Documentation / RelNotes / 2.15.0.txton commit connect: teach client to recognize v1 server response (2609043)
   1Git 2.15 Release Notes
   2======================
   3
   4Backward compatibility notes and other notable changes.
   5
   6 * Use of an empty string as a pathspec element that is used for
   7   'everything matches' is still warned and Git asks users to use a
   8   more explicit '.' for that instead.  The hope is that existing
   9   users will not mind this change, and eventually the warning can be
  10   turned into a hard error, upgrading the deprecation into removal of
  11   this (mis)feature.  That is now scheduled to happen in the upcoming
  12   release.
  13
  14 * Git now avoids blindly falling back to ".git" when the setup
  15   sequence said we are _not_ in Git repository.  A corner case that
  16   happens to work right now may be broken by a call to die("BUG").
  17   We've tried hard to locate such cases and fixed them, but there
  18   might still be cases that need to be addressed--bug reports are
  19   greatly appreciated.
  20
  21 * "branch --set-upstream" that has been deprecated in Git 1.8 has
  22   finally been retired.
  23
  24
  25Updates since v2.14
  26-------------------
  27
  28UI, Workflows & Features
  29
  30 * An example that is now obsolete has been removed from a sample hook,
  31   and an old example in it that added a sign-off manually has been
  32   improved to use the interpret-trailers command.
  33
  34 * The advice message given when "git rebase" stops for conflicting
  35   changes has been improved.
  36
  37 * The "rerere-train" script (in contrib/) learned the "--overwrite"
  38   option to allow overwriting existing recorded resolutions.
  39
  40 * "git contacts" (in contrib/) now lists the address on the
  41   "Reported-by:" trailer to its output, in addition to those on
  42   S-o-b: and other trailers, to make it easier to notify (and thank)
  43   the original bug reporter.
  44
  45 * "git rebase", especially when it is run by mistake and ends up
  46   trying to replay many changes, spent long time in silence.  The
  47   command has been taught to show progress report when it spends
  48   long time preparing these many changes to replay (which would give
  49   the user a chance to abort with ^C).
  50
  51 * "git merge" learned a "--signoff" option to add the Signed-off-by:
  52   trailer with the committer's name.
  53
  54 * "git diff" learned to optionally paint new lines that are the same
  55   as deleted lines elsewhere differently from genuinely new lines.
  56
  57 * "git interpret-trailers" learned to take the trailer specifications
  58   from the command line that overrides the configured values.
  59
  60 * "git interpret-trailers" has been taught a "--parse" and a few
  61   other options to make it easier for scripts to grab existing
  62   trailer lines from a commit log message.
  63
  64 * "gitweb" shows a link to visit the 'raw' contents of blbos in the
  65   history overview page.
  66
  67 * "[gc] rerereResolved = 5.days" used to be invalid, as the variable
  68   is defined to take an integer counting the number of days.  It now
  69   is allowed.
  70
  71 * The code to acquire a lock on a reference (e.g. while accepting a
  72   push from a client) used to immediately fail when the reference is
  73   already locked---now it waits for a very short while and retries,
  74   which can make it succeed if the lock holder was holding it during
  75   a read-only operation.
  76
  77 * "branch --set-upstream" that has been deprecated in Git 1.8 has
  78   finally been retired.
  79
  80 * The codepath to call external process filter for smudge/clean
  81   operation learned to show the progress meter.
  82
  83 * "git rev-parse" learned "--is-shallow-repository", that is to be
  84   used in a way similar to existing "--is-bare-repository" and
  85   friends.
  86
  87
  88Performance, Internal Implementation, Development Support etc.
  89
  90 * Conversion from uchar[20] to struct object_id continues.
  91
  92 * Start using selected c99 constructs in small, stable and
  93   essentialpart of the system to catch people who care about
  94   older compilers that do not grok them.
  95
  96 * The filter-process interface learned to allow a process with long
  97   latency give a "delayed" response.
  98
  99 * Many uses of comparision callback function the hashmap API uses
 100   cast the callback function type when registering it to
 101   hashmap_init(), which defeats the compile time type checking when
 102   the callback interface changes (e.g. gaining more parameters).
