4 Backward compatibility notes.
6 * An empty string used as a pathspec element has always meant
7 'everything matches', but it is too easy to write a script that
8 finds a path to remove in $path and run 'git rm "$paht"' by
9 mistake (when the user meant to give "$path"), which ends up
10 removing everything. This release starts warning about the
11 use of an empty string that is used for 'everything matches' and
12 asks users to use a more explicit '.' for that instead.
14 The hope is that existing users will not mind this change, and
15 eventually the warning can be turned into a hard error, upgrading
16 the deprecation into removal of this (mis)feature.
18 * The historical argument order "git merge <msg> HEAD <commit>..."
19 has been deprecated for quite some time, and will be removed in the
20 next release (not this one).
22 * The default abbreviation length, which has historically been 7, now
23 scales as the repository grows, using the approximate number of
24 objects in the repository and a bit of math around the birthday
25 paradox. The logic suggests to use 12 hexdigits for the Linux
26 kernel, and 9 to 10 for Git itself.
32 UI, Workflows & Features
34 * Comes with new version of git-gui, now at its 0.21.0 tag.
36 * "git format-patch --cover-letter HEAD^" to format a single patch
37 with a separate cover letter now numbers the output as [PATCH 0/1]
38 and [PATCH 1/1] by default.
40 * An incoming "git push" that attempts to push too many bytes can now
41 be rejected by setting a new configuration variable at the receiving
44 * "git nosuchcommand --help" said "No manual entry for gitnosuchcommand",
45 which was not intuitive, given that "git nosuchcommand" said "git:
46 'nosuchcommand' is not a git command".
48 * "git clone --recurse-submodules --reference $path $URL" is a way to
49 reduce network transfer cost by borrowing objects in an existing
50 $path repository when cloning the superproject from $URL; it
51 learned to also peek into $path for presence of corresponding
52 repositories of submodules and borrow objects from there when able.
54 * The "git diff --submodule={short,log}" mechanism has been enhanced
55 to allow "--submodule=diff" to show the patch between the submodule
56 commits bound to the superproject.
58 * Even though "git hash-objects", which is a tool to take an
59 on-filesystem data stream and put it into the Git object store,
60 can perform "outside-world-to-Git" conversions (e.g.
61 end-of-line conversions and application of the clean-filter), and
62 it has had this feature on by default from very early days, its reverse
63 operation "git cat-file", which takes an object from the Git object
64 store and externalizes it for consumption by the outside world,
65 lacked an equivalent mechanism to run the "Git-to-outside-world"
66 conversion. The command learned the "--filters" option to do so.
68 * Output from "git diff" can be made easier to read by intelligently selecting
69 which lines are common and which lines are added/deleted
70 when the lines before and after the changed section
71 are the same. A command line option (--indent-heuristic) and a
72 configuration variable (diff.indentHeuristic) are added to help with the
73 experiment to find good heuristics.
75 * In some projects, it is common to use "[RFC PATCH]" as the subject
76 prefix for a patch meant for discussion rather than application. A
77 new format-patch option "--rfc" is a short-hand for "--subject-prefix=RFC PATCH"
78 to help the participants of such projects.
80 * "git add --chmod={+,-}x <pathspec>" only changed the
81 executable bit for paths that are either new or modified. This has
82 been corrected to change the executable bit for all paths that match
85 * When "git format-patch --stdout" output is placed as an in-body
86 header and it uses RFC2822 header folding, "git am" fails to
87 put the header line back into a single logical line. The
88 underlying "git mailinfo" was taught to handle this properly.
90 * "gitweb" can spawn "highlight" to show blob contents with
91 (programming) language-specific syntax highlighting, but only
92 when the language is known. "highlight" can however be told
93 to guess the language itself by giving it "--force" option, which
96 * "git gui" l10n to Portuguese.
98 * When given an abbreviated object name that is not (or more
99 realistically, "no longer") unique, we gave a fatal error
100 "ambiguous argument". This error is now accompanied by a hint that
101 lists the objects beginning with the given prefix. During the
102 course of development of this new feature, numerous minor bugs were
103 uncovered and corrected, the most notable one of which is that we
104 gave "short SHA1 xxxx is ambiguous." twice without good reason.
