| blob | 67bae07a40012dd9eab83653ce5bdefc8379d347 |
1 Git v2.46 Release Notes
2 =======================
4 Backward Compatibility Notes
6 (None at this moment)
8 UI, Workflows & Features
10 * The "--rfc" option of "git format-patch" learned to take an
11 optional string value to be used in place of "RFC" to tweak the
12 "[PATCH]" on the subject header.
14 * The credential helper protocol, together with the HTTP layer, have
15 been enhanced to support authentication schemes different from
16 username & password pair, like Bearer and NTLM.
18 * Command line completion script (in contrib/) learned to complete
19 "git symbolic-ref" a bit better (you need to enable plumbing
20 commands to be completed with GIT_COMPLETION_SHOW_ALL_COMMANDS).
22 * When the user responds to a prompt given by "git add -p" with an
23 unsupported command, list of available commands were given, which
24 was too much if the user knew what they wanted to type but merely
25 made a typo. Now the user gets a much shorter error message.
27 * The color parsing code learned to handle 12-bit RGB colors, spelled
28 as "#RGB" (in addition to "#RRGGBB" that is already supported).
30 * The operation mode options (like "--get") the "git config" command
31 uses have been deprecated and replaced with subcommands (like "git
32 config get").
34 * "git tag" learned the "--trailer" option to futz with the trailers
35 in the same way as "git commit" does.
37 * A new global "--no-advice" option can be used to disable all advice
38 messages, which is meant to be used only in scripts.
40 * Updates to symbolic refs can now be made as a part of ref
41 transaction.
43 * The trailer API has been reshuffled a bit.
45 * Terminology to call various ref-like things are getting
46 straightened out.
48 * The command line completion script (in contrib/) has been adjusted
49 to the recent update to "git config" that adopted subcommand based
50 UI.
52 * The knobs to tweak how reftable files are written have been made
53 available as configuration variables.
55 * When "git push" notices that the commit at the tip of the ref on
56 the other side it is about to overwrite does not exist locally, it
57 used to first try fetching it if the local repository is a partial
58 clone. The command has been taught not to do so and immediately
59 fail instead.
61 * The promisor.quiet configuration knob can be set to true to make
62 lazy fetching from promisor remotes silent.
64 * The inter/range-diff output has been moved to the end of the patch
65 when format-patch adds it to a single patch, instead of writing it
66 before the patch text, to be consistent with what is done for a
67 cover letter for a multi-patch series.
69 * A new command has been added to migrate a repository that uses the
70 files backend for its ref storage to use the reftable backend, with
71 limitations.
73 * "git diff --exit-code --ext-diff" learned to take the exit status
74 of the external diff driver into account when deciding the exit
75 status of the overall "git diff" invocation when configured to do
76 so.
78 * "git update-ref --stdin" learned to handle transactional updates of
79 symbolic-refs.
81 * "git format-patch --interdiff" for multi-patch series learned to
82 turn on cover letters automatically (unless told never to enable
83 cover letter with "--no-cover-letter" and such).
85 * The "--heads" option of "ls-remote" and "show-ref" has been been
86 deprecated; "--branches" replaces "--heads".
88 * For over a year, setting add.interactive.useBuiltin configuration
89 variable did nothing but giving a "this does not do anything"
90 warning. The warning has been removed.
93 Performance, Internal Implementation, Development Support etc.
95 * Advertise "git contacts", a tool for newcomers to find people to
96 ask review for their patches, a bit more in our developer
97 documentation.
99 * In addition to building the objects needed, try to link the objects
100 that are used in fuzzer tests, to make sure at least they build
101 without bitrot, in Linux CI runs.
103 * Code to write out reftable has seen some optimization and
104 simplification.
106 * Tests to ensure interoperability between reftable written by jgit
107 and our code have been added and enabled in CI.
109 * The singleton index_state instance "the_index" has been eliminated
110 by always instantiating "the_repository" and replacing references
111 to "the_index" with references to its .index member.
113 * Git-GUI has a new maintainer, Johannes Sixt.
115 * The "test-tool" has been taught to run testsuite tests in parallel,
116 bypassing the need to use the "prove" tool.
118 * The "whitespace check" task that was enabled for GitHub Actions CI
119 has been ported to GitLab CI.