 103   The callback implementations have been updated to take "void *"
 104   pointers and cast them to the type they expect instead.
 105
 106 * Because recent Git for Windows do come with a real msgfmt, the
 107   build procedure for git-gui has been updated to use it instead of a
 108   hand-rolled substitute.
 109
 110 * "git grep --recurse-submodules" has been reworked to give a more
 111   consistent output across submodule boundary (and do its thing
 112   without having to fork a separate process).
 113
 114 * A helper function to read a single whole line into strbuf
 115   mistakenly triggered OOM error at EOF under certain conditions,
 116   which has been fixed.
 117   (merge 642956cf45 rs/strbuf-getwholeline-fix later to maint).
 118
 119 * The "ref-store" code reorganization continues.
 120
 121 * "git commit" used to discard the index and re-read from the filesystem
 122   just in case the pre-commit hook has updated it in the middle; this
 123   has been optimized out when we know we do not run the pre-commit hook.
 124   (merge 680ee550d7 kw/commit-keep-index-when-pre-commit-is-not-run later to maint).
 125
 126 * Updates to the HTTP layer we made recently unconditionally used
 127   features of libCurl without checking the existence of them, causing
 128   compilation errors, which has been fixed.  Also migrate the code to
 129   check feature macros, not version numbers, to cope better with
 130   libCurl that vendor ships with backported features.
 131
 132 * The API to start showing progress meter after a short delay has
 133   been simplified.
 134   (merge 8aade107dd jc/simplify-progress later to maint).
 135
 136 * Code clean-up to avoid mixing values read from the .gitmodules file
 137   and values read from the .git/config file.
 138
 139 * We used to spend more than necessary cycles allocating and freeing
 140   piece of memory while writing each index entry out.  This has been
 141   optimized.
 142
 143 * Platforms that ship with a separate sha1 with collision detection
 144   library can link to it instead of using the copy we ship as part of
 145   our source tree.
 146
 147 * Code around "notes" have been cleaned up.
 148   (merge 3964281524 mh/notes-cleanup later to maint).
 149
 150 * The long-standing rule that an in-core lockfile instance, once it
 151   is used, must not be freed, has been lifted and the lockfile and
 152   tempfile APIs have been updated to reduce the chance of programming
 153   errors.
 154
 155 * Our hashmap implementation in hashmap.[ch] is not thread-safe when
 156   adding a new item needs to expand the hashtable by rehashing; add
 157   an API to disable the automatic rehashing to work it around.
 158
 159 * Many of our programs consider that it is OK to release dynamic
 160   storage that is used throughout the life of the program by simply
 161   exiting, but this makes it harder to leak detection tools to avoid
 162   reporting false positives.  Plug many existing leaks and introduce
 163   a mechanism for developers to mark that the region of memory
 164   pointed by a pointer is not lost/leaking to help these tools.
 165
 166 * As "git commit" to conclude a conflicted "git merge" honors the
 167   commit-msg hook, "git merge" that records a merge commit that
 168   cleanly auto-merges should, but it didn't.
 169
 170 * The codepath for "git merge-recursive" has been cleaned up.
 171
 172 * Many leaks of strbuf have been fixed.
 173
 174 * "git imap-send" has our own implementation of the protocol and also
 175   can use more recent libCurl with the imap protocol support.  Update
 176   the latter so that it can use the credential subsystem, and then
 177   make it the default option to use, so that we can eventually
 178   deprecate and remove the former.
 179
 180 * "make style" runs git-clang-format to help developers by pointing
 181   out coding style issues.
 182
 183 * A test to demonstrate "git mv" failing to adjust nested submodules
 184   has been added.
 185   (merge c514167df2 hv/mv-nested-submodules-test later to maint).
 186
 187 * On Cygwin, "ulimit -s" does not report failure but it does not work
 188   at all, which causes an unexpected success of some tests that
 189   expect failures under a limited stack situation.  This has been
 190   fixed.
 191
 192
 193Also contains various documentation updates and code clean-ups.
