| blob | efff5264d883ad14ab036dbc7cc04f8d48e8d71e |
1 Git 2.12 Release Notes
2 ======================
4 Backward compatibility notes.
6 * Use of an empty string that is used for 'everything matches' is
7 still warned and Git asks users to use a more explicit '.' for that
8 instead. The hope is that existing users will not mind this
9 change, and eventually the warning can be turned into a hard error,
10 upgrading the deprecation into removal of this (mis)feature. That
11 is not scheduled to happen in the upcoming release (yet).
13 * The historical argument order "git merge <msg> HEAD <commit>..."
14 has been deprecated for quite some time, and will be removed in the
15 upcoming release.
18 Updates since v2.11
19 -------------------
21 UI, Workflows & Features
23 * Various updates to "git p4".
25 * "git p4" didn't interact with the internal of .git directory
26 correctly in the modern "git-worktree"-enabled world.
28 * "git branch --list" and friends learned "--ignore-case" option to
29 optionally sort branches and tags case insensitively.
31 * In addition to %(subject), %(body), "log --pretty=format:..."
32 learned a new placeholder %(trailers).
34 * "git rebase" learned "--quit" option, which allows a user to
35 remove the metadata left by an earlier "git rebase" that was
36 manually aborted without using "git rebase --abort".
38 * "git clone --reference $there --recurse-submodules $super" has been
39 taught to guess repositories usable as references for submodules of
40 $super that are embedded in $there while making a clone of the
41 superproject borrow objects from $there; extend the mechanism to
42 also allow submodules of these submodules to borrow repositories
43 embedded in these clones of the submodules embedded in the clone of
44 the superproject.
46 * Porcelain scripts written in Perl are getting internationalized.
48 * "git merge --continue" has been added as a synonym to "git commit"
49 to conclude a merge that has stopped due to conflicts.
51 * Finer-grained control of what protocols are allowed for transports
52 during clone/fetch/push have been enabled via a new configuration
53 mechanism.
55 * "git shortlog" learned "--committer" option to group commits by
56 committer, instead of author.
58 * GitLFS integration with "git p4" has been updated.
60 * The isatty() emulation for Windows has been updated to eradicate
61 the previous hack that depended on internals of (older) MSVC
62 runtime.
64 * Some platforms no longer understand "latin-1" that is still seen in
65 the wild in e-mail headers; replace them with "iso-8859-1" that is
66 more widely known when conversion fails from/to it.
67 (merge df3755888b jc/latin-1 later to maint).
69 * "git grep" has been taught to optionally recurse into submodules.
71 * "git rm" used to refuse to remove a submodule when it has its own
72 git repository embedded in its working tree. It learned to move
73 the repository away to $GIT_DIR/modules/ of the superproject
74 instead, and allow the submodule to be deleted (as long as there
75 will be no loss of local modifications, that is).
77 * A recent updates to "git p4" was not usable for older p4 but it
78 could be made to work with minimum changes. Do so.
81 Performance, Internal Implementation, Development Support etc.
83 * Commands that operate on a log message and add lines to the trailer
84 blocks, such as "format-patch -s", "cherry-pick (-x|-s)", and
85 "commit -s", have been taught to use the logic of and share the
86 code with "git interpret-trailer".
88 * The default Travis-CI configuration specifies newer P4 and GitLFS.
90 * The "fast hash" that had disastrous performance issues in some
91 corner cases has been retired from the internal diff.
93 * The character width table has been updated to match Unicode 9.0
95 * Update the procedure to generate "tags" for developer support.
96 (merge 046e4c1c09 jk/make-tags-find-sources-tweak later to maint).
98 * The codeflow of setting NOATIME and CLOEXEC on file descriptors Git
99 opens has been simplified.
100 (merge b4d065df03 jc/git-open-cloexec later to maint).
102 * "git diff" and its family had two experimental heuristics to shift
103 the contents of a hunk to make the patch easier to read. One of
104 them turns out to be better than the other, so leave only the
105 "--indent-heuristic" option and remove the other one.
106 (merge 3cde4e02ee jc/retire-compaction-heuristics later to maint).
108 * A new submodule helper "git submodule embedgitdirs" to make it
109 easier to move embedded .git/ directory for submodules in a
110 superproject to .git/modules/ (and point the latter with the former
111 that is turned into a "gitdir:" file) has been added.
