| blob | fd5aecf8e96ec807c6dd4e131704da8f2ee576ea |
1 Git 2.18 Release Notes
2 ======================
4 Updates since v2.17
5 -------------------
7 UI, Workflows & Features
9 * Rename detection logic in "diff" family that is used in "merge" has
10 learned to guess when all of x/a, x/b and x/c have moved to z/a,
11 z/b and z/c, it is likely that x/d added in the meantime would also
12 want to move to z/d by taking the hint that the entire directory
13 'x' moved to 'z'. A bug causing dirty files involved in a rename
14 to be overwritten during merge has also been fixed as part of this
15 work. Incidentally, this also avoids updating a file in the
16 working tree after a (non-trivial) merge whose result matches what
17 our side originally had.
19 * "git filter-branch" learned to use a different exit code to allow
20 the callers to tell the case where there was no new commits to
21 rewrite from other error cases.
23 * When built with more recent cURL, GIT_SSL_VERSION can now specify
24 "tlsv1.3" as its value.
26 * "git gui" learned that "~/.ssh/id_ecdsa.pub" and
27 "~/.ssh/id_ed25519.pub" are also possible SSH key files.
28 (merge 2e2f0288ef bb/git-gui-ssh-key-files later to maint).
30 * "git gui" performs commit upon CTRL/CMD+ENTER but the
31 CTRL/CMD+KP_ENTER (i.e. enter key on the numpad) did not have the
32 same key binding. It now does.
33 (merge 28a1d94a06 bp/git-gui-bind-kp-enter later to maint).
35 * "git gui" has been taught to work with old versions of tk (like
36 8.5.7) that do not support "ttk::style theme use" as a way to query
37 the current theme.
38 (merge 4891961105 cb/git-gui-ttk-style later to maint).
40 * "git rebase" has learned to honor "--signoff" option when using
41 backends other than "am" (but not "--preserve-merges").
43 * "git branch --list" during an interrupted "rebase -i" now lets
44 users distinguish the case where a detached HEAD is being rebased
45 and a normal branch is being rebased.
47 * "git mergetools" learned talking to guiffy.
49 * The scripts in contrib/emacs/ have outlived their usefulness and
50 have been replaced with a stub that errors out and tells the user
51 there are replacements.
53 * The new "checkout-encoding" attribute can ask Git to convert the
54 contents to the specified encoding when checking out to the working
55 tree (and the other way around when checking in).
57 * The "git config" command uses separate options e.g. "--int",
58 "--bool", etc. to specify what type the caller wants the value to
59 be interpreted as. A new "--type=<typename>" option has been
60 introduced, which would make it cleaner to define new types.
62 * "git config --get" learned the "--default" option, to help the
63 calling script. Building on top of the above changes, the
64 "git config" learns "--type=color" type. Taken together, you can
65 do things like "git config --get foo.color --default blue" and get
66 the ANSI color sequence for the color given to foo.color variable,
67 or "blue" if the variable does not exist.
69 * "git ls-remote" learned an option to allow sorting its output based
70 on the refnames being shown.
72 * The command line completion (in contrib/) has been taught that "git
73 stash save" has been deprecated ("git stash push" is the preferred
74 spelling in the new world) and does not offer it as a possible
75 completion candidate when "git stash push" can be.
77 * "git gc --prune=nonsense" spent long time repacking and then
78 silently failed when underlying "git prune --expire=nonsense"
79 failed to parse its command line. This has been corrected.
81 * Error messages from "git push" can be painted for more visibility.
83 * "git http-fetch" (deprecated) had an optional and experimental
84 "feature" to fetch only commits and/or trees, which nobody used.
85 This has been removed.
87 * The functionality of "$GIT_DIR/info/grafts" has been superseded by
88 the "refs/replace/" mechanism for some time now, but the internal
89 code had support for it in many places, which has been cleaned up
90 in order to drop support of the "grafts" mechanism.
92 * "git worktree add" learned to check out an existing branch.
94 * "git --no-pager cmd" did not have short-and-sweet single letter
95 option. Now it does as "-P".
96 (merge 7213c28818 js/no-pager-shorthand later to maint).
98 * "git rebase" learned "--rebase-merges" to transplant the whole
99 topology of commit graph elsewhere.
101 * "git status" learned to pay attention to UI related diff
102 configuration variables such as diff.renames.
104 * The command line completion mechanism (in contrib/) learned to load
105 custom completion file for "git $command" where $command is a
106 custom "git-$command" that the end user has on the $PATH when using
107 newer version of bash.
