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
14 * The historical argument order "git merge <msg> HEAD <commit>..."
15 has been deprecated for quite some time, and will be removed in a
22 UI, Workflows & Features
24 * "git describe" and "git name-rev" have been taught to take more
25 than one refname patterns to restrict the set of refs to base their
26 naming output on, and also learned to take negative patterns to
27 name refs not to be used for naming via their "--exclude" option.
29 * Deletion of a branch "foo/bar" could remove .git/refs/heads/foo
30 once there no longer is any other branch whose name begins with
31 "foo/", but we didn't do so so far. Now we do.
33 * When "git merge" detects a path that is renamed in one history
34 while the other history deleted (or modified) it, it now reports
35 both paths to help the user understand what is going on in the two
36 histories being merged.
38 * The <url> part in "http.<url>.<variable>" configuration variable
39 can now be spelled with '*' that serves as wildcard.
40 E.g. "http.https://*.example.com.proxy" can be used to specify the
41 proxy used for https://a.example.com, https://b.example.com, etc.,
42 i.e. any host in the example.com domain.
44 * "git tag" did not leave useful message when adding a new entry to
45 reflog; this was left unnoticed for a long time because refs/tags/*
46 doesn't keep reflog by default.
48 * The "negative" pathspec feature was somewhat more cumbersome to use
49 than necessary in that its short-hand used "!" which needed to be
50 escaped from shells, and it required "exclude from what?" specified.
52 * The command line options for ssh invocation needs to be tweaked for
53 some implementations of SSH (e.g. PuTTY plink wants "-P <port>"
54 while OpenSSH wants "-p <port>" to specify port to connect to), and
55 the variant was guessed when GIT_SSH environment variable is used
56 to specify it. The logic to guess now applies to the command
57 specified by the newer GIT_SSH_COMMAND and also core.sshcommand
58 configuration variable, and comes with an escape hatch for users to
59 deal with misdetected cases.
61 * The "--git-path", "--git-common-dir", and "--shared-index-path"
62 options of "git rev-parse" did not produce usable output. They are
63 now updated to show the path to the correct file, relative to where
66 * "git diff -W" has been taught to handle the case where a new
67 function is added at the end of the file better.
69 * "git update-ref -d" and other operations to delete references did
70 not leave any entry in HEAD's reflog when the reference being
71 deleted was the current branch. This is not a problem in practice
72 because you do not want to delete the branch you are currently on,
73 but caused renaming of the current branch to something else not to
74 be logged in a useful way.
76 * "Cc:" on the trailer part does not have to conform to RFC strictly,
77 unlike in the e-mail header. "git send-email" has been updated to
78 ignore anything after '>' when picking addresses, to allow non-address
79 cruft like " # stable 4.4" after the address.
81 * When "git submodule init" decides that the submodule in the working
82 tree is its upstream, it now gives a warning as it is not a very
84 (merge d1b3b81aab sb/submodule-init-url-selection later to maint).
86 * "git stash save" takes a pathspec so that the local changes can be
87 stashed away only partially.
88 (merge 9e140909f6 tg/stash-push later to maint).
90 * Documentation for "git ls-files" did not refer to core.quotePath.
92 * The experimental "split index" feature has gained a few
93 configuration variables to make it easier to use.
95 * From a working tree of a repository, a new option of "rev-parse"
96 lets you ask if the repository is used as a submodule of another
97 project, and where the root level of the working tree of that
98 project (i.e. your superproject) is.
100 * The pathspec mechanism learned to further limit the paths that
101 match the pattern to those that have specified attributes attached
102 via the gitattributes mechanism.
104 * Our source code has used the SHA1_HEADER cpp macro after "#include"
105 in the C code to switch among the SHA-1 implementations. Instead,
106 list the exact header file names and switch among implementations
107 using "#ifdef BLK_SHA1/#include "block-sha1/sha1.h"/.../#endif";
108 this helps some IDE tools.
110 * The start-up sequence of "git" needs to figure out some configured
111 settings before it finds and set itself up in the location of the
112 repository and was quite messy due to its "chicken-and-egg" nature.
113 The code has been restructured.
115 * The command line prompt (in contrib/) learned a new 'tag' style
116 that can be specified with GIT_PS1_DESCRIBE_STYLE, to describe a
117 detached HEAD with "git describe --tags".
119 * The configuration file learned a new "includeIf.<condition>.path"
120 that includes the contents of the given path only when the
121 condition holds. This allows you to say "include this work-related
122 bit only in the repositories under my ~/work/ directory".