113 * "git push \\server\share\dir" has recently regressed and then
114 fixed. A test has retroactively been added for this breakage.
116 * Build updates for Cygwin.
118 * The implementation of "real_path()" was to go there with chdir(2)
119 and call getcwd(3), but this obviously wouldn't be usable in a
120 threaded environment. Rewrite it to manually resolve relative
121 paths including symbolic links in path components.
123 * Adjust documentation to help AsciiDoctor render better while not
124 breaking the rendering done by AsciiDoc.
127 Also contains various documentation updates and code clean-ups.
129 Fixes since v2.10
130 -----------------
132 Unless otherwise noted, all the fixes since v2.9 in the maintenance
133 track are contained in this release (see the maintenance releases'
134 notes for details).
136 * We often decide if a session is interactive by checking if the
137 standard I/O streams are connected to a TTY, but isatty() that
138 comes with Windows incorrectly returned true if it is used on NUL
139 (i.e. an equivalent to /dev/null). This has been fixed.
141 * "git svn" did not work well with path components that are "0", and
142 some configuration variable it uses were not documented.
143 (merge ea9a93dcc2 ew/svn-fixes later to maint).
145 * "git rev-parse --symbolic" failed with a more recent notation like
146 "HEAD^-1" and "HEAD^!".
148 * An empty directory in a working tree that can simply be nuked used
149 to interfere while merging or cherry-picking a change to create a
150 submodule directory there, which has been fixed..
152 * The code in "git push" to compute if any commit being pushed in the
153 superproject binds a commit in a submodule that hasn't been pushed
154 out was overly inefficient, making it unusable even for a small
155 project that does not have any submodule but have a reasonable
156 number of refs.
158 * "git push --dry-run --recurse-submodule=on-demand" wasn't
159 "--dry-run" in the submodules.
161 * The output from "git worktree list" was made in readdir() order,
162 and was unstable.
164 * mergetool.<tool>.trustExitCode configuration variable did not apply
165 to built-in tools, but now it does.
167 * "git p4" LFS support was broken when LFS stores an empty blob.
169 * A corner case in merge-recursive regression that crept in
170 during 2.10 development cycle has been fixed.
172 * Transport with dumb http can be fooled into following foreign URLs
173 that the end user does not intend to, especially with the server
174 side redirects and http-alternates mechanism, which can lead to
175 security issues. Tighten the redirection and make it more obvious
176 to the end user when it happens.
178 * Update the error messages from the dumb-http client when it fails
179 to obtain loose objects; we used to give sensible error message
180 only upon 404 but we now forbid unexpected redirects that needs to
181 be reported with something sensible.
183 * When diff.renames configuration is on (and with Git 2.9 and later,
184 it is enabled by default, which made it worse), "git stash"
185 misbehaved if a file is removed and another file with a very
186 similar content is added.
188 * "git diff --no-index" did not take "--no-abbrev" option.
190 * "git difftool --dir-diff" had a minor regression when started from
191 a subdirectory, which has been fixed.
193 * "git commit --allow-empty --only" (no pathspec) with dirty index
194 ought to be an acceptable way to create a new commit that does not
195 change any paths, but it was forbidden, perhaps because nobody
196 needed it so far.
198 * Git 2.11 had a minor regression in "merge --ff-only" that competed
199 with another process that simultanously attempted to update the
200 index. We used to explain what went wrong with an error message,
201 but the new code silently failed. The error message has been
202 resurrected.
204 * A pathname that begins with "//" or "\\" on Windows is special but
205 path normalization logic was unaware of it.
207 * "git pull --rebase", when there is no new commits on our side since
208 we forked from the upstream, should be able to fast-forward without
209 invoking "git rebase", but it didn't.
211 * The way to specify hotkeys to "xxdiff" that is used by "git
212 mergetool" has been modernized to match recent versions of xxdiff.
214 * Unlike "git am --abort", "git cherry-pick --abort" moved HEAD back
215 to where cherry-pick started while picking multiple changes, when
216 the cherry-pick stopped to ask for help from the user, and the user
217 did "git reset --hard" to a different commit in order to re-attempt
218 the operation.
220 * Code cleanup in shallow boundary computation.
