| blob | 0d983a597546feebc0ddaaa641caee8a2c31349e |
1 Git 2.9 Release Notes
2 =====================
4 Backward compatibility note
5 ---------------------------
7 The end-user facing Porcelain level commands in the "git diff" and
8 "git log" by default enables the rename detection; you can still use
9 "diff.renames" configuration variable to disable this.
11 Merging two branches that have no common ancestor with "git merge" is
12 by default forbidden now to prevent creating such an unusual merge by
13 mistake.
15 The output formats of "git log" that indents the commit log message by
16 4 spaces now expands HT in the log message by default. You can use
17 the "--no-expand-tabs" option to disable this.
20 Updates since v2.8
21 ------------------
23 UI, Workflows & Features
25 * The end-user facing Porcelain level commands like "diff" and "log"
26 now enables the rename detection by default.
28 * The credential.helper configuration variable is cumulative and
29 there is no good way to override it from the command line. As
30 a special case, giving an empty string as its value now serves
31 as the signal to clear the values specified in various files.
33 * A new "interactive.diffFilter" configuration can be used to
34 customize the diff shown in "git add -i" session.
36 * "git p4" now allows P4 author names to be mapped to Git author
37 names.
39 * "git rebase -x" can be used without passing "-i" option.
41 * "git -c credential.<var>=<value> submodule" can now be used to
42 propagate configuration variables related to credential helper
43 down to the submodules.
45 * "git tag" can create an annotated tag without explicitly given an
46 "-a" (or "-s") option (i.e. when a tag message is given). A new
47 configuration variable, tag.forceSignAnnotated, can be used to tell
48 the command to create signed tag in such a situation.
50 * "git merge" used to allow merging two branches that have no common
51 base by default, which led to a brand new history of an existing
52 project created and then get pulled by an unsuspecting maintainer,
53 which allowed an unnecessary parallel history merged into the
54 existing project. The command has been taught not to allow this by
55 default, with an escape hatch "--allow-unrelated-histories" option
56 to be used in a rare event that merges histories of two projects
57 that started their lives independently.
59 * "git pull" has been taught to pass --allow-unrelated-histories
60 option to underlying "git merge".
62 * "git apply -v" learned to report paths in the patch that were
63 skipped via --include/--exclude mechanism or being outside the
64 current working directory.
66 * Shell completion (in contrib/) updates.
68 * The commit object name reported when "rebase -i" stops has been
69 shortened.
71 * "git worktree add" can be given "--no-checkout" option to only
72 create an empty worktree without checking out the files.
74 * "git mergetools" learned to drive ExamDiff.
76 * "git pull --rebase" learned "--[no-]autostash" option, so that
77 the rebase.autostash configuration variable set to true can be
78 overridden from the command line.
80 * When "git log" shows the log message indented by 4-spaces, the
81 remainder of a line after a HT does not align in the way the author
82 originally intended. The command now expands tabs by default in
83 such a case, and allows the users to override it with a new option,
84 "--no-expand-tabs".
86 * "git send-email" now uses a more readable timestamps when
87 formulating a message ID.
89 * "git rerere" can encounter two or more files with the same conflict
90 signature that have to be resolved in different ways, but there was
91 no way to record these separate resolutions.
92 (merge 890fca8 jc/rerere-multi later to maint).
94 * "git p4" learned to record P4 jobs in Git commit that imports from
95 the history in Perforce.
97 * "git describe --contains" often made a hard-to-justify choice of
98 tag to give name to a given commit, because it tried to come up
99 with a name with smallest number of hops from a tag, causing an old
100 commit whose close descendant that is recently tagged were not
101 described with respect to an old tag but with a newer tag. It did
102 not help that its computation of "hop" count was further tweaked to
103 penalize being on a side branch of a merge. The logic has been
104 updated to favor using the tag with the oldest tagger date, which
105 is a lot easier to explain to the end users: "We describe a commit
106 in terms of the (chronologically) oldest tag that contains the
107 commit."
108 (merge 7550424 js/name-rev-use-oldest-ref later to maint).
111 Performance, Internal Implementation, Development Support etc.
113 * The embedded args argv-array in the child process is used to build
114 the command line to run pack-objects instead of using a separate
115 array of strings.
