| blob | d4001c5ae5e82e6955b0de8b4d762c56d3bc3cdc |
1 Git v2.2 Release Notes
2 ======================
4 Updates since v2.1
5 ------------------
7 Ports
9 * Building on older MacOS X systems automatically sets
10 the necessary NO_APPLE_COMMON_CRYPTO build-time option.
12 * The support to build with NO_PTHREADS has been resurrected.
14 * Compilation options has been updated a bit to support z/OS port
15 better.
18 UI, Workflows & Features
20 * "git archive" learned to filter what gets archived with pathspec.
22 * "git config --edit --global" starts from a skeletal per-user
23 configuration file contents, instead of a total blank, when the
24 user does not already have any. This immediately reduces the
25 need for a later "Have you forgotten setting core.user?" and we
26 can add more to the template as we gain more experience.
28 * "git stash list -p" used to be almost always a no-op because each
29 stash entry is represented as a merge commit. It learned to show
30 the difference between the base commit version and the working tree
31 version, which is in line with what "git show" gives.
33 * Sometimes users want to report a bug they experience on their
34 repository, but they are not at liberty to share the contents of
35 the repository. "fast-export" was taught an "--anonymize" option
36 to replace blob contents, names of people and paths and log
37 messages with bland and simple strings to help them.
39 * "git difftool" learned an option to stop feeding paths to the
40 diff backend when it exits with a non-zero status.
42 * "git grep" allows to paint (or not paint) partial matches on
43 context lines when showing "grep -C<num>" output in color.
45 * "log --date=iso" uses a slight variant of ISO 8601 format that is
46 made more human readable. A new "--date=iso-strict" option gives
47 datetime output that is more strictly conformant.
49 * The logic "git prune" uses is more resilient against various corner
50 cases.
52 * A broken reimplementation of Git could write an invalid index that
53 records both stage #0 and higher stage entries for the same path.
54 We now notice and reject such an index, as there is no sensible
55 fallback (we do not know if the broken tool wanted to resolve and
56 forgot to remove higher stage entries, or if it wanted to unresolve
57 and forgot to remove the stage#0 entry).
59 * The temporary files "git mergetool" uses are named to avoid too
60 many dots in them (e.g. a temporary file for "hello.c" used to be
61 named e.g. "hello.BASE.4321.c" but now uses underscore instead,
62 e.g. "hello_BASE_4321.c").
64 * The temporary files "git mergetools" uses can be placed in a newly
65 created temporary directory, instead of the current directory, by
66 setting the mergetool.writeToTemp configuration variable.
68 * "git mergetool" understands "--tool bc" now, as version 4 of
69 BeyondCompare can be driven the same way as its version 3 and it
70 feels awkward to say "--tool bc3".
72 * The "pre-receive" and "post-receive" hooks are no longer required
73 to consume their input fully (not following this requirement used
74 to result in intermittent errors in "git push").
76 * The pretty-format specifier "%d", which expanded to " (tagname)"
77 for a tagged commit, gained a cousin "%D" that just gives the
78 "tagname" without frills.
80 * "git push" learned "--signed" push, that allows a push (i.e.
81 request to update the refs on the other side to point at a new
82 history, together with the transmission of necessary objects) to be
83 signed, so that it can be verified and audited, using the GPG
84 signature of the person who pushed, that the tips of branches at a
85 public repository really point the commits the pusher wanted to,
86 without having to "trust" the server.
88 * "git interpret-trailers" is a new filter to programmatically edit
89 the tail end of the commit log messages.
91 * "git help everyday" shows the "Everyday Git in 20 commands or so"
92 document, whose contents have been updated to more modern Git
93 practice.
95 * On the "git svn" front, work to reduce memory consumption and
96 to improve handling of mergeinfo progresses.
99 Performance, Internal Implementation, etc.
101 * The API to manipulate the "refs" has been restructured to make it
102 more transactional, with the eventual goal to allow all-or-none
103 atomic updates and migrating the storage to something other than
104 the traditional filesystem based one (e.g. databases).
106 * The lockfile API and its users have been cleaned up.
