| blob | 993e49341ead43a5e02dc04b2506ec09747b5bd0 |
1 Git 2.13 Release Notes
2 ======================
4 Backward compatibility notes.
6 * Use of an empty string as a pathspec element that is used for
7 'everything matches' is still warned and Git asks users to use a
8 more explicit '.' for that instead. The hope is that existing
9 users will not mind this change, and eventually the warning can be
10 turned into a hard error, upgrading the deprecation into removal of
11 this (mis)feature. That is not scheduled to happen in the upcoming
12 release (yet).
14 * The historical argument order "git merge <msg> HEAD <commit>..."
15 has been deprecated for quite some time, and will be removed in a
16 future release.
18 * The default location "~/.git-credential-cache/socket" for the
19 socket used to communicate with the credential-cache daemon has
20 been moved to "~/.cache/git/credential/socket".
23 Updates since v2.12
24 -------------------
26 UI, Workflows & Features
28 * "git describe" and "git name-rev" have been taught to take more
29 than one refname patterns to restrict the set of refs to base their
30 naming output on, and also learned to take negative patterns to
31 name refs not to be used for naming via their "--exclude" option.
33 * Deletion of a branch "foo/bar" could remove .git/refs/heads/foo
34 once there no longer is any other branch whose name begins with
35 "foo/", but we didn't do so so far. Now we do.
37 * When "git merge" detects a path that is renamed in one history
38 while the other history deleted (or modified) it, it now reports
39 both paths to help the user understand what is going on in the two
40 histories being merged.
42 * The <url> part in "http.<url>.<variable>" configuration variable
43 can now be spelled with '*' that serves as wildcard.
44 E.g. "http.https://*.example.com.proxy" can be used to specify the
45 proxy used for https://a.example.com, https://b.example.com, etc.,
46 i.e. any host in the example.com domain.
48 * "git tag" did not leave useful message when adding a new entry to
49 reflog; this was left unnoticed for a long time because refs/tags/*
50 doesn't keep reflog by default.
52 * The "negative" pathspec feature was somewhat more cumbersome to use
53 than necessary in that its short-hand used "!" which needed to be
54 escaped from shells, and it required "exclude from what?" specified.
56 * The command line options for ssh invocation needs to be tweaked for
57 some implementations of SSH (e.g. PuTTY plink wants "-P <port>"
58 while OpenSSH wants "-p <port>" to specify port to connect to), and
59 the variant was guessed when GIT_SSH environment variable is used
60 to specify it. The logic to guess now applies to the command
61 specified by the newer GIT_SSH_COMMAND and also core.sshcommand
62 configuration variable, and comes with an escape hatch for users to
63 deal with misdetected cases.
65 * The "--git-path", "--git-common-dir", and "--shared-index-path"
66 options of "git rev-parse" did not produce usable output. They are
67 now updated to show the path to the correct file, relative to where
68 the caller is.
70 * "git diff -W" has been taught to handle the case where a new
71 function is added at the end of the file better.
73 * "git update-ref -d" and other operations to delete references did
74 not leave any entry in HEAD's reflog when the reference being
75 deleted was the current branch. This is not a problem in practice
76 because you do not want to delete the branch you are currently on,
77 but caused renaming of the current branch to something else not to
78 be logged in a useful way.
80 * "Cc:" on the trailer part does not have to conform to RFC strictly,
81 unlike in the e-mail header. "git send-email" has been updated to
82 ignore anything after '>' when picking addresses, to allow non-address
83 cruft like " # stable 4.4" after the address.
85 * When "git submodule init" decides that the submodule in the working
86 tree is its upstream, it now gives a warning as it is not a very
87 common setup.
89 * "git stash push" takes a pathspec so that the local changes can be
90 stashed away only partially.
92 * Documentation for "git ls-files" did not refer to core.quotePath.
94 * The experimental "split index" feature has gained a few
95 configuration variables to make it easier to use.
97 * From a working tree of a repository, a new option of "rev-parse"
98 lets you ask if the repository is used as a submodule of another
99 project, and where the root level of the working tree of that
100 project (i.e. your superproject) is.
102 * The pathspec mechanism learned to further limit the paths that
103 match the pattern to those that have specified attributes attached
104 via the gitattributes mechanism.
106 * Our source code has used the SHA1_HEADER cpp macro after "#include"
107 in the C code to switch among the SHA-1 implementations. Instead,
108 list the exact header file names and switch among implementations
109 using "#ifdef BLK_SHA1/#include "block-sha1/sha1.h"/.../#endif";
110 this helps some IDE tools.
112 * The start-up sequence of "git" needs to figure out some configured
113 settings before it finds and set itself up in the location of the
114 repository and was quite messy due to its "chicken-and-egg" nature.
115 The code has been restructured.
117 * The command line prompt (in contrib/) learned a new 'tag' style
118 that can be specified with GIT_PS1_DESCRIBE_STYLE, to describe a
119 detached HEAD with "git describe --tags".
121 * The configuration file learned a new "includeIf.<condition>.path"
122 that includes the contents of the given path only when the
123 condition holds. This allows you to say "include this work-related
124 bit only in the repositories under my ~/work/ directory".
