9 * Building on older MacOS X systems automatically sets
10 the necessary NO_APPLE_COMMON_CRYPTO build-time option.
13 UI, Workflows & Features
15 * "git archive" learned to filter what gets archived with pathspec.
17 * "git config --edit --global" starts from a skeletal per-user
18 configuration file contents, instead of a total blank, when the
19 user does not already have any. This immediately reduces the
20 need for a later "Have you forgotten setting core.user?" and we
21 can add more to the template as we gain more experience.
23 * "git stash list -p" used to be almost always a no-op because each
24 stash entry is represented as a merge commit. It learned to show
25 the difference between the base commit version and the working tree
26 version, which is in line with what "git show" gives.
28 * Sometimes users want to report a bug they experience on their
29 repository, but they are not at liberty to share the contents of
30 the repository. "fast-export" was taught an "--anonymize" option
31 to replace blob contents, names of people and paths and log
32 messages with bland and simple strings to help them.
34 * "log --date=iso" uses a slight variant of ISO 8601 format that is
35 made more human readable. A new "--date=iso-strict" option gives
36 datetime output that is more strictly conformant.
38 * A broken reimplementation of Git could write an invalid index that
39 records both stage #0 and higher stage entries for the same path.
40 We now notice and reject such an index, as there is no sensible
41 fallback (we do not know if the broken tool wanted to resolve and
42 forgot to remove higher stage entries, or if it wanted to unresolve
43 and forgot to remove the stage#0 entry).
45 * The "pre-receive" and "post-receive" hooks are no longer required
46 to consume their input fully (not following this requirement used
47 to result in intermittent errors in "git push").
49 * The pretty-format specifier "%d", which expanded to " (tagname)"
50 for a tagged commit, gained a cousin "%D" that just gives the
51 "tagname" without frills.
53 * "git push" learned "--signed" push, that allows a push (i.e.
54 request to update the refs on the other side to point at a new
55 history, together with the transmission of necessary objects) to be
56 signed, so that it can be verified and audited, using the GPG
57 signature of the person who pushed, that the tips of branches at a
58 public repository really point the commits the pusher wanted to,
59 without having to "trust" the server.
61 Performance, Internal Implementation, etc.
63 * The API to manipulate the "refs" is currently undergoing a revamp
64 to make it more transactional, with the eventual goal to allow
65 all-or-none atomic updates and migrating the storage to something
66 other than the traditional filesystem based one (e.g. databases).
68 * The lockfile API and its users have been cleaned up.
70 * We no longer attempt to keep track of individual dependencies to
71 the header files in the build procedure, relying on automated
72 dependency generation support from modern compilers.
74 * In tests, we have been using NOT_{MINGW,CYGWIN} test prerequisites
75 long before negated prerequisites e.g. !MINGW were invented.
76 The former has been converted to the latter to avoid confusion.
78 * Looking up remotes configuration in a repository with very many
79 remotes defined has been optimized.
81 * There are cases where you lock and open to write a file, close it
82 to show the updated contents to external processes, and then have
83 to update the file again while still holding the lock, but the
84 lockfile API lacked support for such an access pattern.
86 * The API to allocate the structure to keep track of commit
87 decoration has been updated to make it less cumbersome to use.
89 * An in-core caching layer to let us avoid reading the same
90 configuration files number of times has been added. A few commands
91 have been converted to use this subsystem.
93 * Various code paths have been cleaned up and simplified by using
94 "strbuf", "starts_with()", and "skip_prefix()" APIs more.
96 * A few codepaths that died when large blobs that would not fit in
97 core are involved in their operation have been taught to punt
98 instead, by e.g. marking too large a blob as not to be diffed.
100 * A few more code paths in "commit" and "checkout" have been taught
101 to repopulate the cache-tree in the index, to help speed up later
102 "write-tree" (used in "commit") and "diff-index --cached" (used in
105 * A common programming mistake to assign the same short option name
106 to two separate options is detected by parse_options() API to help
109 * The code path to write out the packed-refs file has been optimized,
110 which especially matters in a repository with a large number of
113 * The check to see if a ref $F can be created by making sure no
114 existing ref has $F/ as its prefix has been optimized, which
115 especially matters in a repository with a large number of existing
118 * "git fsck" was taught to check contents of tag objects a bit more.
120 * "git hash-object" was taught a "--literally" option to help
123 * When running a required clean filter, we do not have to mmap the
124 original before feeding the filter. Instead, stream the file
125 contents directly to the filter and process its output.
