| blob | 431bd5e34ace656c3462e5cb4cf6bd3b77adfece |
1 Git 2.16 Release Notes
2 ======================
4 Backward compatibility notes and other notable changes.
6 * Use of an empty string as a pathspec element that is used for
7 'everything matches' is now an error.
10 Updates since v2.15
11 -------------------
13 UI, Workflows & Features
15 * An empty string as a pathspec element that means "everything"
16 i.e. 'git add ""', is now illegal. We started this by first
17 deprecating and warning a pathspec that has such an element in
18 2.11 (Nov 2016).
20 * A hook script that is set unexecutable is simply ignored. Git
21 notifies when such a file is ignored, unless the message is
22 squelched via advice.ignoredHook configuration.
24 * "git pull" has been taught to accept "--[no-]signoff" option and
25 pass it down to "git merge".
27 * The "--push-option=<string>" option to "git push" now defaults to a
28 list of strings configured via push.pushOption variable.
30 * "gitweb" checks if a directory is searchable with Perl's "-x"
31 operator, which can be enhanced by using "filetest 'access'"
32 pragma, which now we do.
34 * "git stash save" has been deprecated in favour of "git stash push".
36 * The set of paths output from "git status --ignored" was tied
37 closely with its "--untracked=<mode>" option, but now it can be
38 controlled more flexibly. Most notably, a directory that is
39 ignored because it is listed to be ignored in the ignore/exclude
40 mechanism can be handled differently from a directory that ends up
41 to be ignored only because all files in it are ignored.
43 * The remote-helper for talking to MediaWiki has been updated to
44 truncate an overlong pagename so that ".mw" suffix can still be
45 added.
47 * The remote-helper for talking to MediaWiki has been updated to
48 work with mediawiki namespaces.
50 * The "--format=..." option "git for-each-ref" takes learned to show
51 the name of the 'remote' repository and the ref at the remote side
52 that is affected for 'upstream' and 'push' via "%(push:remotename)"
53 and friends.
55 * Doc and message updates to teach users "bisect view" is a synonym
56 for "bisect visualize".
58 * "git bisect run" that did not specify any command to run used to go
59 ahead and treated all commits to be tested as 'good'. This has
60 been corrected by making the command error out.
62 * The SubmittingPatches document has been converted to produce an
63 HTML version via AsciiDoc/Asciidoctor.
65 * We learned to talk to watchman to speed up "git status" and other
66 operations that need to see which paths have been modified.
68 * The "diff" family of commands learned to ignore differences in
69 carriage return at the end of line.
71 * Places that know about "sendemail.to", like documentation and shell
72 completion (in contrib/) have been taught about "sendemail.tocmd",
73 too.
75 * "git add --renormalize ." is a new and safer way to record the fact
76 that you are correcting the end-of-line convention and other
77 "convert_to_git()" glitches in the in-repository data.
79 * "git branch" and "git checkout -b" are now forbidden from creating
80 a branch whose name is "HEAD".
82 * "git branch --list" learned to show its output through the pager by
83 default when the output is going to a terminal, which is controlled
84 by the pager.branch configuration variable. This is similar to a
85 recent change to "git tag --list".
87 * "git grep -W", "git diff -W" and their friends learned a heuristic
88 to extend a pre-context beyond the line that matches the "function
89 pattern" (aka "diff.*.xfuncname") to include a comment block, if
90 exists, that immediately precedes it.
92 * "git config --expiry-date gc.reflogexpire" can read "2.weeks" from
93 the configuration and report it as a timestamp, just like "--int"
94 would read "1k" and report 1024, to help consumption by scripts.
96 * The shell completion (in contrib/) learned that "git pull" can take
97 the "--autostash" option.
100 Performance, Internal Implementation, Development Support etc.
102 * An earlier update made it possible to use an on-stack in-core
103 lockfile structure (as opposed to having to deliberately leak an
104 on-heap one). Many codepaths have been updated to take advantage
105 of this new facility.
