| blob | aa3c1ea3ea4ed66caee1d0abd766a55d4e938341 |
1 Git v1.8.2 Release Notes
2 ========================
4 Backward compatibility notes
5 ----------------------------
7 In the upcoming major release (tentatively called 1.8.2), we will
8 change the behavior of the "git push" command.
10 When "git push [$there]" does not say what to push, we have used the
11 traditional "matching" semantics so far (all your branches were sent
12 to the remote as long as there already are branches of the same name
13 over there). We will use the "simple" semantics that pushes the
14 current branch to the branch with the same name, only when the current
15 branch is set to integrate with that remote branch. There is a user
16 preference configuration variable "push.default" to change this.
18 "git push $there tag v1.2.3" used to allow replacing a tag v1.2.3
19 that already exists in the repository $there, if the rewritten tag
20 you are pushing points at a commit that is a decendant of a commit
21 that the old tag v1.2.3 points at. This was found to be error prone
22 and starting with this release, any attempt to update an existing
23 ref under refs/tags/ hierarchy will fail, without "--force".
26 Updates since v1.8.1
27 --------------------
29 UI, Workflows & Features
31 * Initial ports to QNX and z/OS UNIX System Services have started.
33 * Output from the tests is coloured using "green is okay, yellow is
34 questionable, red is bad and blue is informative" scheme.
36 * In bare repositories, "git shortlog" and other commands now read
37 mailmap files from the tip of the history, to help running these
38 tools in server settings.
40 * Color specifiers, e.g. "%C(blue)Hello%C(reset)", used in the
41 "--format=" option of "git log" and friends can be disabled when
42 the output is not sent to a terminal by prefixing them with
43 "auto,", e.g. "%C(auto,blue)Hello%C(auto,reset)".
45 * Scripts can ask Git that wildcard patterns in pathspecs they give do
46 not have any significance, i.e. take them as literal strings.
48 * The patterns in .gitignore and .gitattributes files can have **/,
49 as a pattern that matches 0 or more levels of subdirectory.
50 E.g. "foo/**/bar" matches "bar" in "foo" itself or in a
51 subdirectory of "foo".
53 * When giving arguments without "--" disambiguation, object names
54 that come earlier on the command line must not be interpretable as
55 pathspecs and pathspecs that come later on the command line must
56 not be interpretable as object names. This disambiguation rule has
57 been tweaked so that ":/" (no other string before or after) is
58 always interpreted as a pathspec; "git cmd -- :/" is no longer
59 needed, you can just say "git cmd :/".
61 * "git blame" (and "git diff") learned the "--no-follow" option.
63 * "git check-ignore" command to help debugging .gitignore files has
64 been added.
66 * "git cherry-pick" can be used to replay a root commit to an unborn
67 branch.
69 * "git commit" can be told to use --cleanup=whitespace by setting the
70 configuration variable commit.cleanup to 'whitespace'.
72 * "git fetch --mirror" and fetch that uses other forms of refspec
73 with wildcard used to attempt to update a symbolic ref that match
74 the wildcard on the receiving end, which made little sense (the
75 real ref that is pointed at by the symbolic ref would be updated
76 anyway). Symbolic refs no longer are affected by such a fetch.
78 * "git format-patch" now detects more cases in which a whole branch
79 is being exported, and uses the description for the branch, when
80 asked to write a cover letter for the series.
82 * "git format-patch" learned "-v $count" option, and prepends a
83 string "v$count-" to the names of its output files, and also
84 automatically sets the subject prefix to "PATCH v$count". This
85 allows patches from rerolled series to be stored under different
86 names and makes it easier to reuse cover letter messsages.
88 * "git log" and friends can be told with --use-mailmap option to
89 rewrite the names and email addresses of people using the mailmap
90 mechanism.
92 * "git push" now requires "-f" to update a tag, even if it is a
93 fast-forward, as tags are meant to be fixed points.
95 * "git push" will stop without doing anything if the new "pre-push"
96 hook exists and exits with a failure.
98 * When "git rebase" fails to generate patches to be applied (e.g. due
99 to oom), it failed to detect the failure and instead behaved as if
100 there were nothing to do. A workaround to use a temporary file has
101 been applied, but we probably would want to revisit this later, as
102 it hurts the common case of not failing at all.
104 * Input and preconditions to "git reset" has been loosened where
105 appropriate. "git reset $fromtree Makefile" requires $fromtree to
106 be any tree (it used to require it to be a commit), for example.
107 "git reset" (without options or parameters) used to error out when
108 you do not have any commits in your history, but it now gives you
109 an empty index (to match non-existent commit you are not even on).
111 * "git submodule" started learning a new mode to integrate with the
112 tip of the remote branch (as opposed to integrating with the commit
113 recorded in the superproject's gitlink).