109 * "git send-email" can sometimes offer confirmation dialog "Send this
110 email?" with choices 'Yes', 'No', 'Quit', and 'All'. A new action
111 'Edit' has been added to this dialog's choice.
113 * With merge.renames configuration set to false, the recursive merge
114 strategy can be told not to spend cycles trying to find renamed
115 paths and merge them accordingly.
117 * "git status" learned to honor a new status.renames configuration to
118 skip rename detection, which could be useful for those who want to
119 do so without disabling the default rename detection done by the
120 "git diff" command.
122 * Command line completion (in contrib/) learned to complete pathnames
123 for various commands better.
125 * "git blame" learns to unhighlight uninteresting metadata from the
126 originating commit on lines that are the same as the previous one,
127 and also paint lines in different colors depending on the age of
128 the commit.
130 * Transfer protocol v2 learned to support the partial clone.
132 * When a short hexadecimal string is used to name an object but there
133 are multiple objects that share the string as the prefix of their
134 names, the code lists these ambiguous candidates in a help message.
135 These object names are now sorted according to their types for
136 easier eyeballing.
138 * "git fetch $there $refspec" that talks over protocol v2 can take
139 advantage of server-side ref filtering; the code has been extended
140 so that this mechanism triggers also when fetching with configured
141 refspec.
143 * Our HTTP client code used to advertise that we accept gzip encoding
144 from the other side; instead, just let cURL library to advertise
145 and negotiate the best one.
148 Performance, Internal Implementation, Development Support etc.
150 * A "git fetch" from a repository with insane number of refs into a
151 repository that is already up-to-date still wasted too many cycles
152 making many lstat(2) calls to see if these objects at the tips
153 exist as loose objects locally. These lstat(2) calls are optimized
154 away by enumerating all loose objects beforehand.
155 It is unknown if the new strategy negatively affects existing use
156 cases, fetching into a repository with many loose objects from a
157 repository with small number of refs.
159 * Git can be built to use either v1 or v2 of the PCRE library, and so
160 far, the build-time configuration USE_LIBPCRE=YesPlease instructed
161 the build procedure to use v1, but now it means v2. USE_LIBPCRE1
162 and USE_LIBPCRE2 can be used to explicitly choose which version to
163 use, as before.
165 * The build procedure learned to optionally use symbolic links
166 (instead of hardlinks and copies) to install "git-foo" for built-in
167 commands, whose binaries are all identical.
169 * Conversion from uchar[20] to struct object_id continues.
171 * The way "git worktree prune" worked internally has been simplified,
172 by assuming how "git worktree move" moves an existing worktree to a
173 different place.
175 * Code clean-up for the "repository" abstraction.
176 (merge 00a3da2a13 nd/remove-ignore-env-field later to maint).
178 * Code to find the length to uniquely abbreviate object names based
179 on packfile content, which is a relatively recent addtion, has been
180 optimized to use the same fan-out table.
182 * The mechanism to use parse-options API to automate the command line
183 completion continues to get extended and polished.
185 * Copies of old scripted Porcelain commands in contrib/examples/ have
186 been removed.
188 * Some tests that rely on the exact hardcoded values of object names
189 have been updated in preparation for hash function migration.
191 * Perf-test update.
193 * Test helper update.
195 * The effort continues to refactor the internal global data structure
196 to make it possible to open multiple repositories, work with and
197 then close them,
199 * Small test-helper programs have been consolidated into a single
200 binary.
202 * API clean-up around ref-filter code.
204 * Shell completion (in contrib) that gives list of paths have been
205 optimized somewhat.
207 * The index file is updated to record the fsmonitor section after a
208 full scan was made, to avoid wasting the effort that has already
209 spent.
211 * Performance measuring framework in t/perf learned to help bisecting
212 performance regressions.
214 * Some multi-word source filenames are being renamed to separate
215 words with dashes instead of underscores.
217 * An reusable "memory pool" implementation has been extracted from
218 fast-import.c, which in turn has become the first user of the
219 mem-pool API.
221 * A build-time option has been added to allow Git to be told to refer
222 to its associated files relative to the main binary, in the same
223 way that has been possible on Windows for quite some time, for
224 Linux, BSDs and Darwin.
226 * Precompute and store information necessary for ancestry traversal
227 in a separate file to optimize graph walking.
229 * The effort to pass the repository in-core structure throughout the
230 API continues. This round deals with the code that implements the
231 refs/replace/ mechanism.
233 * The build procedure "make DEVELOPER=YesPlease" learned to enable a
234 bit more warning options depending on the compiler used to help
235 developers more. There also is "make DEVOPTS=tokens" knob
236 available now, for those who want to help fixing warnings we
237 usually ignore, for example.