126 * Recent update to "rebase -i" started showing a message that is not
127 a warning with "warning:" prefix by mistake. This has been fixed.
129 * Recently we started passing the "--push-options" through the
130 external remote helper interface; now the "smart HTTP" remote
131 helper understands what to do with the passed information.
133 * "git describe --dirty" dies when it cannot be determined if the
134 state in the working tree matches that of HEAD (e.g. broken
135 repository or broken submodule). The command learned a new option
136 "git describe --broken" to give "$name-broken" (where $name is the
137 description of HEAD) in such a case.
139 * "git checkout" is taught the "--recurse-submodules" option.
141 * Recent enhancement to "git stash push" command to support pathspec
142 to allow only a subset of working tree changes to be stashed away
143 was found to be too chatty and exposed the internal implementation
144 detail (e.g. when it uses reset to match the index to HEAD before
145 doing other things, output from reset seeped out). These, and
146 other chattyness has been fixed.
149 Performance, Internal Implementation, Development Support etc.
151 * The code to list branches in "git branch" has been consolidated
152 with the more generic ref-filter API.
154 * Resource usage while enumerating refs from alternate object store
155 has been optimized to help receiving end of "push" that hosts a
156 repository with many "forks".
158 * The gitattributes machinery is being taught to work better in a
159 multi-threaded environment.
161 * "git rebase -i" starts using the recently updated "sequencer" code.
163 * Code and design clean-up for the refs API.
165 * The preload-index code has been taught not to bother with the index
166 entries that are paths that are not checked out by "sparse checkout".
168 * Some warning() messages from "git clean" were updated to show the
169 errno from failed system calls.
171 * The "parse_config_key()" API function has been cleaned up.
173 * A test that creates a confusing branch whose name is HEAD has been
174 corrected not to do so.
176 * The code that parses header fields in the commit object has been
177 updated for (micro)performance and code hygiene.
179 * An helper function to make it easier to append the result from
180 real_path() to a strbuf has been added.
182 * Reduce authentication round-trip over HTTP when the server supports
183 just a single authentication method. This also improves the
184 behaviour when Git is misconfigured to enable http.emptyAuth
185 against a server that does not authenticate without a username
186 (i.e. not using Kerberos etc., which makes http.emptyAuth
187 pointless).
189 * Windows port wants to use OpenSSL's implementation of SHA-1
190 routines, so let them.
192 * The t/perf performance test suite was not prepared to test not so
193 old versions of Git, but now it covers versions of Git that are not
194 so ancient.
196 * Add 32-bit Linux variant to the set of platforms to be tested with
197 Travis CI.
199 * "git branch --list" takes the "--abbrev" and "--no-abbrev" options
200 to control the output of the object name in its "-v"(erbose)
201 output, but a recent update started ignoring them; fix it before
202 the breakage reaches to any released version.
204 * Picking two versions of Git and running tests to make sure the
205 older one and the newer one interoperate happily has now become
206 possible.
208 * "uchar [40]" to "struct object_id" conversion continues.
210 * "git tag --contains" used to (ab)use the object bits to keep track
211 of the state of object reachability without clearing them after
212 use; this has been cleaned up and made to use the newer commit-slab
213 facility.
215 * The "debug" helper used in the test framework learned to run
216 a command under "gdb" interactively.
218 * The "detect attempt to create collisions" variant of SHA-1
219 implementation by Marc Stevens (CWI) and Dan Shumow (Microsoft)
220 has been integrated and made the default.
222 * The test framework learned to detect unterminated here documents.
224 * The name-hash used for detecting paths that are different only in
225 cases (which matter on case insensitive filesystems) has been
226 optimized to take advantage of multi-threading when it makes sense.
228 * An earlier version of sha1dc/sha1.c that was merged to 'master'
229 compiled incorrectly on Windows, which has been fixed.
232 Also contains various documentation updates and code clean-ups.
235 Fixes since v2.12
236 -----------------
238 Unless otherwise noted, all the fixes since v2.12 in the maintenance
239 track are contained in this release (see the maintenance releases'
240 notes for details).
242 * "git repack --depth=<n>" for a long time busted the specified depth
243 when reusing delta from existing packs. This has been corrected.
245 * The code to parse the command line "git grep <patterns>... <rev>
246 [[--] <pathspec>...]" has been cleaned up, and a handful of bugs
247 have been fixed (e.g. we used to check "--" if it is a rev).
249 * "git ls-remote" and "git archive --remote" are designed to work
250 without being in a directory under Git's control. However, recent
251 updates revealed that we randomly look into a directory called
252 .git/ without actually doing necessary set-up when working in a
253 repository. Stop doing so.
255 * "git show-branch" expected there were only very short branch names
256 in the repository and used a fixed-length buffer to hold them
257 without checking for overflow.
259 * A caller of tempfile API that uses stdio interface to write to
260 files may ignore errors while writing, which is detected when
261 tempfile is closed (with a call to ferror()). By that time, the
262 original errno that may have told us what went wrong is likely to
263 be long gone and was overwritten by an irrelevant value.