106 * "git log rev^..rev" is an often-used revision range specification
107 to show what was done on a side branch merged at rev. This has
108 gained a short-hand "rev^-1". In general "rev^-$n" is the same as
109 "^rev^$n rev", i.e. what has happened on other branches while the
110 history leading to nth parent was looking the other way.
112 * In recent versions of cURL, GSSAPI credential delegation is
113 disabled by default due to CVE-2011-2192; introduce a http.delegation
114 configuration variable to selectively allow enabling this.
115 (merge 26a7b23429 ps/http-gssapi-cred-delegation later to maint).
117 * "git mergetool" learned to honor "-O<orderfile>" to control the
118 order of paths to present to the end user.
120 * "git diff/log --ws-error-highlight=<kind>" lacked the corresponding
121 configuration variable (diff.wsErrorHighlight) to set it by default.
123 * "git ls-files" learned the "--recurse-submodules" option
124 to get a listing of tracked files across submodules (i.e. this
125 only works with the "--cached" option, not for listing untracked or
126 ignored files). This would be a useful tool to sit on the upstream
127 side of a pipe that is read with xargs to work on all working tree
128 files from the top-level superproject.
130 * A new credential helper that talks via "libsecret" with
131 implementations of XDG Secret Service API has been added to
134 * The GPG verification status shown by the "%G?" pretty format specifier
135 was not rich enough to differentiate a signature made by an expired
136 key, a signature made by a revoked key, etc. New output letters
137 have been assigned to express them.
139 * In addition to purely abbreviated commit object names, "gitweb"
140 learned to turn "git describe" output (e.g. v2.9.3-599-g2376d31787)
141 into clickable links in its output.
143 * "git commit" created an empty commit when invoked with an index
144 consisting solely of intend-to-add paths (added with "git add -N").
145 It now requires the "--allow-empty" option to create such a commit.
146 The same logic prevented "git status" from showing such paths as "new files" in the
147 "Changes not staged for commit" section.
149 * The smudge/clean filter API spawns an external process
150 to filter the contents of each path that has a filter defined. A
151 new type of "process" filter API has been added to allow the first
152 request to run the filter for a path to spawn a single process, and
153 all filtering is served by this single process for multiple
154 paths, reducing the process creation overhead.
156 * The user always has to say "stash@{$N}" when naming a single
157 element in the default location of the stash, i.e. reflogs in
158 refs/stash. The "git stash" command learned to accept "git stash
159 apply 4" as a short-hand for "git stash apply stash@{4}".
162 Performance, Internal Implementation, Development Support etc.
164 * The delta-base-cache mechanism has been a key to the performance in
165 a repository with a tightly packed packfile, but it did not scale
166 well even with a larger value of core.deltaBaseCacheLimit.
168 * Enhance "git status --porcelain" output by collecting more data on
169 the state of the index and the working tree files, which may
170 further be used to teach git-prompt (in contrib/) to make fewer
173 * Extract a small helper out of the function that reads the authors
174 script file "git am" internally uses.
175 (merge a77598e jc/am-read-author-file later to maint).
177 * Lift calls to exit(2) and die() higher in the callchain in
178 sequencer.c files so that more helper functions in it can be used
179 by callers that want to handle error conditions themselves.
181 * "git am" has been taught to make an internal call to "git apply"'s
182 innards without spawning the latter as a separate process.
184 * The ref-store abstraction was introduced to the refs API so that we
185 can plug in different backends to store references.
187 * The "unsigned char sha1[20]" to "struct object_id" conversion
188 continues. Notable changes in this round includes that ce->sha1,
189 i.e. the object name recorded in the cache_entry, turns into an
192 * JGit can show a fake ref "capabilities^{}" to "git fetch" when it
193 does not advertise any refs, but "git fetch" was not prepared to
194 see such an advertisement. When the other side disconnects without
195 giving any ref advertisement, we used to say "there may not be a
196 repository at that URL", but we may have seen other advertisements
197 like "shallow" and ".have" in which case we definitely know that a
198 repository is there. The code to detect this case has also been
201 * Some codepaths in "git pack-objects" were not ready to use an
202 existing pack bitmap; now they are and as a result they have
205 * The codepath in "git fsck" to detect malformed tree objects has
206 been updated not to die but keep going after detecting them.