107 * Calling cmd_foo() as if it is a general purpose helper function is
108 a no-no. Correct two instances of such to set an example.
110 * We try to see if somebody runs our test suite with a shell that
111 does not support "local" like bash/dash does.
113 * An early part of piece-by-piece rewrite of "git bisect" in C.
115 * GSoC to piece-by-piece rewrite "git submodule" in C.
117 * Optimize the code to find shortest unique prefix of object names.
119 * Pathspec-limited revision traversal was taught not to keep finding
120 unneeded differences once it knows two trees are different inside
121 given pathspec.
123 * Conversion from uchar[20] to struct object_id continues.
125 * Code cleanup.
127 * A single-word "unsigned flags" in the diff options is being split
128 into a structure with many bitfields.
130 * TravisCI build updates.
132 * Parts of a test to drive the long-running content filter interface
133 has been split into its own module, hopefully to eventually become
134 reusable.
136 * Drop (perhaps overly cautious) sanity check before using the index
137 read from the filesystem at runtime.
139 * The build procedure has been taught to avoid some unnecessary
140 instability in the build products.
142 * A new mechanism to upgrade the wire protocol in place is proposed
143 and demonstrated that it works with the older versions of Git
144 without harming them.
146 Also contains various documentation updates and code clean-ups.
149 Fixes since v2.15
150 -----------------
152 * "auto" as a value for the columnar output configuration ought to
153 judge "is the output consumed by humans?" with the same criteria as
154 "auto" for coloured output configuration, i.e. either the standard
155 output stream is going to tty, or a pager is in use. We forgot the
156 latter, which has been fixed.
158 * The experimental "color moved lines differently in diff output"
159 feature was buggy around "ignore whitespace changes" edges, which
160 has been corrected.
162 * Instead of using custom line comparison and hashing functions to
163 implement "moved lines" coloring in the diff output, use the pair
164 of these functions from lower-layer xdiff/ code.
166 * Some codepaths did not check for errors when asking what branch the
167 HEAD points at, which have been fixed.
169 * "git commit", after making a commit, did not check for errors when
170 asking on what branch it made the commit, which has been corrected.
172 * "git status --ignored -u" did not stop at a working tree of a
173 separate project that is embedded in an ignored directory and
174 listed files in that other project, instead of just showing the
175 directory itself as ignored.
177 * A broken access to object databases in recent update to "git grep
178 --recurse-submodules" has been fixed.
180 * A recent regression in "git rebase -i" that broke execution of git
181 commands from subdirectories via "exec" instruction has been fixed.
183 * A (possibly flakey) test fix.
185 * "git check-ref-format --branch @{-1}" bit a "BUG()" when run
186 outside a repository for obvious reasons; clarify the documentation
187 and make sure we do not even try to expand the at-mark magic in
188 such a case, but still call the validation logic for branch names.
190 * "git fetch --recurse-submodules" now knows that submodules can be
191 moved around in the superproject in addition to getting updated,
192 and finds the ones that need to be fetched accordingly.
194 * Command line completion (in contrib/) update.
196 * Description of blame.{showroot,blankboundary,showemail,date}
197 configuration variables have been added to "git config --help".
199 * After an error from lstat(), diff_populate_filespec() function
200 sometimes still went ahead and used invalid data in struct stat,
201 which has been fixed.
203 * UNC paths are also relevant in Cygwin builds and they are now
204 tested just like Mingw builds.
206 * Correct start-up sequence so that a repository could be placed
207 immediately under the root directory again (which was broken at
208 around Git 2.13).
210 * The credential helper for libsecret (in contrib/) has been improved
211 to allow possibly prompting the end user to unlock secrets that are
212 currently locked (otherwise the secrets may not be loaded).
214 * MinGW updates.
216 * Error checking in "git imap-send" for empty response has been
217 improved.
219 * Recent update to the refs infrastructure implementation started
220 rewriting packed-refs file more often than before; this has been
221 optimized again for most trivial cases.