124 * Recent update to "rebase -i" started showing a message that is not
125 a warning with "warning:" prefix by mistake. This has been fixed.
128 Performance, Internal Implementation, Development Support etc.
130 * The code to list branches in "git branch" has been consolidated
131 with the more generic ref-filter API.
133 * Resource usage while enumerating refs from alternate object store
134 has been optimized to help receiving end of "push" that hosts a
135 repository with many "forks".
137 * The gitattributes machinery is being taught to work better in a
138 multi-threaded environment.
140 * "git rebase -i" starts using the recently updated "sequencer" code.
142 * Code and design clean-up for the refs API.
144 * The preload-index code has been taught not to bother with the index
145 entries that are paths that are not checked out by "sparse checkout".
147 * Some warning() messages from "git clean" were updated to show the
148 errno from failed system calls.
150 * The "parse_config_key()" API function has been cleaned up.
151 (merge ad8c7cdadd jk/parse-config-key-cleanup later to maint).
153 * A test that creates a confusing branch whose name is HEAD has been
154 corrected not to do so.
156 * The code that parses header fields in the commit object has been
157 updated for (micro)performance and code hygiene.
159 * An helper function to make it easier to append the result from
160 real_path() to a strbuf has been added.
161 (merge 33ad9ddd0b rs/strbuf-add-real-path later to maint).
163 * Reduce authentication round-trip over HTTP when the server supports
164 just a single authentication method. This also improves the
165 behaviour when Git is misconfigured to enable http.emptyAuth
166 against a server that does not authenticate without a username
167 (i.e. not using Kerberos etc., which makes http.emptyAuth
170 * Windows port wants to use OpenSSL's implementation of SHA-1
171 routines, so let them.
173 * The t/perf performance test suite was not prepared to test not so
174 old versions of Git, but now it covers versions of Git that are not
176 (merge 28e1fb5466 jt/perf-updates later to maint).
178 * Add 32-bit Linux variant to the set of platforms to be tested with
181 * "git branch --list" takes the "--abbrev" and "--no-abbrev" options
182 to control the output of the object name in its "-v"(erbose)
183 output, but a recent update started ignoring them; fix it before
184 the breakage reaches to any released version.
186 * Picking two versions of Git and running tests to make sure the
187 older one and the newer one interoperate happily has now become
189 (merge bd4d9d993c jk/interop-test later to maint).
191 * "uchar [40]" to "struct object_id" conversion continues.
193 * "git tag --contains" used to (ab)use the object bits to keep track
194 of the state of object reachability without clearing them after
195 use; this has been cleaned up and made to use the newer commit-slab
199 Also contains various documentation updates and code clean-ups.
205 Unless otherwise noted, all the fixes since v2.9 in the maintenance
206 track are contained in this release (see the maintenance releases'
209 * "git repack --depth=<n>" for a long time busted the specified depth
210 when reusing delta from existing packs. This has been corrected.
211 (merge 42b766d765 jk/delta-chain-limit later to maint).
213 * The code to parse the command line "git grep <patterns>... <rev>
214 [[--] <pathspec>...]" has been cleaned up, and a handful of bugs
215 have been fixed (e.g. we used to check "--" if it is a rev).
216 (merge 131f3c96d2 jk/grep-no-index-fix later to maint).
218 * "git ls-remote" and "git archive --remote" are designed to work
219 without being in a directory under Git's control. However, recent
220 updates revealed that we randomly look into a directory called
221 .git/ without actually doing necessary set-up when working in a
222 repository. Stop doing so.
223 (merge 4b0c3c7735 jn/remote-helpers-with-git-dir later to maint).
225 * "git show-branch" expected there were only very short branch names
226 in the repository and used a fixed-length buffer to hold them
227 without checking for overflow.
229 * A caller of tempfile API that uses stdio interface to write to
230 files may ignore errors while writing, which is detected when
231 tempfile is closed (with a call to ferror()). By that time, the
232 original errno that may have told us what went wrong is likely to
233 be long gone and was overwritten by an irrelevant value.
234 close_tempfile() now resets errno to EIO to make errno at least
237 * "git remote rm X", when a branch has remote X configured as the
238 value of its branch.*.remote, tried to remove branch.*.remote and
239 branch.*.merge and failed if either is unset.
241 * A "gc.log" file left by a backgrounded "gc --auto" disables further
242 automatic gc; it has been taught to run at least once a day (by
243 default) by ignoring a stale "gc.log" file that is too old.
245 * The code to parse "git -c VAR=VAL cmd" and set configuration
246 variable for the duration of cmd had two small bugs, which have
248 (merge 1274a155af jc/config-case-cmdline-take-2 later to maint).
