 5025425dff ms/remote-error-message-update later to maint).
* "git update-ref" learned to make both "--no-deref" and "--stdin"
work at the same time.
(merge d345e9fbe7 en/update-ref-no-deref-stdin later to maint).
* Recently added "range-diff" had a corner-case bug to cause it
segfault, which has been corrected.
(merge e467a90c7a tg/range-diff-corner-case-fix later to maint).
* The recently introduced commit-graph auxiliary data is incompatible
with mechanisms such as replace & grafts that "breaks" immutable
nature of the object reference relationship. Disable optimizations
based on its use (and updating existing commit-graph) when these
incompatible features are in use in the repository.
(merge 829a321569 ds/commit-graph-with-grafts later to maint).
* The mailmap file update.
(merge 255eb03edf jn/mailmap-update later to maint).
* The code in "git status" sometimes hit an assertion failure. This
was caused by a structure that was reused without cleaning the data
used for the first run, which has been corrected.
(merge 3e73cc62c0 en/status-multiple-renames-to-the-same-target-fix later to maint).
* "git fetch $repo $object" in a partial clone did not correctly
fetch the asked-for object that is referenced by an object in
promisor packfile, which has been fixed.
* A corner-case bugfix.
(merge c5cbb27cb5 sm/show-superproject-while-conflicted later to maint).
* Various fixes to "diff --color-moved-ws".
* A partial clone that is configured to lazily fetch missing objects
will on-demand issue a "git fetch" request to the originating
repository to fill not-yet-obtained objects. The request has been
optimized for requesting a tree object (and not the leaf blob
objects contained in it) by telling the originating repository that
no blobs are needed.
(merge 4c7f9567ea jt/non-blob-lazy-fetch later to maint).
* The codepath to support the experimental split-index mode had
remaining "racily clean" issues fixed.
(merge 4c490f3d32 sg/split-index-racefix later to maint).
* "git log --graph" showing an octopus merge sometimes miscounted the
number of display columns it is consuming to show the merge and its
parent commits, which has been corrected.
(merge 04005834ed np/log-graph-octopus-fix later to maint).
* "git range-diff" did not work well when the compared ranges had
changes in submodules and the "--submodule=log" was used.
* The implementation of run_command() API on the UNIX platforms had a
bug that caused a command not on $PATH to be found in the current
(merge f67b980771 jk/run-command-notdot later to maint).
* A mutex used in "git pack-objects" were not correctly initialized
and this caused "git repack" to dump core on Windows.
(merge 34204c8166 js/pack-objects-mutex-init-fix later to maint).
* Under certain circumstances, "git diff D:/a/b/c D:/a/b/d" on
Windows would strip initial parts from the paths because they
were not recognized as absolute, which has been corrected.
(merge ffd04e92e2 js/diff-notice-has-drive-prefix later to maint).
* The receive.denyCurrentBranch=updateInstead codepath kicked in even
when the push should have been rejected due to other reasons, such
as it does not fast-forward or the update-hook rejects it, which
has been corrected.
(merge b072a25fad jc/receive-deny-current-branch-fix later to maint).
* The logic to determine the archive type "git archive" uses did not
correctly kick in for "git archive --remote", which has been
* "git repack" in a shallow clone did not correctly update the
shallow points in the repository, leading to a repository that
does not pass fsck.
(merge 5dcfbf564c js/shallow-and-fetch-prune later to maint).
* Some codepaths failed to form a proper URL when .gitmodules record
the URL to a submodule repository as relative to the repository of
superproject, which has been corrected.
(merge e0a862fdaf sb/submodule-url-to-absolute later to maint).
* "git fetch" over protocol v2 into a shallow repository failed to
fetch full history behind a new tip of history that was diverged
before the cut-off point of the history that was previously fetched
* The command line completion machinery (in contrib/) has been
updated to allow the completion script to tweak the list of options
that are reported by the parse-options machinery correctly.
(merge 276b49ff34 nd/completion-negation later to maint).
* Operations on promisor objects make sense in the context of only a
small subset of the commands that internally use the revisions
machinery, but the "--exclude-promisor-objects" option were taken
and led to nonsense results by commands like "log", to which it
didn't make much sense. This has been corrected.
(merge 669b1d2aae md/exclude-promisor-objects-fix later to maint).
* The "container" mode of TravisCI is going away. Our .travis.yml
file is getting prepared for the transition.
(merge 32ee384be8 ss/travis-ci-force-vm-mode later to maint).
* Our test scripts can now take the '-V' option as a synonym for the
'--verbose-log' option.
(merge a5f52c6dab sg/test-verbose-log later to maint).
* A regression in Git 2.12 era made "git fsck" fall into an infinite
loop while processing truncated loose objects.
(merge 18ad13e5b2 jk/detect-truncated-zlib-input later to maint).