223 * Some error messages did not quote filenames shown in it, which have
224 been fixed.
226 * "git rebase -i" recently started misbehaving when a submodule that
227 is configured with 'submodule.<name>.ignore' is dirty; this has
228 been corrected.
230 * Building with NO_LIBPCRE1_JIT did not disable it, which has been fixed.
232 * We used to add an empty alternate object database to the system
233 that does not help anything; it has been corrected.
235 * Doc update around use of "format-patch --subject-prefix" etc.
237 * A fix for an ancient bug in "git apply --ignore-space-change" codepath.
239 * Clarify and enhance documentation for "merge-base --fork-point", as
240 it was clear what it computed but not why/what for.
242 * A few scripts (both in production and tests) incorrectly redirected
243 their error output. These have been corrected.
245 * "git notes" sent its error message to its standard output stream,
246 which was corrected.
248 * The three-way merge performed by "git cherry-pick" was confused
249 when a new submodule was added in the meantime, which has been
250 fixed (or "papered over").
252 * The sequencer machinery (used by "git cherry-pick A..B", and "git
253 rebase -i", among other things) would have lost a commit if stopped
254 due to an unlockable index file, which has been fixed.
256 * "git apply --inaccurate-eof" when used with "--ignore-space-change"
257 triggered an internal sanity check, which has been fixed.
259 * Command line completion (in contrib/) has been taught about the
260 "--copy" option of "git branch".
262 * When "git rebase" prepared an mailbox of changes and fed it to "git
263 am" to replay them, it was confused when a stray "From " happened
264 to be in the log message of one of the replayed changes. This has
265 been corrected.
267 * There was a recent semantic mismerge in the codepath to write out a
268 section of a configuration section, which has been corrected.
270 * Mentions of "git-rebase" and "git-am" (dashed form) still remained
271 in end-user visible strings emitted by the "git rebase" command;
272 they have been corrected.
274 * Contrary to the documentation, "git pull -4/-6 other-args" did not
275 ask the underlying "git fetch" to go over IPv4/IPv6, which has been
276 corrected.
278 * "git checkout --recursive" may overwrite and rewind the history of
279 the branch that happens to be checked out in submodule
280 repositories, which might not be desirable. Detach the HEAD but
281 still allow the recursive checkout to succeed in such a case.
282 (merge 57f22bf997 sb/submodule-recursive-checkout-detach-head later to maint).
284 * "git branch --set-upstream" has been deprecated and (sort of)
285 removed, as "--set-upstream-to" is the preferred one these days.
286 The documentation still had "--set-upstream" listed on its
287 synopsys section, which has been corrected.
288 (merge a060f3d3d8 tz/branch-doc-remove-set-upstream later to maint).
290 * Internaly we use 0{40} as a placeholder object name to signal the
291 codepath that there is no such object (e.g. the fast-forward check
292 while "git fetch" stores a new remote-tracking ref says "we know
293 there is no 'old' thing pointed at by the ref, as we are creating
294 it anew" by passing 0{40} for the 'old' side), and expect that a
295 codepath to locate an in-core object to return NULL as a sign that
296 the object does not exist. A look-up for an object that does not
297 exist however is quite costly with a repository with large number
298 of packfiles. This access pattern has been optimized.
299 (merge 87b5e236a1 jk/fewer-pack-rescan later to maint).
301 * In addition to "git stash -m message", the command learned to
302 accept "git stash -mmessage" form.
303 (merge 5675473fcb ph/stash-save-m-option-fix later to maint).
305 * Other minor doc, test and build updates and code cleanups.
306 (merge 1a1fc2d5b5 rd/man-prune-progress later to maint).
307 (merge 0ba014035a rd/man-reflog-add-n later to maint).
308 (merge e54b63359f rd/doc-notes-prune-fix later to maint).
309 (merge ff4c9b413a sp/doc-info-attributes later to maint).