| blob | 0ae40f3b25a4971b93e7396abf3501cb08b92e19 |
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.
19 Updates since v1.8.1
20 --------------------
22 UI, Workflows & Features
24 * Initial ports to QNX and z/OS UNIX System Services have started.
26 * Output from the tests is coloured using "green is okay, yellow is
27 questionable, red is bad and blue is informative" scheme.
29 * In bare repositories, "git shortlog" and other commands now read
30 mailmap files from the tip of the history, to help running these
31 tools in server settings.
33 * Color specifiers, e.g. "%C(blue)Hello%C(reset)", used in the
34 "--format=" option of "git log" and friends can be disabled when
35 the output is not sent to a terminal by prefixing them with
36 "auto,", e.g. "%C(auto,blue)Hello%C(auto,reset)".
38 * Scripts can ask Git that wildcard patterns in pathspecs they give do
39 not have any significance, i.e. take them as literal strings.
41 * The pathspec code learned to grok "foo/**/bar" as a pattern that
42 matches "bar" in 0-or-more levels of subdirectory in "foo".
44 * "git blame" (and "git diff") learned the "--no-follow" option.
46 * "git cherry-pick" can be used to replay a root commit to an unborn
47 branch.
49 * "git commit" can be told to use --cleanup=whitespace by setting the
50 configuration variable commit.cleanup to 'whitespace'.
52 * "git fetch --mirror" and fetch that uses other forms of refspec
53 with wildcard used to attempt to update a symbolic ref that match
54 the wildcard on the receiving end, which made little sense (the
55 real ref that is pointed at by the symbolic ref would be updated
56 anyway). Symbolic refs no longer are affected by such a fetch.
58 * "git format-patch" now detects more cases in which a whole branch
59 is being exported, and uses the description for the branch, when
60 asked to write a cover letter for the series.
62 * "git format-patch" learned "-v $count" option, and prepends a
63 string "v$count-" to the names of its output files, and also
64 automatically sets the subject prefix to "PATCH v$count". This
65 allows patches from rerolled series to be stored under different
66 names and makes it easier to reuse cover letter messsages.
68 * "git log" and friends can be told with --use-mailmap option to
69 rewrite the names and email addresses of people using the mailmap
70 mechanism.
72 * "git push" now requires "-f" to update a tag, even if it is a
73 fast-forward, as tags are meant to be fixed points.
75 * When "git rebase" fails to generate patches to be applied (e.g. due
76 to oom), it failed to detect the failure and instead behaved as if
77 there were nothing to do. A workaround to use a temporary file has
78 been applied, but we probably would want to revisit this later, as
79 it hurts the common case of not failing at all.
81 * Input and preconditions to "git reset" has been loosened where
82 appropriate. "git reset $fromtree Makefile" requires $fromtree to
83 be any tree (it used to require it to be a commit), for example.
84 "git reset" (without options or parameters) used to error out when
85 you do not have any commits in your history, but it now gives you
86 an empty index (to match non-existent commit you are not even on).
88 * "git submodule" started learning a new mode to integrate with the
89 tip of the remote branch (as opposed to integrating with the commit
90 recorded in the superproject's gitlink).
93 Foreign Interface
95 * "git fast-export" has been updated for its use in the context of
96 the remote helper interface.
98 * A new remote helper to interact with bzr has been added to contrib/.
100 * "git p4" got various bugfixes around its branch handling.
102 * The remote helper to interact with Hg in contrib/ has seen a few
103 fixes.
106 Performance, Internal Implementation, etc.
108 * "git fsck" has been taught to be pickier about entries in tree
109 objects that should not be there, e.g. ".", ".git", and "..".
111 * Matching paths with common forms of pathspecs that contain wildcard
112 characters has been optimized further.
114 * "git reset" internals has been reworked and should be faster in
115 general. We tried to be careful not to break any behaviour but
116 there could be corner cases, especially when running the command
117 from a conflicted state, that we may have missed.
119 * The implementation of "imap-send" has been updated to reuse xml
120 quoting code from http-push codepath.