 194
 195
 196Fixes since v2.14
 197-----------------
 198
 199 * "%C(color name)" in the pretty print format always produced ANSI
 200   color escape codes, which was an early design mistake.  They now
 201   honor the configuration (e.g. "color.ui = never") and also tty-ness
 202   of the output medium.
 203
 204 * The http.{sslkey,sslCert} configuration variables are to be
 205   interpreted as a pathname that honors "~[username]/" prefix, but
 206   weren't, which has been fixed.
 207
 208 * Numerous bugs in walking of reflogs via "log -g" and friends have
 209   been fixed.
 210
 211 * "git commit" when seeing an totally empty message said "you did not
 212   edit the message", which is clearly wrong.  The message has been
 213   corrected.
 214
 215 * When a directory is not readable, "gitweb" fails to build the
 216   project list.  Work this around by skipping such a directory.
 217
 218 * Some versions of GnuPG fails to kill gpg-agent it auto-spawned
 219   and such a left-over agent can interfere with a test.  Work it
 220   around by attempting to kill one before starting a new test.
 221
 222 * A recently added test for the "credential-cache" helper revealed
 223   that EOF detection done around the time the connection to the cache
 224   daemon is torn down were flaky.  This was fixed by reacting to
 225   ECONNRESET and behaving as if we got an EOF.
 226
 227 * "git log --tag=no-such-tag" showed log starting from HEAD, which
 228   has been fixed---it now shows nothing.
 229
 230 * The "tag.pager" configuration variable was useless for those who
 231   actually create tag objects, as it interfered with the use of an
 232   editor.  A new mechanism has been introduced for commands to enable
 233   pager depending on what operation is being carried out to fix this,
 234   and then "git tag -l" is made to run pager by default.
 235
 236 * "git push --recurse-submodules $there HEAD:$target" was not
 237   propagated down to the submodules, but now it is.
 238
 239 * Commands like "git rebase" accepted the --rerere-autoupdate option
 240   from the command line, but did not always use it.  This has been
 241   fixed.
 242
 243 * "git clone --recurse-submodules --quiet" did not pass the quiet
 244   option down to submodules.
 245
 246 * Test portability fix for OBSD.
 247
 248 * Portability fix for OBSD.
 249
 250 * "git am -s" has been taught that some input may end with a trailer
 251   block that is not Signed-off-by: and it should refrain from adding
 252   an extra blank line before adding a new sign-off in such a case.
 253
 254 * "git svn" used with "--localtime" option did not compute the tz
 255   offset for the timestamp in question and instead always used the
 256   current time, which has been corrected.
 257
 258 * Memory leak in an error codepath has been plugged.
 259
 260 * "git stash -u" used the contents of the committed version of the
 261   ".gitignore" file to decide which paths are ignored, even when the
 262   file has local changes.  The command has been taught to instead use
 263   the locally modified contents.
 264
 265 * bash 4.4 or newer gave a warning on NUL byte in command
 266   substitution done in "git stash"; this has been squelched.
 267
 268 * "git grep -L" and "git grep --quiet -L" reported different exit
 269   codes; this has been corrected.
 270
 271 * When handshake with a subprocess filter notices that the process
 272   asked for an unknown capability, Git did not report what program
 273   the offending subprocess was running.  This has been corrected.
 274
 275 * "git apply" that is used as a better "patch -p1" failed to apply a
 276   taken from a file with CRLF line endings to a file with CRLF line
 277   endings.  The root cause was because it misused convert_to_git()
 278   that tried to do "safe-crlf" processing by looking at the index
 279   entry at the same path, which is a nonsense---in that mode, "apply"
 280   is not working on the data in (or derived from) the index at all.
 281   This has been fixed.
 282
 283 * Killing "git merge --edit" before the editor returns control left
 284   the repository in a state with MERGE_MSG but without MERGE_HEAD,
 285   which incorrectly tells the subsequent "git commit" that there was
 286   a squash merge in progress.  This has been fixed.
 287
 288 * "git archive" did not work well with pathspecs and the
 289   export-ignore attribute.
 290
 291 * In addition to "cc: <a@dd.re.ss> # cruft", "cc: a@dd.re.ss # cruft"
 292   was taught to "git send-email" as a valid way to tell it that it
 293   needs to also send a carbon copy to <a@dd.re.ss> in the trailer
 294   section.