108 * We no longer attempt to keep track of individual dependencies to
109 the header files in the build procedure, relying on automated
110 dependency generation support from modern compilers.
112 * In tests, we have been using NOT_{MINGW,CYGWIN} test prerequisites
113 long before negated prerequisites e.g. !MINGW were invented.
114 The former has been converted to the latter to avoid confusion.
116 * Looking up remotes configuration in a repository with very many
117 remotes defined has been optimized.
119 * There are cases where you lock and open to write a file, close it
120 to show the updated contents to external processes, and then have
121 to update the file again while still holding the lock, but the
122 lockfile API lacked support for such an access pattern.
124 * The API to allocate the structure to keep track of commit
125 decoration has been updated to make it less cumbersome to use.
127 * An in-core caching layer to let us avoid reading the same
128 configuration files number of times has been added. A few commands
129 have been converted to use this subsystem.
131 * Various code paths have been cleaned up and simplified by using
132 "strbuf", "starts_with()", and "skip_prefix()" APIs more.
134 * A few codepaths that died when large blobs that would not fit in
135 core are involved in their operation have been taught to punt
136 instead, by e.g. marking too large a blob as not to be diffed.
138 * A few more code paths in "commit" and "checkout" have been taught
139 to repopulate the cache-tree in the index, to help speed up later
140 "write-tree" (used in "commit") and "diff-index --cached" (used in
141 "status").
143 * A common programming mistake to assign the same short option name
144 to two separate options is detected by parse_options() API to help
145 developers.
147 * The code path to write out the packed-refs file has been optimized,
148 which especially matters in a repository with a large number of
149 refs.
151 * The check to see if a ref $F can be created by making sure no
152 existing ref has $F/ as its prefix has been optimized, which
153 especially matters in a repository with a large number of existing
154 refs.
156 * "git fsck" was taught to check contents of tag objects a bit more.
158 * "git hash-object" was taught a "--literally" option to help
159 debugging.
161 * When running a required clean filter, we do not have to mmap the
162 original before feeding the filter. Instead, stream the file
163 contents directly to the filter and process its output.
165 * The scripts in the test suite can be run with "-x" option to show
166 a shell-trace of each command run in them.
168 * The "run-command" API learned to manage the argv and environment
169 array for child process, alleviating the need for the callers to
170 allocate and deallocate them.
172 * Some people use AsciiDoctor, instead of AsciiDoc, to format our
173 documentation set; the documentation has been adjusted, as
174 AsciiDoctor is pickier than AsciiDoc in its input mark-up.
177 Also contains various documentation updates and code clean-ups.
180 Fixes since v2.1
181 ----------------
183 Unless otherwise noted, all the fixes since v2.1 in the maintenance
184 track are contained in this release (see the maintenance releases'
185 notes for details).
187 * "git log --pretty/format=" with an empty format string did not
188 mean the more obvious "No output whatsoever" but "Use default
189 format", which was counterintuitive.
191 * "git -c section.var command" and "git -c section.var= command"
192 should pass the configuration differently (the former should be a
193 boolean true, the latter should be an empty string).
195 * Applying a patch not generated by Git in a subdirectory used to
196 check the whitespace breakage using the attributes for incorrect
197 paths. Also whitespace checks were performed even for paths
198 excluded via "git apply --exclude=<path>" mechanism.
200 * "git bundle create" with date-range specification were meant to
201 exclude tags outside the range, but it didn't.
203 * "git add x" where x that used to be a directory has become a
204 symbolic link to a directory misbehaved.
206 * The prompt script checked $GIT_DIR/ref/stash file to see if there
207 is a stash, which was a no-no.
209 * Pack-protocol documentation had a minor typo.
211 * "git checkout -m" did not switch to another branch while carrying
212 the local changes forward when a path was deleted from the index.
214 * "git daemon" (with NO_IPV6 build configuration) used to incorrectly
215 use the hostname even when gethostbyname() reported that the given
216 hostname is not found.
217 (merge 107efbe rs/daemon-fixes later to maint).
219 * With sufficiently long refnames, "git fast-import" could have
220 overflown an on-stack buffer.