116 Foreign Interface
118 * "git fast-export" has been updated for its use in the context of
119 the remote helper interface.
121 * A new remote helper to interact with bzr has been added to contrib/.
123 * "git p4" got various bugfixes around its branch handling. It is
124 also made usable with Python 2.4/2.5.
126 * The remote helper to interact with Hg in contrib/ has seen a few
127 fixes.
130 Performance, Internal Implementation, etc.
132 * "git fsck" has been taught to be pickier about entries in tree
133 objects that should not be there, e.g. ".", ".git", and "..".
135 * Matching paths with common forms of pathspecs that contain wildcard
136 characters has been optimized further.
138 * We stopped paying attention to $GIT_CONFIG environment that points
139 at a single configuration file from any command other than "git config"
140 quite a while ago, but "git clone" internally set, exported, and
141 then unexported the variable during its operation unnecessarily.
143 * "git reset" internals has been reworked and should be faster in
144 general. We tried to be careful not to break any behaviour but
145 there could be corner cases, especially when running the command
146 from a conflicted state, that we may have missed.
148 * The implementation of "imap-send" has been updated to reuse xml
149 quoting code from http-push codepath, and lost a lot of unused
150 code.
152 * There is a simple-minded checker for the test scripts in t/
153 directory to catch most common mistakes (it is not enabled by
154 default).
156 * You can build with USE_WILDMATCH=YesPlease to use a replacement
157 implementation of pattern matching logic used for pathname-like
158 things, e.g. refnames and paths in the repository. This new
159 implementation is not expected change the existing behaviour of Git
160 in this release, except for "git for-each-ref" where you can now
161 say "refs/**/master" and match with both refs/heads/master and
162 refs/remotes/origin/master. We plan to use this new implementation
163 in wider places (e.g. "git ls-files '**/Makefile' may find Makefile
164 at the top-level, and "git log '**/t*.sh'" may find commits that
165 touch a shell script whose name begins with "t" at any level) in
166 future versions of Git, but we are not there yet. By building with
167 USE_WILDMATCH, using the resulting Git daily and reporting when you
168 find breakages, you can help us get closer to that goal.
170 * Some reimplementations of Git do not write all the stat info back
171 to the index due to their implementation limitations (e.g. jgit).
172 A configuration option can tell Git to ignore changes to most of
173 the stat fields and only pay attention to mtime and size, which
174 these implementations can reliably update. This can be used to
175 avoid excessive revalidation of contents.
178 Also contains minor documentation updates and code clean-ups.
181 Fixes since v1.8.1
182 ------------------
184 Unless otherwise noted, all the fixes since v1.8.1 in the maintenance
185 track are contained in this release (see release notes to them for
186 details).
188 * An element on GIT_CEILING_DIRECTORIES list that does not name the
189 real path to a directory (i.e. a symbolic link) could have caused
190 the GIT_DIR discovery logic to escape the ceiling.
192 * When attempting to read the XDG-style $HOME/.config/git/config and
193 finding that $HOME/.config/git is a file, we gave a wrong error
194 message, instead of treating the case as "a custom config file does
195 not exist there" and moving on.
197 * The behaviour visible to the end users was confusing, when they
198 attempt to kill a process spawned in the editor that was in turn
199 launched by Git with SIGINT (or SIGQUIT), as Git would catch that
200 signal and die. We ignore these signals now.
201 (merge 1250857 pf/editor-ignore-sigint later to maint).
203 * A child process that was killed by a signal (e.g. SIGINT) was
204 reported in an inconsistent way depending on how the process was
205 spawned by us, with or without a shell in between.
207 * After failing to create a temporary file using mkstemp(), failing
208 pathname was not reported correctly on some platforms.
210 * We used to stuff "user@" and then append what we read from
211 /etc/mailname to come up with a default e-mail ident, but a bug
212 lost the "user@" part.
213 (merge dc342a2 jn/do-not-drop-username-when-reading-from-etc-mailname later to maint).
215 * The attribute mechanism didn't allow limiting attributes to be
216 applied to only a single directory itself with "path/" like the
217 exclude mechanism does. The initial implementation of this that
218 was merged to 'maint' and 1.8.1.2 was with a severe performance
219 degradations and needs to merge a fix-up topic.
220 (merge 9db9eec nd/fix-directory-attrs-off-by-one later to maint).
222 * "git am" did not parse datestamp correctly from Hg generated patch,
223 when it is run in a locale outside C (or en).
224 (merge 5185b97 dl/am-hg-locale later to maint).
226 * "git apply" misbehaved when fixing whitespace breakages by removing
227 excess trailing blank lines.
229 * A tar archive created by "git archive" recorded a directory in a
230 way that made NetBSD's implementation of "tar" sometimes unhappy.