122 * There is a simple-minded checker for the test scripts in t/
123 directory to catch most common mistakes (it is not enabled by
124 default).
127 Also contains minor documentation updates and code clean-ups.
130 Fixes since v1.8.1
131 ------------------
133 Unless otherwise noted, all the fixes since v1.8.1 in the maintenance
134 track are contained in this release (see release notes to them for
135 details).
137 * An element on GIT_CEILING_DIRECTORIES list that does not name the
138 real path to a directory (i.e. a symbolic link) could have caused
139 the GIT_DIR discovery logic to escape the ceiling.
140 (merge 059b379 mh/ceiling later to maint).
142 * When attempting to read the XDG-style $HOME/.config/git/config and
143 finding that $HOME/.config/git is a file, we gave a wrong error
144 message, instead of treating the case as "a custom config file does
145 not exist there" and moving on.
146 (merge 8f2bbe4 jn/warn-on-inaccessible-loosen later to maint).
148 * The behaviour visible to the end users was confusing, when they
149 attempt to kill a process spawned in the editor that was in turn
150 launched by Git with SIGINT (or SIGQUIT), as Git would catch that
151 signal and die. We ignore these signals now.
152 (merge 1250857 pf/editor-ignore-sigint later to maint).
154 * A child process that was killed by a signal (e.g. SIGINT) was
155 reported in an inconsistent way depending on how the process was
156 spawned by us, with or without a shell in between.
157 (merge 709ca73 jk/unify-exit-code-by-receiving-signal later to maint).
159 * After failing to create a temporary file using mkstemp(), failing
160 pathname was not reported correctly on some platforms.
161 (merge f7be59b jc/mkstemp-more-careful-error-reporting later to maint).
163 * The attribute mechanism didn't allow limiting attributes to be
164 applied to only a single directory itself with "path/" like the
165 exclude mechanism does. The initial implementation of this that
166 was merged to 'maint' and 1.8.1.2 was with a severe performance
167 degradations and needs to merge a fix-up topic.
168 (merge 9db9eec nd/fix-directory-attrs-off-by-one later to maint).
170 * "git apply" misbehaved when fixing whitespace breakages by removing
171 excess trailing blank lines.
172 (merge 5de7166 jc/apply-trailing-blank-removal later to maint).
174 * A tar archive created by "git archive" recorded a directory in a
175 way that made NetBSD's implementation of "tar" sometimes unhappy.
176 (merge 22f0dcd rs/leave-base-name-in-name-field-of-tar later to maint).
178 * "git archive" did not record uncompressed size in the header when
179 streaming a zip archive, which confused some implementations of unzip.
180 (merge 5ea2c84 rs/zip-with-uncompressed-size-in-the-header later to maint).
182 * "git clean" showed what it was going to do, but sometimes end up
183 finding that it was not allowed to do so, which resulted in a
184 confusing output (e.g. after saying that it will remove an
185 untracked directory, it found an embedded git repository there
186 which it is not allowed to remove). It now performs the actions
187 and then reports the outcome more faithfully.
188 (merge f538a91 zk/clean-report-failure later to maint).
190 * When "git clone --separate-git-dir=$over_there" is interrupted, it
191 failed to remove the real location of the $GIT_DIR it created.
192 This was most visible when interrupting a submodule update.
193 (merge 9be1980 jl/interrupt-clone-remove-separate-git-dir later to maint).
195 * The way "git svn" asked for password using SSH_ASKPASS and
196 GIT_ASKPASS was not in line with the rest of the system.
197 (merge e9263e4 ss/svn-prompt later to maint).
199 * The --graph code fell into infinite loop when asked to do what the
200 code did not expect.
201 (merge 656197a mk/maint-graph-infinity-loop later to maint).
203 * http transport was wrong to ask for the username when the
204 authentication is done by certificate identity.
205 (merge 75e9a40 rb/http-cert-cred-no-username-prompt later to maint).
207 * "git pack-refs" that ran in parallel to another process that
208 created new refs had a nasty race.
209 (merge b3f1280 jk/repack-ref-racefix later to maint).
211 * After "git add -N" and then writing a tree object out of the
212 index, the cache-tree data structure got corrupted.