* "git ls-remote $there foo" was broken by recent update for the
protocol v2 and stopped showing refs that match 'foo' that are not
refs/{heads,tags}/foo, which has been fixed.
(merge 6a139cdd74 jk/proto-v2-ref-prefix-fix later to maint).
* Additional comment on a tricky piece of code to help developers.
(merge 0afbe3e806 jk/stream-pack-non-delta-clarification later to maint).
* A couple of tests used to leave the repository in a state that is
deliberately corrupt, which have been corrected.
(merge aa984dbe5e ab/pack-tests-cleanup later to maint).
* The submodule support has been updated to read from the blob at
HEAD:.gitmodules when the .gitmodules file is missing from the
working tree.
(merge 2b1257e463 ao/submodule-wo-gitmodules-checked-out later to maint).
* "git fetch" was a bit loose in parsing responses from the other side
when talking over the protocol v2.
* "git rev-parse --exclude=* --branches --branches" (i.e. first
saying "add only things that do not match '*' out of all branches"
and then adding all branches, without any exclusion this time")
worked as expected, but "--exclude=* --all --all" did not work the
same way, which has been fixed.
(merge 5221048092 ag/rev-parse-all-exclude-fix later to maint).
* "git send-email --transfer-encoding=..." in recent versions of Git
sometimes produced an empty "Content-Transfer-Encoding:" header,
which has been corrected.
(merge 3c88e46f1a al/send-email-auto-cte-fixup later to maint).
* The interface into "xdiff" library used to discover the offset and
size of a generated patch hunk by first formatting it into the
textual hunk header "@@ -n,m +k,l @@" and then parsing the numbers
out. A new interface has been introduced to allow callers a more
direct access to them.
(merge 5eade0746e jk/xdiff-interface later to maint).
* Pathspec matching against a tree object were buggy when negative
pathspec elements were involved, which has been fixed.
(merge b7845cebc0 nd/tree-walk-path-exclusion later to maint).
* Code cleanup, docfix, build fix, etc.
(merge 96a7501aad ts/doc-build-manpage-xsl-quietly later to maint).
(merge b9b07efdb2 tg/conflict-marker-size later to maint).
(merge fa0aeea770 sg/doc-trace-appends later to maint).
(merge d64324cb60 tb/void-check-attr later to maint).
(merge c3b9bc94b9 en/double-semicolon-fix later to maint).
(merge 79336116f5 sg/t3701-tighten-trace later to maint).
(merge 801fa63a90 jk/dev-build-format-security later to maint).
(merge 0597dd62ba sb/string-list-remove-unused later to maint).
(merge db2d36fad8 bw/protocol-v2 later to maint).
(merge 456d7cd3a9 sg/split-index-test later to maint).
(merge 7b6057c852 tq/refs-internal-comment-fix later to maint).
(merge 29e8dc50ad tg/t5551-with-curl-7.61.1 later to maint).
(merge 55f6bce2c9 fe/doc-updates later to maint).
(merge 7987d2232d jk/check-everything-connected-is-long-gone later to maint).
(merge 4ba3c9be47 dz/credential-doc-url-matching-rules later to maint).
(merge 4c399442f7 ma/commit-graph-docs later to maint).
(merge fc0503b04e ma/t1400-undebug-test later to maint).
(merge e56b53553a nd/packobjectshook-doc-fix later to maint).
(merge c56170a0c4 ma/mailing-list-address-in-git-help later to maint).
(merge 6e8fc70fce rs/sequencer-oidset-insert-avoids-dups later to maint).
(merge ad0b8f9575 mw/doc-typofixes later to maint).
(merge d9f079ad1a jc/how-to-document-api later to maint).
(merge b1492bf315 ma/t7005-bash-workaround later to maint).
(merge ac1f98a0df du/rev-parse-is-plumbing later to maint).
(merge ca8ed443a5 mm/doc-no-dashed-git later to maint).
(merge ce366a8144 du/get-tar-commit-id-is-plumbing later to maint).
(merge 61018fe9e0 du/cherry-is-plumbing later to maint).
(merge c7e5fe79b9 sb/strbuf-h-update later to maint).
(merge 8d2008196b tq/branch-create-wo-branch-get later to maint).
(merge 2e3c894f4b tq/branch-style-fix later to maint).
(merge c5d844af9c sg/doc-show-branch-typofix later to maint).
(merge 081d91618b ah/doc-updates later to maint).
(merge b84c783882 jc/cocci-preincr later to maint).
(merge 5e495f8122 uk/merge-subtree-doc-update later to maint).
(merge aaaa881822 jk/uploadpack-packobjectshook-fix later to maint).
(merge 3063477445 tb/char-may-be-unsigned later to maint).
(merge 8c64bc9420 sg/test-rebase-editor-fix later to maint).
(merge 71571cd7d6 ma/sequencer-do-reset-saner-loop-termination later to maint).