239 * A new version of the transport protocol is being worked on.
241 * The code to interface to GPG has been restructured somewhat to make
242 it cleaner to integrate with other types of signature systems later.
244 * The code has been taught to use the duplicated information stored
245 in the commit-graph file to learn the tree object name for a commit
246 to avoid opening and parsing the commit object when it makes sense
247 to do so.
249 * "git gc" in a large repository takes a lot of time as it considers
250 to repack all objects into one pack by default. The command has
251 been taught to pretend as if the largest existing packfile is
252 marked with ".keep" so that it is left untouched while objects in
253 other packs and loose ones are repacked.
255 * The transport protocol v2 is getting updated further.
257 * The codepath around object-info API has been taught to take the
258 repository object (which in turn tells the API which object store
259 the objects are to be located).
261 * "git pack-objects" needs to allocate tons of "struct object_entry"
262 while doing its work, and shrinking its size helps the performance
263 quite a bit.
265 * The implementation of "git rebase -i --root" has been updated to use
266 the sequencer machinery more.
268 * Developer support update, by using BUG() macro instead of die() to
269 mark codepaths that should not happen more clearly.
271 * Developer support. Use newer GCC on one of the builds done at
272 TravisCI.org to get more warnings and errors diagnosed.
274 * Conversion from uchar[20] to struct object_id continues.
276 * By code restructuring of submodule merge in merge-recursive,
277 informational messages from the codepath are now given using the
278 same mechanism as other output, and honor the merge.verbosity
279 configuration. The code also learned to give a few new messages
280 when a submodule three-way merge resolves cleanly when one side
281 records a descendant of the commit chosen by the other side.
283 * Avoid unchecked snprintf() to make future code auditing easier.
284 (merge ac4896f007 jk/snprintf-truncation later to maint).
286 * Many tests hardcode the raw object names, which would change once
287 we migrate away from SHA-1. While some of them must test against
288 exact object names, most of them do not have to use hardcoded
289 constants in the test. The latter kind of tests have been updated
290 to test the moral equivalent of the original without hardcoding the
291 actual object names.
294 Also contains various documentation updates and code clean-ups.
297 Fixes since v2.17
298 -----------------
300 * "git shortlog cruft" aborted with a BUG message when run outside a
301 Git repository. The command has been taught to complain about
302 extra and unwanted arguments on its command line instead in such a
303 case.
304 (merge 4aa0161e83 ma/shortlog-revparse later to maint).
306 * "git stash push -u -- <pathspec>" gave an unnecessary and confusing
307 error message when there was no tracked files that match the
308 <pathspec>, which has been fixed.
309 (merge 353278687e tg/stash-untracked-with-pathspec-fix later to maint).
311 * "git tag --contains no-such-commit" gave a full list of options
312 after giving an error message.
313 (merge 3bb0923f06 ps/contains-id-error-message later to maint).
315 * "diff-highlight" filter (in contrib/) learned to undertand "git log
316 --graph" output better.
317 (merge 4551fbba14 jk/diff-highlight-graph-fix later to maint).
319 * when refs that do not point at committish are given, "git
320 filter-branch" gave a misleading error messages. This has been
321 corrected.
322 (merge f78ab355e7 yk/filter-branch-non-committish-refs later to maint).
324 * "git submodule status" misbehaved on a submodule that has been
325 removed from the working tree.
326 (merge 74b6bda32f rs/status-with-removed-submodule later to maint).
328 * When credential helper exits very quickly without reading its
329 input, it used to cause Git to die with SIGPIPE, which has been
330 fixed.
331 (merge a0d51e8d0e eb/cred-helper-ignore-sigpipe later to maint).
333 * "git rebase --keep-empty" still removed an empty commit if the
334 other side contained an empty commit (due to the "does an
335 equivalent patch exist already?" check), which has been corrected.
336 (merge 3d946165e1 pw/rebase-keep-empty-fixes later to maint).
338 * Some codepaths, including the refs API, get and keep relative
339 paths, that go out of sync when the process does chdir(2). The
340 chdir-notify API is introduced to let these codepaths adjust these
341 cached paths to the new current directory.
342 (merge fb9c2d2703 jk/relative-directory-fix later to maint).
344 * "cd sub/dir && git commit ../path" ought to record the changes to
345 the file "sub/path", but this regressed long time ago.
346 (merge 86238e07ef bw/commit-partial-from-subdirectory-fix later to maint).
348 * Recent introduction of "--log-destination" option to "git daemon"
349 did not work well when the daemon was run under "--inetd" mode.
350 (merge e67d906d73 lw/daemon-log-destination later to maint).