264 close_tempfile() now resets errno to EIO to make errno at least
265 predictable.
267 * "git remote rm X", when a branch has remote X configured as the
268 value of its branch.*.remote, tried to remove branch.*.remote and
269 branch.*.merge and failed if either is unset.
271 * A "gc.log" file left by a backgrounded "gc --auto" disables further
272 automatic gc; it has been taught to run at least once a day (by
273 default) by ignoring a stale "gc.log" file that is too old.
275 * The code to parse "git -c VAR=VAL cmd" and set configuration
276 variable for the duration of cmd had two small bugs, which have
277 been fixed.
279 * user.email that consists of only cruft chars should consistently
280 error out, but didn't.
282 * "git upload-pack", which is a counter-part of "git fetch", did not
283 report a request for a ref that was not advertised as invalid.
284 This is generally not a problem (because "git fetch" will stop
285 before making such a request), but is the right thing to do.
287 * A leak in a codepath to read from a packed object in (rare) cases
288 has been plugged.
290 * When a redirected http transport gets an error during the
291 redirected request, we ignored the error we got from the server,
292 and ended up giving a not-so-useful error message.
294 * The patch subcommand of "git add -i" was meant to have paths
295 selection prompt just like other subcommand, unlike "git add -p"
296 directly jumps to hunk selection. Recently, this was broken and
297 "add -i" lost the paths selection dialog, but it now has been
298 fixed.
300 * Git v2.12 was shipped with an embarrassing breakage where various
301 operations that verify paths given from the user stopped dying when
302 seeing an issue, and instead later triggering segfault.
304 * There is no need for Python only to give a few messages to the
305 standard error stream, but we somehow did.
307 * The code to parse "git log -L..." command line was buggy when there
308 are many ranges specified with -L; overrun of the allocated buffer
309 has been fixed.
311 * The command-line parsing of "git log -L" copied internal data
312 structures using incorrect size on ILP32 systems.
314 * "git diff --quiet" relies on the size field in diff_filespec to be
315 correctly populated, but diff_populate_filespec() helper function
316 made an incorrect short-cut when asked only to populate the size
317 field for paths that need to go through convert_to_git() (e.g. CRLF
318 conversion).
320 * A few tests were run conditionally under (rare) conditions where
321 they cannot be run (like running cvs tests under 'root' account).
323 * "git branch @" created refs/heads/@ as a branch, and in general the
324 code that handled @{-1} and @{upstream} was a bit too loose in
325 disambiguating.
327 * "git fetch" that requests a commit by object name, when the other
328 side does not allow such an request, failed without much
329 explanation.
331 * "git filter-branch --prune-empty" drops a single-parent commit that
332 becomes a no-op, but did not drop a root commit whose tree is empty.
334 * Recent versions of Git treats http alternates (used in dumb http
335 transport) just like HTTP redirects and requires the client to
336 enable following it, due to security concerns. But we forgot to
337 give a warning when we decide not to honor the alternates.
339 * "git push" had a handful of codepaths that could lead to a deadlock
340 when unexpected error happened, which has been fixed.
342 * "Dumb http" transport used to misparse a nonsense http-alternates
343 response, which has been fixed.
345 * "git add -p <pathspec>" unnecessarily expanded the pathspec to a
346 list of individual files that matches the pathspec by running "git
347 ls-files <pathspec>", before feeding it to "git diff-index" to see
348 which paths have changes, because historically the pathspec
349 language supported by "diff-index" was weaker. These days they are
350 equivalent and there is no reason to internally expand it. This
351 helps both performance and avoids command line argument limit on
352 some platforms.
353 (merge 7288e12cce jk/add-i-use-pathspecs later to maint).
355 * "git status --porcelain" is supposed to give a stable output, but a
356 few strings were left as translatable by mistake.
358 * "git revert -m 0 $merge_commit" complained that reverting a merge
359 needs to say relative to which parent the reversion needs to
360 happen, as if "-m 0" weren't given. The correct diagnosis is that
361 "-m 0" does not refer to the first parent ("-m 1" does). This has
362 been fixed.
364 * Code to read submodule.<name>.ignore config did not state the
365 variable name correctly when giving an error message diagnosing
366 misconfiguration.
368 * Fix for NO_PTHREADS build.
370 * Fix for potential segv introduced in v2.11.0 and later (also
371 v2.10.2) to "git log --pickaxe-regex -S".
373 * A few unterminated here documents in tests were fixed, which in
374 turn revealed incorrect expectations the tests make. These tests
375 have been updated.
377 * Fix for NO_PTHREADS option.
378 (merge 2225e1ea20 bw/grep-recurse-submodules later to maint).
380 * Other minor doc, test and build updates and code cleanups.
381 (merge df2a6e38b7 jk/pager-in-use later to maint).
382 (merge 75ec4a6cb0 ab/branch-list-doc later to maint).
383 (merge 3e5b36c637 sg/skip-prefix-in-prettify-refname later to maint).
384 (merge 2c5e2865cc jk/fast-import-cleanup later to maint).