232 * "git archive" did not record uncompressed size in the header when
233 streaming a zip archive, which confused some implementations of unzip.
235 * Attempt to "branch --edit-description" an existing branch, while
236 being on a detached HEAD, errored out.
237 (merge 75135b2 nd/edit-branch-desc-while-detached later to maint).
239 * "git clean" showed what it was going to do, but sometimes end up
240 finding that it was not allowed to do so, which resulted in a
241 confusing output (e.g. after saying that it will remove an
242 untracked directory, it found an embedded git repository there
243 which it is not allowed to remove). It now performs the actions
244 and then reports the outcome more faithfully.
245 (merge f538a91 zk/clean-report-failure later to maint).
247 * When "git clone --separate-git-dir=$over_there" is interrupted, it
248 failed to remove the real location of the $GIT_DIR it created.
249 This was most visible when interrupting a submodule update.
251 * "git fetch --depth" was broken in at least three ways. The
252 resulting history was deeper than specified by one commit, it was
253 unclear how to wipe the shallowness of the repository with the
254 command, and documentation was misleading.
255 (merge cfb70e1 nd/fetch-depth-is-broken later to maint).
257 * The way "git svn" asked for password using SSH_ASKPASS and
258 GIT_ASKPASS was not in line with the rest of the system.
260 * The --graph code fell into infinite loop when asked to do what the
261 code did not expect.
263 * http transport was wrong to ask for the username when the
264 authentication is done by certificate identity.
266 * "git pack-refs" that ran in parallel to another process that
267 created new refs had a nasty race.
269 * After "git add -N" and then writing a tree object out of the
270 index, the cache-tree data structure got corrupted.
272 * "git clone" used to allow --bare and --separate-git-dir=$there
273 options at the same time, which was nonsensical.
274 (merge 95b63f1 nd/clone-no-separate-git-dir-with-bare later to maint).
276 * "git rebase --preserve-merges" lost empty merges in recent versions
277 of Git.
278 (merge 9869778 ph/rebase-preserve-all-merges later to maint).
280 * "git merge --no-edit" computed who were involved in the work done
281 on the side branch, even though that information is to be discarded
282 without getting seen in the editor.
284 * "git merge" started calling prepare-commit-msg hook like "git
285 commit" does some time ago, but forgot to pay attention to the exit
286 status of the hook.
288 * When users spell "cc:" in lowercase in the fake "header" in the
289 trailer part, "git send-email" failed to pick up the addresses from
290 there. As e-mail headers field names are case insensitive, this
291 script should follow suit and treat "cc:" and "Cc:" the same way.
293 * Output from "git status --ignored" showed an unexpected interaction
294 with "--untracked".
296 * "gitweb", when sorting by age to show repositories with new
297 activities first, used to sort repositories with absolutely
298 nothing in it early, which was not very useful.
300 * "gitweb"'s code to sanitize control characters before passing it to
301 "highlight" filter lost known-to-be-safe control characters by
302 mistake.
304 * When a line to be wrapped has a solid run of non space characters
305 whose length exactly is the wrap width, "git shortlog -w" failed
306 to add a newline after such a line.
308 * Command line completion leaked an unnecessary error message while
309 looking for possible matches with paths in <tree-ish>.
311 * Command line completion for "tcsh" emitted an unwanted space
312 after completing a single directory name.
314 * Command line completion code was inadvertently made incompatible with
315 older versions of bash by using a newer array notation.
316 (merge 50c5885 bc/fix-array-syntax-for-3.0-in-completion-bash later to maint).
318 * Some shells do not behave correctly when IFS is unset; work it
319 around by explicitly setting it to the default value.
321 * Some scripted programs written in Python did not get updated when
322 PYTHON_PATH changed.
323 (cherry-pick 96a4647fca54031974cd6ad1 later to maint).
325 * When autoconf is used, any build on a different commit always ran
326 "config.status --recheck" even when unnecessary.
328 * We have been carrying a translated and long-unmaintained copy of an
329 old version of the tutorial; removed.
331 * t0050 had tests expecting failures from a bug that was fixed some
332 time ago.
333 (merge 336e2e2 tb/t0050-maint later to maint).
335 * t4014, t9502 and t0200 tests had various portability issues that
336 broke on OpenBSD.
338 * t9020 and t3600 tests had various portability issues.
340 * t9200 runs "cvs init" on a directory that already exists, but a
341 platform can configure this fail for the current user (e.g. you
342 need to be in the cvsadmin group on NetBSD 6.0).
344 * t9020 and t9810 had a few non-portable shell script construct.
346 * Scripts to test bash completion was inherently flaky as it was
347 affected by whatever random things the user may have on $PATH.
348 (merge 5047822 jc/do-not-let-random-file-interfere-with-completion-tests later to maint).