352 * Small fix to the autoconf build procedure.
353 (merge 249482daf0 es/fread-reads-dir-autoconf-fix later to maint).
355 * Fix an unexploitable (because the oversized contents are not under
356 attacker's control) buffer overflow.
357 (merge d8579accfa bp/fsmonitor-bufsize-fix later to maint).
359 * Recent simplification of build procedure forgot a bit of tweak to
360 the build procedure of contrib/mw-to-git/
361 (merge d8698987f3 ab/simplify-perl-makefile later to maint).
363 * Moving a submodule that itself has submodule in it with "git mv"
364 forgot to make necessary adjustment to the nested sub-submodules;
365 now the codepath learned to recurse into the submodules.
367 * "git config --unset a.b", when "a.b" is the last variable in an
368 otherwise empty section "a", left an empty section "a" behind, and
369 worse yet, a subsequent "git config a.c value" did not reuse that
370 empty shell and instead created a new one. These have been
371 (partially) corrected.
372 (merge c71d8bb38a js/empty-config-section-fix later to maint).
374 * "git worktree remove" learned that "-f" is a shorthand for
375 "--force" option, just like for "git worktree add".
376 (merge d228eea514 sb/worktree-remove-opt-force later to maint).
378 * The completion script (in contrib/) learned to clear cached list of
379 command line options upon dot-sourcing it again in a more efficient
380 way.
381 (merge 94408dc71c sg/completion-clear-cached later to maint).
383 * "git svn" had a minor thinko/typo which has been fixed.
384 (merge 51db271587 ab/git-svn-get-record-typofix later to maint).
386 * During a "rebase -i" session, the code could give older timestamp
387 to commits created by later "pick" than an earlier "reword", which
388 has been corrected.
389 (merge 12f7babd6b js/ident-date-fix later to maint).
391 * "git submodule status" did not check the symbolic revision name it
392 computed for the submodule HEAD is not the NULL, and threw it at
393 printf routines, which has been corrected.
394 (merge 0b5e2ea7cf nd/submodule-status-fix later to maint).
396 * When fed input that already has In-Reply-To: and/or References:
397 headers and told to add the same information, "git send-email"
398 added these headers separately, instead of appending to an existing
399 one, which is a violation of the RFC. This has been corrected.
400 (merge 256be1d3f0 sa/send-email-dedup-some-headers later to maint).
402 * "git fast-export" had a regression in v2.15.0 era where it skipped
403 some merge commits in certain cases, which has been corrected.
404 (merge be011bbe00 ma/fast-export-skip-merge-fix later to maint).
406 * The code did not propagate the terminal width to subprocesses via
407 COLUMNS environment variable, which it now does. This caused
408 trouble to "git column" helper subprocess when "git tag --column=row"
409 tried to list the existing tags on a display with non-default width.
410 (merge b5d5a567fb nd/term-columns later to maint).
412 * We learned that our source files with ".pl" and ".py" extensions
413 are Perl and Python files respectively and changes to them are
414 better viewed as such with appropriate diff drivers.
415 (merge 7818b619e2 ab/perl-python-attrs later to maint).
417 * "git rebase -i" sometimes left intermediate "# This is a
418 combination of N commits" message meant for the human consumption
419 inside an editor in the final result in certain corner cases, which
420 has been fixed.
421 (merge 15ef69314d js/rebase-i-clean-msg-after-fixup-continue later to maint).
423 * A test to see if the filesystem normalizes UTF-8 filename has been
424 updated to check what we need to know in a more direct way, i.e. a
425 path created in NFC form can be accessed with NFD form (or vice
426 versa) to cope with APFS as well as HFS.
427 (merge 742ae10e35 tb/test-apfs-utf8-normalization later to maint).
429 * "git format-patch --cover --attach" created a broken MIME multipart
430 message for the cover letter, which has been fixed by keeping the
431 cover letter as plain text file.
432 (merge 50cd54ef4e bc/format-patch-cover-no-attach later to maint).
434 * The split-index feature had a long-standing and dormant bug in
435 certain use of the in-core merge machinery, which has been fixed.
436 (merge 7db118303a en/unpack-trees-split-index-fix later to maint).
438 * Asciidoctor gives a reasonable imitation for AsciiDoc, but does not
439 render illustration in a literal block correctly when indented with
440 HT by default. The problem is fixed by forcing 8-space tabs.
441 (merge 379805051d bc/asciidoctor-tab-width later to maint).
443 * Code clean-up to adjust to a more recent lockfile API convention that
444 allows lockfile instances kept on the stack.
445 (merge 0fa5a2ed8d ma/lockfile-cleanup later to maint).
