| blob | 9158899d60a394e998c8f011cea137b2fb4dae25 |
1 Git v1.8.0 Release Notes
2 ========================
4 Backward compatibility notes
5 ----------------------------
7 In the next major release, we will change the behavior of the "git
8 push" command. When "git push [$there]" does not say what to push, we
9 have used the traditional "matching" semantics so far (all your branches were
10 sent to the remote as long as there already are branches of the same
11 name over there). We will now use the "simple" semantics, that pushes the
12 current branch to the branch with the same name only when the current
13 branch is set to integrate with that remote branch. There is a user
14 preference configuration variable "push.default" to change this, and
15 "git push" will warn about the upcoming change until you set this
16 variable.
18 "git branch --set-upstream" is deprecated and may be removed in a
19 relatively distant future. "git branch [-u|--set-upstream-to]" has
20 been introduced with a saner order of arguments.
23 Updates since v1.7.12
24 ---------------------
26 UI, Workflows & Features
28 * A credential helper for Win32 to allow access to the keychain of
29 the logged-in user has been added.
31 * An initial port to HP NonStop.
33 * A credential helper to allow access to the Gnome keyring has been
34 added.
36 * When "git am" sanitizes the "Subject:" line, we strip the prefix from
37 "Re: subject" and also from a less common "re: subject", but left
38 the even less common "RE: subject" intact. We strip that now, too.
40 * It was tempting to say "git branch --set-upstream origin/master",
41 but that tells Git to arrange the local branch "origin/master" to
42 integrate with the currently checked out branch, which is highly
43 unlikely what the user meant. The option is deprecated; use the
44 new "--set-upstream-to" (with a short-and-sweet "-u") option
45 instead.
47 * "git cherry-pick" learned the "--allow-empty-message" option to
48 allow it to replay a commit without any log message.
50 * After "git cherry-pick -s" gave control back to the user asking
51 help to resolve conflicts, concluding "git commit" used to need to
52 be run with "-s" if the user wants to sign it off; now the command
53 leaves the sign-off line in the log template.
55 * "git daemon" learned the "--access-hook" option to allow an
56 external command to decline service based on the client address,
57 repository path, etc.
59 * "git difftool --dir-diff" learned to use symbolic links to prepare
60 a temporary copy of the working tree when available.
62 * "git grep" learned to use a non-standard pattern type by default if
63 a configuration variable tells it to.
65 * "git log -g" learned the "--grep-reflog=<pattern>" option to limit
66 its output to commits with a reflog message that matches the given
67 pattern.
69 * "git merge-base" learned the "--is-ancestor A B" option to tell if A is
70 an ancestor of B. The result is indicated by its exit status code.
72 * "git mergetool" now allows users to override the actual command used
73 with the mergetool.$name.cmd configuration variable even for built-in
74 mergetool backends.
76 * The "-Xours" backend option to "git merge -s recursive" now takes
77 effect even on binary files.
79 * "git rebase -i" learned the "--edit-todo" option to open an editor
80 to edit the instruction sheet.
83 Foreign Interface
85 * "git svn" has been updated to work with SVN 1.7.
87 * "git p4" learned the "--conflicts" option to specify what to do when
88 encountering a conflict during "p4 submit".
91 Performance, Internal Implementation, etc. (please report possible regressions)
93 * Git ships with a fall-back regexp implementation for platforms with
94 buggy regexp library, but it was easy for people to keep using their
95 platform regexp by mistake. A new test has been added to check this.
97 * The "check-docs" build target has been updated and greatly
98 simplified.
100 * The test suite is run under MALLOC_CHECK_ when running with a glibc
101 that supports the feature.
103 * The documentation in the TeXinfo format was using indented output
104 for materials meant to be examples that are better typeset in
105 monospace.
107 * Compatibility wrapper around some mkdir(2) implementations that
108 reject parameters with trailing slash has been introduced.
110 * Compatibility wrapper for systems that lack usable setitimer() has
111 been added.
113 * The option parsing of "git checkout" had error checking, dwim and
114 defaulting missing options, all mixed in the code, and issuing an
115 appropriate error message with useful context was getting harder.
116 The code has been reorganized to allow giving a proper diagnosis
117 when the user says "git checkout -b -t foo bar" (e.g. "-t" is not a
118 good name for a branch).
120 * Many internal uses of a "git merge-base" equivalent were only to see
121 if one commit fast-forwards to the other, which did not need the
122 full set of merge bases to be computed. They have been updated to
123 use less expensive checks.
125 * The heuristics to detect and silently convert latin1 to utf8 when
126 we were told to use utf-8 in the log message has been transplanted
127 from "mailinfo" to "commit" and "commit-tree".
129 * Messages given by "git <subcommand> -h" from many subcommands have
130 been marked for translation.
133 Also contains minor documentation updates and code clean-ups.
136 Fixes since v1.7.12
137 -------------------
139 Unless otherwise noted, all the fixes since v1.7.12 in the
140 maintenance track are contained in this release (see release notes
141 to them for details).