208 * We call "qsort(array, nelem, sizeof(array[0]), fn)", and most of
209 the time third parameter is redundant. A new QSORT() macro lets us
212 * "git pack-objects" in a repository with many packfiles used to
213 spend a lot of time looking for/at objects in them; the accesses to
214 the packfiles are now optimized by checking the most-recently-used
216 (merge c9af708b1a jk/pack-objects-optim-mru later to maint).
218 * Codepaths involved in interacting alternate object stores have
221 * In order for the receiving end of "git push" to inspect the
222 received history and decide to reject the push, the objects sent
223 from the sending end need to be made available to the hook and
224 the mechanism for the connectivity check, and this was done
225 traditionally by storing the objects in the receiving repository
226 and letting "git gc" expire them. Instead, store the newly
227 received objects in a temporary area, and make them available by
228 reusing the alternate object store mechanism to them only while we
229 decide if we accept the check, and once we decide, either migrate
230 them to the repository or purge them immediately.
232 * The require_clean_work_tree() helper was recreated in C when "git
233 pull" was rewritten from shell; the helper is now made available to
234 other callers in preparation for upcoming "rebase -i" work.
236 * "git upload-pack" had its code cleaned-up and performance improved
237 by reducing use of timestamp-ordered commit-list, which was
238 replaced with a priority queue.
240 * "git diff --no-index" codepath has been updated not to try to peek
241 into a .git/ directory that happens to be under the current
242 directory, when we know we are operating outside any repository.
244 * Update of the sequencer codebase to make it reusable to reimplement
245 "rebase -i" continues.
247 * Git generally does not explicitly close file descriptors that were
248 open in the parent process when spawning a child process, but most
249 of the time the child does not want to access them. As Windows does
250 not allow removing or renaming a file that has a file descriptor
251 open, a slow-to-exit child can even break the parent process by
252 holding onto them. Use O_CLOEXEC flag to open files in various
255 * Update "interpret-trailers" machinery and teach it that people in
256 the real world write all sorts of cruft in the "trailer" that was
257 originally designed to have the neat-o "Mail-Header: like thing"
261 Also contains various documentation updates and code clean-ups.
267 Unless otherwise noted, all the fixes since v2.9 in the maintenance
268 track are contained in this release (see the maintenance releases'
271 * Clarify various ways to specify the "revision ranges" in the
274 * "diff-highlight" script (in contrib/) learned to work better with
275 "git log -p --graph" output.
277 * The test framework left the number of tests and success/failure
278 count in the t/test-results directory, keyed by the name of the
279 test script plus the process ID. The latter however turned out not
280 to serve any useful purpose. The process ID part of the filename
283 * Having a submodule whose ".git" repository is somehow corrupt
284 caused a few commands that recurse into submodules to loop forever.
286 * "git symbolic-ref -d HEAD" happily removes the symbolic ref, but
287 the resulting repository becomes an invalid one. Teach the command
288 to forbid removal of HEAD.
290 * A test spawned a short-lived background process, which sometimes
291 prevented the test directory from getting removed at the end of the
292 script on some platforms.
294 * Update a few tests that used to use GIT_CURL_VERBOSE to use the
295 newer GIT_TRACE_CURL.
297 * "git pack-objects --include-tag" was taught that when we know that
298 we are sending an object C, we want a tag B that directly points at
299 C but also a tag A that points at the tag B. We used to miss the
300 intermediate tag B in some cases.
302 * Update Japanese translation for "git-gui".
304 * "git fetch http::/site/path" did not die correctly and segfaulted
307 * "git commit-tree" stopped reading commit.gpgsign configuration
308 variable that was meant for Porcelain "git commit" in Git 2.9; we
309 forgot to update "git gui" to look at the configuration to match
312 * "git add --chmod={+,-}x" added recently lacked documentation, which has
315 * "git log --cherry-pick" used to include merge commits as candidates
316 to be matched up with other commits, resulting a lot of wasted time.