250 * user.email that consists of only cruft chars should consistently
251 error out, but didn't.
252 (merge 94425552f3 jk/ident-empty later to maint).
254 * "git upload-pack", which is a counter-part of "git fetch", did not
255 report a request for a ref that was not advertised as invalid.
256 This is generally not a problem (because "git fetch" will stop
257 before making such a request), but is the right thing to do.
259 * A leak in a codepath to read from a packed object in (rare) cases
262 * When a redirected http transport gets an error during the
263 redirected request, we ignored the error we got from the server,
264 and ended up giving a not-so-useful error message.
266 * The patch subcommand of "git add -i" was meant to have paths
267 selection prompt just like other subcommand, unlike "git add -p"
268 directly jumps to hunk selection. Recently, this was broken and
269 "add -i" lost the paths selection dialog, but it now has been
272 * Git v2.12 was shipped with an embarrassing breakage where various
273 operations that verify paths given from the user stopped dying when
274 seeing an issue, and instead later triggering segfault.
276 * There is no need for Python only to give a few messages to the
277 standard error stream, but we somehow did.
279 * The code to parse "git log -L..." command line was buggy when there
280 are many ranges specified with -L; overrun of the allocated buffer
283 * The command-line parsing of "git log -L" copied internal data
284 structures using incorrect size on ILP32 systems.
286 * "git diff --quiet" relies on the size field in diff_filespec to be
287 correctly populated, but diff_populate_filespec() helper function
288 made an incorrect short-cut when asked only to populate the size
289 field for paths that need to go through convert_to_git() (e.g. CRLF
292 * A few tests were run conditionally under (rare) conditions where
293 they cannot be run (like running cvs tests under 'root' account).
294 (merge c6507484a2 ab/cond-skip-tests later to maint).
296 * "git branch @" created refs/heads/@ as a branch, and in general the
297 code that handled @{-1} and @{upstream} was a bit too loose in
299 (merge fd4692ff70 jk/interpret-branch-name later to maint).
301 * "git fetch" that requests a commit by object name, when the other
302 side does not allow such an request, failed without much
304 (merge d56583ded6 mm/fetch-show-error-message-on-unadvertised-object later to maint).
306 * "git filter-branch --prune-empty" drops a single-parent commit that
307 becomes a no-op, but did not drop a root commit whose tree is empty.
308 (merge 32da7467eb dp/filter-branch-prune-empty later to maint).
310 * Recent versions of Git treats http alternates (used in dumb http
311 transport) just like HTTP redirects and requires the client to
312 enable following it, due to security concerns. But we forgot to
313 give a warning when we decide not to honor the alternates.
314 (merge 5cae73d5d2 ew/http-alternates-as-redirects-warning later to maint).
316 * "git push" had a handful of codepaths that could lead to a deadlock
317 when unexpected error happened, which has been fixed.
318 (merge d1a13d3fcb jk/push-deadlock-regression-fix later to maint).
320 * "Dumb http" transport used to misparse a nonsense http-alternates
321 response, which has been fixed.
323 * "git add -p <pathspec>" unnecessarily expanded the pathspec to a
324 list of individual files that matches the pathspec by running "git
325 ls-files <pathspec>", before feeding it to "git diff-index" to see
326 which paths have changes, because historically the pathspec
327 language supported by "diff-index" was weaker. These days they are
328 equivalent and there is no reason to internally expand it. This
329 helps both performance and avoids command line argument limit on
331 (merge 7288e12cce jk/add-i-use-pathspecs later to maint).
333 * "git status --porcelain" is supposed to give a stable output, but a
334 few strings were left as translatable by mistake.
336 * Code to read submodule.<name>.ignore config did not state the
337 variable name correctly when giving an error message diagnosing
339 (merge 5ea304896e sb/submodule-config-parse-ignore-fix later to maint).
341 * Other minor doc, test and build updates and code cleanups.
342 (merge dfa3ad3238 rs/blame-code-cleanup later to maint).
343 (merge ffddfc6328 jk/rev-parse-cleanup later to maint).
344 (merge f20754802a jk/pack-name-cleanups later to maint).
345 (merge d4aae459cd sb/wt-status-cleanup later to maint).
346 (merge 2c7ee986c7 ab/doc-no-option-notation-fix later to maint).
347 (merge e4e016f65d ab/push-default-doc-fix later to maint).
348 (merge baced9e4e5 nd/commit-hook-doc-fix later to maint).
projects / git / blob