213 (merge eec3e7e nd/invalidate-i-t-a-cache-tree later to maint).
215 * "git clone" used to allow --bare and --separate-git-dir=$there
216 options at the same time, which was nonsensical.
217 (merge 95b63f1 nd/clone-no-separate-git-dir-with-bare later to maint).
219 * "git rebase --preserve-merges" lost empty merges in recent versions
220 of Git.
221 (merge 9869778 ph/rebase-preserve-all-merges later to maint).
223 * "git merge --no-edit" computed who were involved in the work done
224 on the side branch, even though that information is to be discarded
225 without getting seen in the editor.
226 (merge 9bcbb1c jc/maint-fmt-merge-msg-no-edit-lose-credit later to maint).
228 * "git merge" started calling prepare-commit-msg hook like "git
229 commit" does some time ago, but forgot to pay attention to the exit
230 status of the hook.
231 (merge 3e4141d ap/merge-stop-at-prepare-commit-msg-failure later to maint).
233 * When users spell "cc:" in lowercase in the fake "header" in the
234 trailer part, "git send-email" failed to pick up the addresses from
235 there. As e-mail headers field names are case insensitive, this
236 script should follow suit and treat "cc:" and "Cc:" the same way.
237 (merge 6310071 nz/send-email-headers-are-case-insensitive later to maint).
239 * Output from "git status --ignored" showed an unexpected interaction
240 with "--untracked".
241 (merge a45fb69 ap/status-ignored-in-ignored-directory later to maint).
243 * "gitweb", when sorting by age to show repositories with new
244 activities first, used to sort repositories with absolutely
245 nothing in it early, which was not very useful.
246 (merge 28dae18 md/gitweb-sort-by-age later to maint).
248 * "gitweb"'s code to sanitize control characters before passing it to
249 "highlight" filter lost known-to-be-safe control characters by
250 mistake.
251 (merge 0e901d2 os/gitweb-highlight-uncaptured later to maint).
253 * When a line to be wrapped has a solid run of non space characters
254 whose length exactly is the wrap width, "git shortlog -w" failed
255 to add a newline after such a line.
256 (merge e0db176 sp/shortlog-missing-lf later to maint).
258 * Command line completion leaked an unnecessary error message while
259 looking for possible matches with paths in <tree-ish>.
260 (merge ca87dd6 ds/completion-silence-in-tree-path-probe later to maint).
262 * Command line completion for "tcsh" emitted an unwanted space
263 after completing a single directory name.
264 (merge 92f1c04 mk/complete-tcsh later to maint).
266 * Some shells do not behave correctly when IFS is unset; work it
267 around by explicitly setting it to the default value.
268 (merge 393050c jc/maint-fbsd-sh-ifs-workaround later to maint).
270 * Some scripted programs written in Python did not get updated when
271 PYTHON_PATH changed.
272 (cherry-pick 96a4647fca54031974cd6ad1 later to maint).
274 * When autoconf is used, any build on a different commit always ran
275 "config.status --recheck" even when unnecessary.
276 (merge 1226504 jn/less-reconfigure later to maint).
278 * We have been carrying a translated and long-unmaintained copy of an
279 old version of the tutorial; removed.
280 (merge 0a85441 ta/remove-stale-translated-tut later to maint).
282 * t4014, t9502 and t0200 tests had various portability issues that
283 broke on OpenBSD.
284 (merge 27f6342 jc/maint-test-portability later to maint).
286 * t9020 and t3600 tests had various portability issues.
287 (merge 5a02966 jc/test-portability later to maint).
289 * t9200 runs "cvs init" on a directory that already exists, but a
290 platform can configure this fail for the current user (e.g. you
291 need to be in the cvsadmin group on NetBSD 6.0).
292 (merge 8666df0 jc/test-cvs-no-init-in-existing-dir later to maint).
294 * t9020 and t9810 had a few non-portable shell script construct.
295 (merge 2797914 tb/test-t9020-no-which later to maint).
296 (merge 6f4e505 tb/test-t9810-no-sed-i later to maint).