447 * the_repository->index is not a allocated piece of memory but
448 repo_clear() indiscriminately attempted to free(3) it, which has
449 been corrected.
450 (merge 74373b5f10 nd/repo-clear-keep-the-index later to maint).
452 * Code clean-up to avoid non-standard-conformant pointer arithmetic.
453 (merge c112084af9 rs/no-null-ptr-arith-in-fast-export later to maint).
455 * Code clean-up to turn history traversal more robust in a
456 semi-corrupt repository.
457 (merge 8702b30fd7 jk/unavailable-can-be-missing later to maint).
459 * "git update-ref A B" is supposed to ensure that ref A does not yet
460 exist when B is a NULL OID, but this check was not done correctly
461 for pseudo-refs outside refs/ hierarchy, e.g. MERGE_HEAD.
463 * "git submodule update" and "git submodule add" supported the
464 "--reference" option to borrow objects from a neighbouring local
465 repository like "git clone" does, but lacked the more recent
466 invention "--dissociate". Also "git submodule add" has been taught
467 to take the "--progress" option.
468 (merge a0ef29341a cf/submodule-progress-dissociate later to maint).
470 * Update credential-netrc helper (in contrib/) to allow customizing
471 the GPG used to decrypt the encrypted .netrc file.
472 (merge 786ef50a23 lm/credential-netrc later to maint).
474 * "git submodule update" attempts two different kinds of "git fetch"
475 against the upstream repository to grab a commit bound at the
476 submodule's path, but it incorrectly gave up if the first kind
477 (i.e. a normal fetch) failed, making the second "last resort" one
478 (i.e. fetching an exact commit object by object name) ineffective.
479 This has been corrected.
480 (merge e30d833671 sb/submodule-update-try-harder later to maint).
482 * Error behaviour of "git grep" when it cannot read the index was
483 inconsistent with other commands that uses the index, which has
484 been corrected to error out early.
485 (merge b2aa84c789 sb/grep-die-on-unreadable-index later to maint).
487 * We used to call regfree() after regcomp() failed in some codepaths,
488 which have been corrected.
489 (merge 17154b1576 ma/regex-no-regfree-after-comp-fail later to maint).
491 * Other minor doc, test and build updates and code cleanups.
492 (merge 248f66ed8e nd/trace-with-env later to maint).
493 (merge 14ced5562c ys/bisect-object-id-missing-conversion-fix later to maint).
494 (merge 5988eb631a ab/doc-hash-brokenness later to maint).
495 (merge a4d4e32a70 pk/test-avoid-pipe-hiding-exit-status later to maint).
496 (merge 05e293c1ac jk/flockfile-stdio later to maint).
497 (merge e9184b0789 jk/t5561-missing-curl later to maint).
498 (merge b1801b85a3 nd/worktree-move later to maint).
499 (merge bbd374dd20 ak/bisect-doc-typofix later to maint).
500 (merge 4855f06fb3 mn/send-email-credential-doc later to maint).
501 (merge 8523b1e355 en/doc-typoes later to maint).
502 (merge 43b44ccfe7 js/t5404-path-fix later to maint).
503 (merge decf711fc1 ps/test-chmtime-get later to maint).
504 (merge 22d11a6e8e es/worktree-docs later to maint).
505 (merge 92a5dbbc22 tg/use-git-contacts later to maint).
506 (merge adc887221f tq/t1510 later to maint).
507 (merge bed21a8ad6 sg/doc-gc-quote-mismatch-fix later to maint).
508 (merge 73364e4f10 tz/doc-git-urls-reference later to maint).
509 (merge cd1e606bad bc/mailmap-self later to maint).
510 (merge f7997e3682 ao/config-api-doc later to maint).
511 (merge ee930754d8 jk/apply-p-doc later to maint).
512 (merge 011b648646 nd/pack-format-doc later to maint).
513 (merge 87a6bb701a sg/t5310-jgit-bitmap-test later to maint).
514 (merge f6b82970aa sg/t5516-fixes later to maint).
515 (merge 4362da078e sg/t7005-spaces-in-filenames-cleanup later to maint).
516 (merge 7d0ee47c11 js/test-unset-prereq later to maint).
517 (merge 5356a3c354 ah/misc-doc-updates later to maint).
518 (merge 92c4a7a129 nd/completion-aliasfiletype-typofix later to maint).
519 (merge 58bd77b66a nd/pack-unreachable-objects-doc later to maint).
520 (merge 4ed79d5203 sg/t6500-no-redirect-of-stdin later to maint).
521 (merge 17b8a2d6cd jk/config-blob-sans-repo later to maint).