222 * A recent update to receive-pack to make it easier to drop garbage
223 objects made it clear that GIT_ALTERNATE_OBJECT_DIRECTORIES cannot
224 have a pathname with a colon in it (no surprise!), and this in turn
225 made it impossible to push into a repository at such a path. This
226 has been fixed by introducing a quoting mechanism used when
227 appending such a path to the colon-separated list.
229 * The function usage_msg_opt() has been updated to say "fatal:"
230 before the custom message programs give, when they want to die
231 with a message about wrong command line options followed by the
232 standard usage string.
234 * "git index-pack --stdin" needs an access to an existing repository,
235 but "git index-pack file.pack" to generate an .idx file that
236 corresponds to a packfile does not.
238 * Fix for NDEBUG builds.
240 * A lazy "git push" without refspec did not internally use a fully
241 specified refspec to perform 'current', 'simple', or 'upstream'
242 push, causing unnecessary "ambiguous ref" errors.
244 * "git p4" misbehaved when swapping a directory and a symbolic link.
246 * Even though an fix was attempted in Git 2.9.3 days, but running
247 "git difftool --dir-diff" from a subdirectory never worked. This
248 has been fixed.
250 * "git p4" that tracks multile p4 paths imported a single changelist
251 that touches files in these multiple paths as one commit, followed
252 by many empty commits. This has been fixed.
254 * A potential but unlikely buffer overflow in Windows port has been
255 fixed.
257 * When the http server gives an incomplete response to a smart-http
258 rpc call, it could lead to client waiting for a full response that
259 will never come. Teach the client side to notice this condition
260 and abort the transfer.
262 * Compression setting for producing packfiles were spread across
263 three codepaths, one of which did not honor any configuration.
264 Unify these so that all of them honor core.compression and
265 pack.compression variables the same way.
266 (merge 8de7eeb54b jc/compression-config later to maint).
268 * "git fast-import" sometimes mishandled while rebalancing notes
269 tree, which has been fixed.
270 (merge 405d7f4af6 mh/fast-import-notes-fix-new later to maint).
272 * Recent update to the default abbreviation length that auto-scales
273 lacked documentation update, which has been corrected.
274 (merge 48d5014dd4 jc/abbrev-autoscale-config later to maint).
276 * Leakage of lockfiles in the config subsystem has been fixed.
277 (merge c06fa62dfc nd/config-misc-fixes later to maint).
279 * It is natural that "git gc --auto" may not attempt to pack
280 everything into a single pack, and there is no point in warning
281 when the user has configured the system to use the pack bitmap,
282 leading to disabling further "gc".
283 (merge 1c409a705c dt/disable-bitmap-in-auto-gc later to maint).
285 * "git archive" did not read the standard configuration files, and
286 failed to notice a file that is marked as binary via the userdiff
287 driver configuration.
288 (merge 965cba2e7e jk/archive-zip-userdiff-config later to maint).
290 * "git blame --porcelain" misidentified the "previous" <commit, path>
291 pair (aka "source") when contents came from two or more files.
292 (merge 4e76832984 jk/blame-fixes later to maint).
294 * "git rebase -i" with a recent update started showing an incorrect
295 count when squashing more than 10 commits.
296 (merge 356b8ecff1 jk/rebase-i-squash-count-fix later to maint).
298 * "git <cmd> @{push}" on a detached HEAD used to segfault; it has
299 been corrected to error out with a message.
300 (merge b10731f43d km/branch-get-push-while-detached later to maint).
302 * Running "git add a/b" when "a" is a submodule correctly errored
303 out, but without a meaningful error message.
304 (merge 2d81c48fa7 sb/pathspec-errors later to maint).
306 * Typing ^C to pager, which usually does not kill it, killed Git and
307 took the pager down as a collateral damage in certain process-tree
308 structure. This has been fixed.
309 (merge 46df6906f3 jk/execv-dashed-external later to maint).
311 * "git mergetool" without any pathspec on the command line that is
312 run from a subdirectory became no-op in Git v2.11 by mistake, which
313 has been fixed.
315 * Retire long unused/unmaintained gitview from the contrib/ area.
316 (merge 3120925c25 sb/remove-gitview later to maint).
318 * Other minor doc, test and build updates and code cleanups.
319 (merge f2627d9b19 sb/submodule-config-cleanup later to maint).
320 (merge 384f1a167b sb/unpack-trees-cleanup later to maint).