317 The patch-id generation logic has been updated to ignore merges and
320 * The http transport (with curl-multi option, which is the default
321 these days) failed to remove curl-easy handle from a curlm session,
322 which led to unnecessary API failures.
324 * There were numerous corner cases in which the configuration files
325 are read and used or not read at all depending on the directory a
326 Git command was run, leading to inconsistent behaviour. The code
327 to set-up repository access at the beginning of a Git process has
328 been updated to fix them.
329 (merge 4d0efa1 jk/setup-sequence-update later to maint).
331 * "git diff -W" output needs to extend the context backward to
332 include the header line of the current function and also forward to
333 include the body of the entire current function up to the header
334 line of the next one. This process may have to merge two adjacent
335 hunks, but the code forgot to do so in some cases.
337 * Performance tests done via "t/perf" did not use the right
338 build configuration if the user relied on autoconf generated
341 * "git format-patch --base=..." feature that was recently added
342 showed the base commit information after the "-- " e-mail signature
343 line, which turned out to be inconvenient. The base information
344 has been moved above the signature line.
348 * Even when "git pull --rebase=preserve" (and the underlying "git
349 rebase --preserve") can complete without creating any new commits
350 (i.e. fast-forwards), it still insisted on having usable ident
351 information (read: user.email is set correctly), which was less
352 than nice. As the underlying commands used inside "git rebase"
353 would fail with a more meaningful error message and advice text
354 when the bogus ident matters, this extra check was removed.
356 * "git gc --aggressive" used to limit the delta-chain length to 250,
357 which is way too deep for gaining additional space savings and is
358 detrimental for runtime performance. The limit has been reduced to
361 * Documentation for individual configuration variables to control use
362 of color (like `color.grep`) said that their default value is
363 'false', instead of saying their default is taken from `color.ui`.
364 When we updated the default value for color.ui from 'false' to
365 'auto' quite a while ago, all of them broke. This has been
368 * The pretty-format specifier "%C(auto)" used by the "log" family of
369 commands to enable coloring of the output is taught to also issue a
370 color-reset sequence to the output.
372 * A shell script example in check-ref-format documentation has been
375 * "git checkout <word>" does not follow the usual disambiguation
376 rules when the <word> can be both a rev and a path, to allow
377 checking out a branch 'foo' in a project that happens to have a
378 file 'foo' in the working tree without having to disambiguate.
379 This was poorly documented and the check was incorrect when the
380 command was run from a subdirectory.
382 * Some codepaths in "git diff" used regexec(3) on a buffer that was
383 mmap(2)ed, which may not have a terminating NUL, leading to a read
384 beyond the end of the mapped region. This was fixed by introducing
385 a regexec_buf() helper that takes a <ptr,len> pair with REG_STARTEND
388 * The procedure to build Git on Mac OS X for Travis CI hardcoded the
389 internal directory structure we assumed HomeBrew uses, which was a
390 no-no. The procedure has been updated to ask HomeBrew things we
391 need to know to fix this.
393 * When "git rebase -i" is given a broken instruction, it told the
394 user to fix it with "--edit-todo", but didn't say what the step
395 after that was (i.e. "--continue").
397 * Documentation around tools to import from CVS was fairly outdated.
399 * "git clone --recurse-submodules" lost the progress eye-candy in
400 a recent update, which has been corrected.
402 * A low-level function verify_packfile() was meant to show errors
403 that were detected without dying itself, but under some conditions
404 it didn't and died instead, which has been fixed.
406 * When "git fetch" tries to find where the history of the repository
407 it runs in has diverged from what the other side has, it has a
408 mechanism to avoid digging too deep into irrelevant side branches.
409 This however did not work well over the "smart-http" transport due
410 to a design bug, which has been fixed.
412 * In the codepath that comes up with the hostname to be used in an
413 e-mail when the user didn't tell us, we looked at the ai_canonname
414 field in struct addrinfo without making sure it is not NULL first.
416 * "git worktree", even though it used the default_abbrev setting that
417 ought to be affected by the core.abbrev configuration variable, ignored
418 the variable setting. The command has been taught to read the
419 default set of configuration variables to correct this.