117 * A test for tags has been restructured so that more parts of it can
118 easily be run on a platform without a working GnuPG.
120 * The startup_info data, which records if we are working inside a
121 repository (among other things), are now uniformly available to Git
122 subcommand implementations, and Git avoids attempting to touch
123 references when we are not in a repository.
125 * The command line argument parser for "receive-pack" has been
126 rewritten to use parse-options.
128 * A major part of "git submodule update" has been ported to C to take
129 advantage of the recently added framework to run download tasks in
130 parallel.
132 * Rename bunch of tests on "git clone" for better organization.
134 * The tests that involve running httpd leaked the system-wide
135 configuration in /etc/gitconfig to the tested environment.
137 * Build updates for MSVC.
139 * The repository set-up sequence has been streamlined (the biggest
140 change is that there is no longer git_config_early()), so that we
141 do not attempt to look into refs/* when we know we do not have a
142 Git repository.
144 * Code restructuring around the "refs" area to prepare for pluggable
145 refs backends.
147 * Sources to many test helper binaries (and the generated helpers)
148 have been moved to t/helper/ subdirectory to reduce clutter at the
149 top level of the tree.
151 Note that this can break your tests if you check out revisions
152 across the merge boundary of this topic, e0b58519 (Merge branch
153 'nd/test-helpers', 2016-04-29), as bin-wrappers/test-* are not
154 rebuilt to point the underlying executables. For now, "make
155 distclean" is your friend.
157 * Unify internal logic between "git tag -v" and "git verify-tag"
158 commands by making one directly call into the other.
159 (merge bef234b st/verify-tag later to maint).
161 * "merge-recursive" strategy incorrectly checked if a path that is
162 involved in its internal merge exists in the working tree.
165 Also contains various documentation updates and code clean-ups.
168 Fixes since v2.8
169 ----------------
171 Unless otherwise noted, all the fixes since v2.8 in the maintenance
172 track are contained in this release (see the maintenance releases'
173 notes for details).
175 * "git config --get-urlmatch", unlike other variants of the "git
176 config --get" family, did not signal error with its exit status
177 when there was no matching configuration.
179 * The "--local-env-vars" and "--resolve-git-dir" options of "git
180 rev-parse" failed to work outside a repository when the command's
181 option parsing was rewritten in 1.8.5 era.
183 * "git index-pack --keep[=<msg>] pack-$name.pack" simply did not work.
185 * Fetching of history by naming a commit object name directly didn't
186 work across remote-curl transport.
188 * A small memory leak in an error codepath has been plugged in xdiff
189 code.
191 * strbuf_getwholeline() did not NUL-terminate the buffer on certain
192 corner cases in its error codepath.
194 * "git mergetool" did not work well with conflicts that both sides
195 deleted.
197 * "git send-email" had trouble parsing alias file in mailrc format
198 when lines in it had trailing whitespaces on them.
200 * When "git merge --squash" stopped due to conflict, the concluding
201 "git commit" failed to read in the SQUASH_MSG that shows the log
202 messages from all the squashed commits.
204 * "git merge FETCH_HEAD" dereferenced NULL pointer when merging
205 nothing into an unborn history (which is arguably unusual usage,
206 which perhaps was the reason why nobody noticed it).
208 * When "git worktree" feature is in use, "git branch -d" allowed
209 deletion of a branch that is checked out in another worktree,
210 which was wrong.
212 * When "git worktree" feature is in use, "git branch -m" renamed a
213 branch that is checked out in another worktree without adjusting
214 the HEAD symbolic ref for the worktree.
216 * "git diff -M" used to work better when two originally identical
217 files A and B got renamed to X/A and X/B by pairing A to X/A and B
218 to X/B, but this was broken in the 2.0 timeframe.
220 * "git send-pack --all <there>" was broken when its command line
221 option parsing was written in the 2.6 timeframe.
223 * "git format-patch --help" showed `-s` and `--no-patch` as if these
224 are valid options to the command. We already hide `--patch` option
225 from the documentation, because format-patch is about showing the
226 diff, and the documentation now hides these options as well.
228 * When running "git blame $path" with unnormalized data in the index
229 for the path, the data in the working tree was blamed, even though
230 "git add" would not have changed what is already in the index, due
231 to "safe crlf" that disables the line-end conversion. It has been
232 corrected.