121 * The refs API lost functions that implicitly assumes to work on the
122 primary ref_store by forcing the callers to pass a ref_store as an
123 argument.
125 * Code clean-up to reduce inter-function communication inside
126 builtin/config.c done via the use of global variables.
128 * The pack bitmap code saw some clean-up to prepare for a follow-up topic.
130 * Preliminary code clean-up for "git send-email".
132 * The default "creation-factor" used by "git format-patch" has been
133 raised to make it more aggressively find matching commits.
135 * Before discovering the repository details, We used to assume SHA-1
136 as the "default" hash function, which has been corrected. Hopefully
137 this will smoke out codepaths that rely on such an unwarranted
138 assumptions.
140 * The project decision making policy has been documented.
142 * The strcmp-offset tests have been rewritten using the unit test
143 framework.
145 * "git add -p" learned to complain when an answer with more than one
146 letter is given to a prompt that expects a single letter answer.
148 * The alias-expanded command lines are logged to the trace output.
150 * A new test was added to ensure git commands that are designed to
151 run outside repositories do work.
153 * Basic unit tests for reftable have been reimplemented under the
154 unit test framework.
156 * A pair of test helpers that essentially are unit tests on hash
157 algorithms have been rewritten using the unit-tests framework.
159 * A test helper that essentially is unit tests on the "decorate"
160 logic has been rewritten using the unit-tests framework.
162 * Many memory leaks in the sparse-checkout code paths have been
163 plugged.
165 * "make check-docs" noticed problems and reported to its output but
166 failed to signal its findings with its exit status, which has been
167 corrected.
169 * Building with "-Werror -Wwrite-strings" is now supported.
171 * To help developers, the build procedure now allows builders to use
172 CFLAGS_APPEND to specify additional CFLAGS.
174 * "oidtree" tests were rewritten to use the unit test framework.
176 * The structure of the document that records longer-term project
177 decisions to deprecate/remove/update various behaviour has been
178 outlined.
180 * The pseudo-merge reachability bitmap to help more efficient storage
181 of the reachability bitmap in a repository with too many refs has
182 been added.
184 * When "git merge" sees that the index cannot be refreshed (e.g. due
185 to another process doing the same in the background), it died but
186 after writing MERGE_HEAD etc. files, which was useless for the
187 purpose to recover from the failure.
189 * The output from "git cat-file --batch-check" and "--batch-command
190 (info)" should not be unbuffered, for which some tests have been
191 added.
193 * A CPP macro USE_THE_REPOSITORY_VARIABLE is introduced to help
194 transition the codebase to rely less on the availability of the
195 singleton the_repository instance.
197 * "git version --build-options" reports the version information of
198 OpenSSL and other libraries (if used) in the build.
200 * Memory ownership rules for the in-core representation of
201 remote.*.url configuration values have been straightened out, which
202 resulted in a few leak fixes and code clarification.
205 Fixes since v2.45
206 -----------------
208 * "git rebase --signoff" used to forget that it needs to add a
209 sign-off to the resulting commit when told to continue after a
210 conflict stops its operation.
212 * The procedure to build multi-pack-index got confused by the
213 replace-refs mechanism, which has been corrected by disabling the
214 latter.
216 * The "-k" and "--rfc" options of "format-patch" will now error out
217 when used together, as one tells us not to add anything to the
218 title of the commit, and the other one tells us to add "RFC" in
219 addition to "PATCH".
221 * "git stash -S" did not handle binary files correctly, which has
222 been corrected.
224 * A scheduled "git maintenance" job is expected to work on all
225 repositories it knows about, but it stopped at the first one that
226 errored out. Now it keeps going.
228 * zsh can pretend to be a normal shell pretty well except for some
229 glitches that we tickle in some of our scripts. Work them around
230 so that "vimdiff" and our test suite works well enough with it.
232 * Command line completion support for zsh (in contrib/) has been
233 updated to stop exposing internal state to end-user shell
234 interaction.
236 * Tests that try to corrupt in-repository files in chunked format did
237 not work well on macOS due to its broken "mv", which has been
238 worked around.
240 * The maximum size of attribute files is enforced more consistently.
242 * Unbreak CI jobs so that we do not attempt to use Python 2 that has
243 been removed from the platform.