222 * After "pack-refs --prune" packed refs at the top-level, it failed
223 to prune them.
225 * Progress output from "git gc --auto" was visible in "git fetch -q".
227 * We used to pass -1000 to poll(2), expecting it to also mean "no
228 timeout", which should be spelled as -1.
230 * "git rebase" documentation was unclear that it is required to
231 specify on what <upstream> the rebase is to be done when telling it
232 to first check out <branch>.
233 (merge 95c6826 so/rebase-doc later to maint).
235 * "git push" over HTTP transport had an artificial limit on number of
236 refs that can be pushed imposed by the command line length.
237 (merge 26be19b jk/send-pack-many-refspecs later to maint).
239 * When receiving an invalid pack stream that records the same object
240 twice, multiple threads got confused due to a race.
241 (merge ab791dd jk/index-pack-threading-races later to maint).
243 * An attempt to remove the entire tree in the "git fast-import" input
244 stream caused it to misbehave.
245 (merge 2668d69 mb/fast-import-delete-root later to maint).
247 * Reachability check (used in "git prune" and friends) did not add a
248 detached HEAD as a starting point to traverse objects still in use.
249 (merge c40fdd0 mk/reachable-protect-detached-head later to maint).
251 * "git config --add section.var val" used to lose existing
252 section.var whose value was an empty string.
253 (merge c1063be ta/config-add-to-empty-or-true-fix later to maint).
255 * "git fsck" failed to report that it found corrupt objects via its
256 exit status in some cases.
257 (merge 30d1038 jk/fsck-exit-code-fix later to maint).
259 * Use of "--verbose" option used to break "git branch --merged".
260 (merge 12994dd jk/maint-branch-verbose-merged later to maint).
262 * Some MUAs mangled a line in a message that begins with "From " to
263 ">From " when writing to a mailbox file and feeding such an input
264 to "git am" used to lose such a line.
265 (merge 85de86a jk/mbox-from-line later to maint).
267 * "rev-parse --verify --quiet $name" is meant to quietly exit with a
268 non-zero status when $name is not a valid object name, but still
269 gave error messages in some cases.
271 * A handful of C source files have been updated to include
272 "git-compat-util.h" as the first thing, to conform better to our
273 coding guidelines.
274 (merge 1c4b660 da/include-compat-util-first-in-c later to maint).
276 * t7004 test, which tried to run Git with small stack space, has been
277 updated to give a bit larger stack to avoid false breakage on some
278 platforms.
279 (merge b9a1907 sk/tag-contains-wo-recursion later to maint).
281 * A few documentation pages had example sections marked up not quite
282 correctly, which passed AsciiDoc but failed with AsciiDoctor.
283 (merge c30c43c bc/asciidoc-pretty-formats-fix later to maint).
284 (merge f8a48af bc/asciidoc later to maint).
286 * "gitweb" used deprecated CGI::startfrom, which was removed from
287 CGI.pm as of 4.04; use CGI::start_from instead.
288 (merge 4750f4b rm/gitweb-start-form later to maint).
290 * Newer versions of 'meld' breaks the auto-detection we use to see if
291 they are new enough to support the `--output` option.
292 (merge b12d045 da/mergetool-meld later to maint).
294 * "git pack-objects" forgot to disable the codepath to generate
295 object reachability bitmap when it needs to split the resulting
296 pack.
297 (merge 2113471 jk/pack-objects-no-bitmap-when-splitting later to maint).
299 * The code to use cache-tree trusted the on-disk data too much and
300 fell into an infinite loop upon seeing an incorrectly recorded
301 index file.
302 (merge 729dbbd jk/cache-tree-protect-from-broken-libgit2 later to maint).
304 * "git fetch" into a repository where branch B was deleted earlier
305 back when it had reflog enabled, and then branch B/C is fetched
306 into it without reflog enabled, which is arguably an unlikely
307 corner case, unnecessarily failed.
308 (merge aae828b jk/fetch-reflog-df-conflict later to maint).
310 * "git log --first-parent -L..." used to crash.
311 (merge a8787c5 tm/line-log-first-parent later to maint).