234 * A change back in version 2.7 to "git branch" broke display of a
235 symbolic ref in a non-standard place in the refs/ hierarchy (we
236 expect symbolic refs to appear in refs/remotes/*/HEAD to point at
237 the primary branch the remote has, and as .git/HEAD to point at the
238 branch we locally checked out).
240 * A partial rewrite of "git submodule" in the 2.7 timeframe changed
241 the way the gitdir: pointer in the submodules point at the real
242 repository location to use absolute paths by accident. This has
243 been corrected.
245 * "git commit" misbehaved in a few minor ways when an empty message
246 is given via -m '', all of which has been corrected.
248 * Support for CRAM-MD5 authentication method in "git imap-send" did
249 not work well.
251 * Upcoming OpenSSL 1.1.0 will break compilation b updating a few APIs
252 we use in imap-send, which has been adjusted for the change.
253 (merge 1245c74 ky/imap-send-openssl-1.1.0 later to maint).
255 * The socks5:// proxy support added back in 2.6.4 days was not aware
256 that socks5h:// proxies behave differently.
258 * "git config" had a codepath that tried to pass a NULL to
259 printf("%s"), which nobody seems to have noticed.
261 * On Cygwin, object creation uses the "create a temporary and then
262 rename it to the final name" pattern, not "create a temporary,
263 hardlink it to the final name and then unlink the temporary"
264 pattern.
266 This is necessary to use Git on Windows shared directories, and is
267 already enabled for the MinGW and plain Windows builds. It also
268 has been used in Cygwin packaged versions of Git for quite a while.
269 See http://thread.gmane.org/gmane.comp.version-control.git/291853
271 * "merge-octopus" strategy did not ensure that the index is clean
272 when merge begins.
274 * When "git merge" notices that the merge can be resolved purely at
275 the tree level (without having to merge blobs) and the resulting
276 tree happens to already exist in the object store, it forgot to
277 update the index, which lead to an inconsistent state for later
278 operations.
280 * "git submodule" reports the paths of submodules the command
281 recurses into, but this was incorrect when the command was not run
282 from the root level of the superproject.
283 (merge 2ab5660 sb/submodule-path-misc-bugs later to maint).
285 * The "user.useConfigOnly" configuration variable makes it an error
286 if users do not explicitly set user.name and user.email. However,
287 its check was not done early enough and allowed another error to
288 trigger, reporting that the default value we guessed from the
289 system setting was unusable. This was a suboptimal end-user
290 experience as we want the users to set user.name/user.email without
291 relying on the auto-detection at all.
292 (merge d3c06c1 da/user-useconfigonly later to maint).
294 * "git mv old new" did not adjust the path for a submodule that lives
295 as a subdirectory inside old/ directory correctly.
296 (merge a127331 sb/mv-submodule-fix later to maint).
298 * "git replace -e" did not honour "core.editor" configuration.
299 (merge 36b1437 js/replace-edit-use-editor-configuration later to maint).
301 * "git push" from a corrupt repository that attempts to push a large
302 number of refs deadlocked; the thread to relay rejection notices
303 for these ref updates blocked on writing them to the main thread,
304 after the main thread at the receiving end notices that the push
305 failed and decides not to read these notices and return a failure.
306 (merge c4b2751 jk/push-client-deadlock-fix later to maint).
308 * mmap emulation on Windows has been optimized and work better without
309 consuming paging store when not needed.
310 (merge d5425d1 js/win32-mmap later to maint).
312 * A question by "git send-email" to ask the identity of the sender
313 has been updated.
314 (merge 0d6b21e jd/send-email-to-whom later to maint).
316 * UI consistency improvements for "git mergetool".
317 (merge cce076e nf/mergetool-prompt later to maint).
319 * Other minor clean-ups and documentation updates
320 (merge 8b5a3e9 kn/for-each-tag-branch later to maint).
321 (merge 9c60d9f sb/misc-cleanups later to maint).
322 (merge 7a6a44c cc/apply later to maint).
323 (merge 6594883 nd/remove-unused later to maint).
324 (merge 0ff7410 sg/test-lib-simplify-expr-away later to maint).
325 (merge 060e776 jk/fix-attribute-macro-in-2.5 later to maint).
326 (merge d16df0c rt/string-list-lookup-cleanup later to maint).