421 * "git init" tried to record core.worktree in the repository's
422 'config' file when the GIT_WORK_TREE environment variable was set and
423 it was different from where GIT_DIR appears as ".git" at its top,
424 but the logic was faulty when .git is a "gitdir:" file that points
425 at the real place, causing trouble in working trees that are
426 managed by "git worktree". This has been corrected.
428 * Codepaths that read from an on-disk loose object were too loose in
429 validating that they are reading a proper object file and
430 sometimes read past the data they read from the disk, which has
431 been corrected. H/t to Gustavo Grieco for reporting.
433 * The original command line syntax for "git merge", which was "git
434 merge <msg> HEAD <parent>...", has been deprecated for quite some
435 time, and "git gui" was the last in-tree user of the syntax. This
436 is finally fixed, so that we can move forward with the deprecation.
438 * An author name that has a backslash-quoted double quote in the
439 human readable part ("My \"double quoted\" name"), was not unquoted
440 correctly while applying a patch from a piece of e-mail.
442 * Doc update to clarify what "log -3 --reverse" does.
444 * Almost everybody uses DEFAULT_ABBREV to refer to the default
445 setting for the abbreviation, but "git blame" peeked into
446 underlying variable bypassing the macro for no good reason.
448 * The "graph" API used in "git log --graph" miscounted the number of
449 output columns consumed so far when drawing a padding line, which
450 has been fixed; this did not affect any existing code as nobody
451 tried to write anything after the padding on such a line, though.
453 * The code that parses the format parameter of the for-each-ref command
454 has seen a micro-optimization.
456 * When we started to use cURL to talk to an imap server, we forgot to explicitly add
457 imap(s):// before the destination. To some folks, that didn't work
458 and the library tried to make HTTP(s) requests instead.
460 * The ./configure script generated from configure.ac was taught how
461 to detect support of SSL by libcurl better.
463 * The command-line completion script (in contrib/) learned to
464 complete "git cmd ^mas<HT>" to complete the negative end of
465 reference to "git cmd ^master".
466 (merge 49416ad22a cp/completion-negative-refs later to maint).
468 * The existing "git fetch --depth=<n>" option was hard to use
469 correctly when making the history of an existing shallow clone
470 deeper. A new option, "--deepen=<n>", has been added to make this
471 easier to use. "git clone" also learned "--shallow-since=<date>"
472 and "--shallow-exclude=<tag>" options to make it easier to specify
473 "I am interested only in the recent N months worth of history" and
474 "Give me only the history since that version".
475 (merge cccf74e2da nd/shallow-deepen later to maint).
477 * "git blame --reverse OLD path" is now DWIMmed to show how lines
478 in path in an old revision OLD have survived up to the current
480 (merge e1d09701a4 jc/blame-reverse later to maint).
482 * The http.emptyauth configuration variable is a way to allow an empty username to
483 pass when attempting to authenticate using mechanisms like
484 Kerberos. We took an unspecified (NULL) username and sent ":"
485 (i.e. no username, no password) to CURLOPT_USERPWD, but did not do
486 the same when the username is explicitly set to an empty string.
488 * "git clone" of a local repository can be done at the filesystem
489 level, but the codepath did not check errors while copying and
490 adjusting the file that lists alternate object stores.
492 * Documentation for "git commit" was updated to clarify that "commit
493 -p <paths>" adds to the current contents of the index to come up
496 * A stray symbolic link in the $GIT_DIR/refs/ directory could make name
497 resolution loop forever, which has been corrected.
499 * The "submodule.<name>.path" stored in .gitmodules is never copied
500 to .git/config and such a key in .git/config has no meaning, but
501 the documentation described it next to submodule.<name>.url
502 as if both belong to .git/config. This has been fixed.
504 * In a worktree created via "git
505 worktree", "git checkout" attempts to protect users from confusion
506 by refusing to check out a branch that is already checked out in
507 another worktree. However, this also prevented checking out a
508 branch which is designated as the primary branch of a bare
509 repository, in a worktree that is connected to the bare
510 repository. The check has been corrected to allow it.