245 * Git 2.43 started using the tree of HEAD as the source of attributes
246 in a bare repository, which has severe performance implications.
247 For now, revert the change, without ripping out a more explicit
248 support for the attr.tree configuration variable.
250 * The "--exit-code" option of "git diff" command learned to work with
251 the "--ext-diff" option.
253 * Windows CI running in GitHub Actions started complaining about the
254 order of arguments given to calloc(); the imported regex code uses
255 the wrong order almost consistently, which has been corrected.
257 * Expose "name conflict" error when a ref creation fails due to D/F
258 conflict in the ref namespace, to improve an error message given by
259 "git fetch".
260 (merge 9339fca23e it/refs-name-conflict later to maint).
262 * The SubmittingPatches document now refers folks to manpages
263 translation project.
265 * The documentation for "git diff --name-only" has been clarified
266 that it is about showing the names in the post-image tree.
268 * The credential helper that talks with osx keychain learned to avoid
269 storing back the authentication material it just got received from
270 the keychain.
271 (merge e1ab45b2da kn/osxkeychain-skip-idempotent-store later to maint).
273 * The chainlint script (invoked during "make test") did nothing when
274 it failed to detect the number of available CPUs. It now falls
275 back to 1 CPU to avoid the problem.
277 * Revert overly aggressive "layered defence" that went into 2.45.1
278 and friends, which broke "git-lfs", "git-annex", and other use
279 cases, so that we can rebuild necessary counterparts in the open.
281 * "git init" in an already created directory, when the user
282 configuration has includeif.onbranch, started to fail recently,
283 which has been corrected.
285 * Memory leaks in "git mv" has been plugged.
287 * The safe.directory configuration knob has been updated to
288 optionally allow leading path matches.
290 * An overly large ".gitignore" files are now rejected silently.
292 * Upon expiration event, the credential subsystem forgot to clear
293 in-core authentication material other than password (whose support
294 was added recently), which has been corrected.
296 * Fix for an embarrassing typo that prevented Python2 tests from running
297 anywhere.
299 * Varargs functions that are unannotated as printf-like or execl-like
300 have been annotated as such.
302 * "git am" has a safety feature to prevent it from starting a new
303 session when there already is a session going. It reliably
304 triggers when a mbox is given on the command line, but it has to
305 rely on the tty-ness of the standard input. Add an explicit way to
306 opt out of this safety with a command line option.
307 (merge 62c71ace44 jk/am-retry later to maint).
309 * A leak in "git imap-send" that somehow escapes LSan has been
310 plugged.
312 * Setting core.abbrev too early before the repository set-up
313 (typically in "git clone") caused segfault, which as been
314 corrected.
316 * When the user adds to "git rebase -i" instruction to "pick" a merge
317 commit, the error experience is not pleasant. Such an error is now
318 caught earlier in the process that parses the todo list.
320 * We forgot to normalize the result of getcwd() to NFC on macOS where
321 all other paths are normalized, which has been corrected. This still
322 does not address the case where core.precomposeUnicode configuration
323 is not defined globally.
325 * Earlier we stopped using the tree of HEAD as the default source of
326 attributes in a bare repository, but failed to document it. This
327 has been corrected.
329 * "git update-server-info" and "git commit-graph --write" have been
330 updated to use the tempfile API to avoid leaving cruft after
331 failing.
333 * An unused extern declaration for mingw has been removed to prevent
334 it from causing build failure.
336 * A helper function shared between two tests had a copy-paste bug,
337 which has been corrected.
339 * "git fetch-pack -k -k" without passing "--lock-pack" (which we
340 never do ourselves) did not work at all, which has been corrected.
342 * CI job to build minimum fuzzers learned to pass NO_CURL=NoThanks to
343 the build procedure, as its build environment does not offer, or
344 the rest of the build needs, anything cURL.
345 (merge 4e66b5a990 jc/fuzz-sans-curl later to maint).
347 * "git diff --no-ext-diff" when diff.external is configured ignored
348 the "--color-moved" option.
349 (merge 0f4b0d4cf0 rs/diff-color-moved-w-no-ext-diff-fix later to maint).
351 * Other code cleanup, docfix, build fix, etc.
352 (merge 493fdae046 ew/object-convert-leakfix later to maint).