127 Also contains various documentation updates and code clean-ups.
133 Unless otherwise noted, all the fixes since v2.1 in the maintenance
134 track are contained in this release (see the maintenance releases'
137 * "git log --pretty/format=" with an empty format string did not
138 mean the more obvious "No output whatsoever" but "Use default
139 format", which was counterintuitive.
141 * Implementations of "tar" that do not understand an extended pax
142 header would extract the contents of it in a regular file; make
143 sure the permission bits of this file follows the same tar.umask
144 configuration setting.
146 * "git -c section.var command" and "git -c section.var= command"
147 should pass the configuration differently (the former should be a
148 boolean true, the latter should be an empty string).
150 * Applying a patch not generated by Git in a subdirectory used to
151 check the whitespace breakage using the attributes for incorrect
152 paths. Also whitespace checks were performed even for paths
153 excluded via "git apply --exclude=<path>" mechanism.
155 * "git bundle create" with date-range specification were meant to
156 exclude tags outside the range, but it didn't.
158 * "git add x" where x that used to be a directory has become a
159 symbolic link to a directory misbehaved.
161 * The prompt script checked $GIT_DIR/ref/stash file to see if there
162 is a stash, which was a no-no.
164 * Pack-protocol documentation had a minor typo.
166 * "git checkout -m" did not switch to another branch while carrying
167 the local changes forward when a path was deleted from the index.
169 * "git daemon" (with NO_IPV6 build configuration) used to incorrectly
170 use the hostname even when gethostbyname() reported that the given
171 hostname is not found.
172 (merge 107efbe rs/daemon-fixes later to maint).
174 * With sufficiently long refnames, "git fast-import" could have
175 overflown an on-stack buffer.
177 * After "pack-refs --prune" packed refs at the top-level, it failed
180 * Progress output from "git gc --auto" was visible in "git fetch -q".
182 * We used to pass -1000 to poll(2), expecting it to also mean "no
183 timeout", which should be spelled as -1.
185 * "git rebase" documentation was unclear that it is required to
186 specify on what <upstream> the rebase is to be done when telling it
187 to first check out <branch>.
188 (merge 95c6826 so/rebase-doc later to maint).
190 * "git push" over HTTP transport had an artificial limit on number of
191 refs that can be pushed imposed by the command line length.
192 (merge 26be19b jk/send-pack-many-refspecs later to maint).
194 * When receiving an invalid pack stream that records the same object
195 twice, multiple threads got confused due to a race.
196 (merge ab791dd jk/index-pack-threading-races later to maint).
198 * An attempt to remove the entire tree in the "git fast-import" input
199 stream caused it to misbehave.
200 (merge 2668d69 mb/fast-import-delete-root later to maint).
202 * Reachability check (used in "git prune" and friends) did not add a
203 detached HEAD as a starting point to traverse objects still in use.
204 (merge c40fdd0 mk/reachable-protect-detached-head later to maint).
206 * "git config --add section.var val" used to lose existing
207 section.var whose value was an empty string.
208 (merge c1063be ta/config-add-to-empty-or-true-fix later to maint).
210 * "git fsck" failed to report that it found corrupt objects via its
211 exit status in some cases.
212 (merge 30d1038 jk/fsck-exit-code-fix later to maint).
214 * Use of "--verbose" option used to break "git branch --merged".
215 (merge 12994dd jk/maint-branch-verbose-merged later to maint).
217 * Some MUAs mangled a line in a message that begins with "From " to
218 ">From " when writing to a mailbox file and feeding such an input
219 to "git am" used to lose such a line.
220 (merge 85de86a jk/mbox-from-line later to maint).
222 * "rev-parse --verify --quiet $name" is meant to quietly exit with a
223 non-zero status when $name is not a valid object name, but still
224 gave error messages in some cases.
226 * A handful of C source files have been updated to include
227 "git-compat-util.h" as the first thing, to conform better to our
229 (merge 1c4b660 da/include-compat-util-first-in-c later to maint).
231 * t7004 test, which tried to run Git with small stack space, has been
232 updated to give a bit larger stack to avoid false breakage on some
234 (merge b9a1907 sk/tag-contains-wo-recursion later to maint).
236 * A few documentation pages had example sections marked up not quite
237 correctly, which passed AsciiDoc but failed with AsciiDoctor.
238 (merge c30c43c bc/asciidoc-pretty-formats-fix later to maint).
projects / git / blob