512 * "git rebase" immediately after "git clone" failed to find the fork
513 point from the upstream.
515 * When fetching from a remote that has many tags that are irrelevant
516 to branches we are following, we used to waste way too many cycles
517 checking if the object pointed at by a tag (that we are not
518 going to fetch!) exists in our repository too carefully.
520 * Protect our code from over-eager compilers.
522 * Recent git allows submodule.<name>.branch to use a special token
523 "." instead of the branch name; the documentation has been updated
526 * "git send-email" attempts to pick up valid e-mails from the
527 trailers, but people in the real world write non-addresses there, like
528 "Cc: Stable <add@re.ss> # 4.8+", which broke the output depending
529 on the availability and vintage of the Mail::Address perl module.
530 (merge dcfafc5214 mm/send-email-cc-cruft-after-address later to maint).
532 * The Travis CI configuration we ship ran the tests with the --verbose
533 option but this risks non-TAP output that happens to be "ok" to be
534 misinterpreted as TAP signalling a test that passed. This resulted
535 in unnecessary failures. This has been corrected by introducing a
536 new mode to run our tests in the test harness to send the verbose
537 output separately to the log file.
539 * Some AsciiDoc formatters mishandle a displayed illustration with
540 tabs in it. Adjust a few of them in merge-base documentation to
543 * Fixed a minor regression in "git submodule" that was introduced
544 when more helper functions were reimplemented in C.
545 (merge 77b63ac31e sb/submodule-ignore-trailing-slash later to maint).
547 * The code that we have used for the past 10+ years to cycle
548 4-element ring buffers turns out to be not quite portable in
550 (merge bb84735c80 rs/ring-buffer-wraparound later to maint).
552 * "git daemon" used fixed-length buffers to turn URLs to the
553 repository the client asked for into the server side directory
554 paths, using snprintf() to avoid overflowing these buffers, but
555 allowed possibly truncated paths to the directory. This has been
556 tightened to reject such a request that causes an overlong path to be
558 (merge 6bdb0083be jk/daemon-path-ok-check-truncation later to maint).
560 * Recent update to git-sh-setup (a library of shell functions that
561 are used by our in-tree scripted Porcelain commands) included
562 another shell library git-sh-i18n without specifying where it is,
563 relying on the $PATH. This has been fixed to be more explicit by
564 prefixing with $(git --exec-path) output.
565 (merge 1073094f30 ak/sh-setup-dot-source-i18n-fix later to maint).
567 * Fix for a racy false-positive test failure.
568 (merge fdf4f6c79b as/merge-attr-sleep later to maint).
570 * Portability update and workaround for builds on recent Mac OS X.
571 (merge a296bc0132 ls/macos-update later to maint).
573 * Using a %(HEAD) placeholder in "for-each-ref --format=" option
574 caused the command to segfault when on an unborn branch.
575 (merge 84679d470d jc/for-each-ref-head-segfault-fix later to maint).
577 * "git rebase -i" did not work well with the core.commentchar
578 configuration variable for two reasons, both of which have been
580 (merge 882cd23777 js/rebase-i-commentchar-fix later to maint).
582 * Other minor doc, test and build updates and code cleanups.
583 (merge 5c238e29a8 jk/common-main later to maint).
584 (merge 5a5749e45b ak/pre-receive-hook-template-modefix later to maint).
585 (merge 6d834ac8f1 jk/rebase-config-insn-fmt-docfix later to maint).
586 (merge de9f7fa3b0 rs/commit-pptr-simplify later to maint).
587 (merge 4259d693fc sc/fmt-merge-msg-doc-markup-fix later to maint).
588 (merge 28fab7b23d nd/test-helpers later to maint).
589 (merge c2bb0c1d1e rs/cocci later to maint).
590 (merge 3285b7badb ps/common-info-doc later to maint).
591 (merge 2b090822e8 nd/worktree-lock later to maint).
592 (merge 4bd488ea7c jk/create-branch-remove-unused-param later to maint).
593 (merge 974e0044d6 tk/diffcore-delta-remove-unused later to maint).
projects / git / blob