143 * The attribute system may be asked for a path that itself or its
144 leading directories no longer exists in the working tree, and it is
145 fine if we cannot open .gitattribute file in such a case. Failure
146 to open per-directory .gitattributes with error status other than
147 ENOENT and ENOTDIR should be diagnosed, but it wasn't.
149 * When looking for $HOME/.gitconfig etc., it is OK if we cannot read
150 them because they do not exist, but we did not diagnose existing
151 files that we cannot read.
153 * When "git am" is fed an input that has multiple "Content-type: ..."
154 header, it did not grok charset= attribute correctly.
156 * "git am" mishandled a patch attached as application/octet-stream
157 (e.g. not text/*); Content-Transfer-Encoding (e.g. base64) was not
158 honored correctly.
160 * "git blame MAKEFILE" run in a history that has "Makefile" but not
161 "MAKEFILE" should say "No such file MAKEFILE in HEAD", but got
162 confused on a case insensitive filesystem and failed to do so.
164 * Even during a conflicted merge, "git blame $path" always meant to
165 blame uncommitted changes to the "working tree" version; make it
166 more useful by showing cleanly merged parts as coming from the other
167 branch that is being merged.
169 * It was unclear in the documentation for "git blame" that it is
170 unnecessary for users to use the "--follow" option.
172 * Output from "git branch -v" contains "(no branch)" that could be
173 localized, but the code to align it along with the names of
174 branches was counting in bytes, not in display columns.
176 * "git cherry-pick A C B" used to replay changes in A and then B and
177 then C if these three commits had committer timestamps in that
178 order, which is not what the user who said "A C B" naturally
179 expects.
181 * A repository created with "git clone --single" had its fetch
182 refspecs set up just like a clone without "--single", leading the
183 subsequent "git fetch" to slurp all the other branches, defeating
184 the whole point of specifying "only this branch".
186 * Documentation talked about "first line of commit log" when it meant
187 the title of the commit. The description was clarified by defining
188 how the title is decided and rewording the casual mention of "first
189 line" to "title".
191 * "git cvsimport" did not thoroughly cleanse tag names that it
192 inferred from the names of the tags it obtained from CVS, which
193 caused "git tag" to barf and stop the import in the middle.
195 * Earlier we made the diffstat summary line that shows the number of
196 lines added/deleted localizable, but it was found irritating having
197 to see them in various languages on a list whose discussion language
198 is English, and this change has been reverted.
200 * "git fetch --all", when passed "--no-tags", did not honor the
201 "--no-tags" option while fetching from individual remotes (the same
202 issue existed with "--tags", but the combination "--all --tags" makes
203 much less sense than "--all --no-tags").
205 * "git fetch" over http had an old workaround for an unlikely server
206 misconfiguration; it turns out that this hurts debuggability of the
207 configuration in general, and has been reverted.
209 * "git fetch" over http advertised that it supports "deflate", which
210 is much less common, and did not advertise the more common "gzip" on
211 its Accept-Encoding header.
213 * "git fetch" over the dumb-http revision walker could segfault when
214 curl's multi interface was used.
215 (merge 188923f jk/maint-http-half-auth-push later to maint).
217 * "git gc --auto" notified the user that auto-packing has triggered
218 even under the "--quiet" option.
220 * After "gitk" showed the contents of a tag, neither "Reread
221 references" nor "Reload" updated what is shown as the
222 contents of it when the user overwrote the tag with "git tag -f".
224 * "git log --all-match --grep=A --grep=B" ought to show commits that
225 mention both A and B, but when these three options are used with
226 --author or --committer, it showed commits that mention either A or
227 B (or both) instead.
229 * "git p4", when "--use-client-spec" and "--detect-branches" are used
230 together, misdetected branches.
232 * "git receive-pack" (the counterpart to "git push") did not give
233 progress output while processing objects it received to the puser
234 when run over the smart-http protocol.
236 * When you misspell the command name you give to the "exec" action in
237 the "git rebase -i" instruction sheet you were told that 'rebase' is not a
238 git subcommand from "git rebase --continue".
240 * The subcommand in "git remote" to remove a defined remote was
241 "rm" and the command did not take a fully-spelled "remove".
243 * The interactive prompt that "git send-email" gives was error prone. It
244 asked "What e-mail address do you want to use?" with the address it
245 guessed (correctly) the user would want to use in its prompt,
246 tempting the user to say "y". But the response was taken as "No,
247 please use 'y' as the e-mail address instead", which is most
248 certainly not what the user meant.
250 * "git show --format='%ci'" did not give the timestamp correctly for
251 commits created without human readable name on the "committer" line.
253 * "git show --quiet" ought to be a synonym for "git show -s", but
254 wasn't.
256 * "git submodule frotz" was not diagnosed as "frotz" being an unknown
257 subcommand to "git submodule"; the user instead got a complaint
258 that "git submodule status" was run with an unknown path "frotz".
260 * "git status" honored the ignore=dirty settings in .gitmodules but
261 "git commit" didn't.
263 * "gitweb" did not give the correct committer timezone in its feed
264 output due to a typo.