 295   (merge cc90750677 mm/send-email-cc-cruft later to maint).
 296
 297 * "git branch -M a b" while on a branch that is completely unrelated
 298   to either branch a or branch b misbehaved when multiple worktree
 299   was in use.  This has been fixed.
 300   (merge 31824d180d nd/worktree-kill-parse-ref later to maint).
 301
 302 * "git gc" and friends when multiple worktrees are used off of a
 303   single repository did not consider the index and per-worktree refs
 304   of other worktrees as the root for reachability traversal, making
 305   objects that are in use only in other worktrees to be subject to
 306   garbage collection.
 307
 308 * A regression to "gitk --bisect" by a recent update has been fixed.
 309   (merge 1d0538e486 mh/packed-ref-store-prep later to maint).
 310
 311 * "git -c submodule.recurse=yes pull" did not work as if the
 312   "--recurse-submodules" option was given from the command line.
 313   This has been corrected.
 314
 315 * Unlike "git commit-tree < file", "git commit-tree -F file" did not
 316   pass the contents of the file verbatim and instead completed an
 317   incomplete line at the end, if exists.  The latter has been updated
 318   to match the behaviour of the former.
 319   (merge c818e74332 rk/commit-tree-make-F-verbatim later to maint).
 320
 321 * Many codepaths did not diagnose write failures correctly when disks
 322   go full, due to their misuse of write_in_full() helper function,
 323   which have been corrected.
 324   (merge f48ecd38cb jk/write-in-full-fix later to maint).
 325
 326 * "git help co" now says "co is aliased to ...", not "git co is".
 327   (merge b3a8076e0d ks/help-alias-label later to maint).
 328
 329 * "git archive", especially when used with pathspec, stored an empty
 330   directory in its output, even though Git itself never does so.
 331   This has been fixed.
 332   (merge 4318094047 rs/archive-excluded-directory later to maint).
 333
 334 * API error-proofing which happens to also squelch warnings from GCC.
 335   (merge c788c54cde tg/refs-allowed-flags later to maint).
 336
 337 * The explanation of the cut-line in the commit log editor has been
 338   slightly tweaked.
 339   (merge 8c4b1a3593 ks/commit-do-not-touch-cut-line later to maint).
 340
 341 * "git gc" tries to avoid running two instances at the same time by
 342   reading and writing pid/host from and to a lock file; it used to
 343   use an incorrect fscanf() format when reading, which has been
 344   corrected.
 345   (merge afe2fab72c aw/gc-lockfile-fscanf-fix later to maint).
 346
 347 * The scripts to drive TravisCI has been reorganized and then an
 348   optimization to avoid spending cycles on a branch whose tip is
 349   tagged has been implemented.
 350   (merge 8376eb4a8f ls/travis-scriptify later to maint).
 351
 352 * The test linter has been taught that we do not like "echo -e".
 353   (merge 1a6d46895d tb/test-lint-echo-e later to maint).
 354
 355 * Code cmp.std.c nitpick.
 356   (merge ac7da78ede mh/for-each-string-list-item-empty-fix later to maint).
 357
 358 * A regression fix for 2.11 that made the code to read the list of
 359   alternate object stores overrun the end of the string.
 360   (merge f0f7bebef7 jk/info-alternates-fix later to maint).
 361
 362 * Other minor doc, test and build updates and code cleanups.
 363   (merge f094b89a4d ma/parse-maybe-bool later to maint).
 364   (merge 39b00fa4d4 jk/drop-sha1-entry-pos later to maint).
 365   (merge 6cdf8a7929 ma/ts-cleanups later to maint).
 366   (merge 7560f547e6 ma/up-to-date later to maint).
 367   (merge 0db3dc75f3 rs/apply-epoch later to maint).
 368   (merge 74f1bd912b dw/diff-highlight-makefile-fix later to maint).
 369   (merge f991761eb8 jk/config-lockfile-leak-fix later to maint).
 370   (merge 150efef1e7 ma/pkt-line-leakfix later to maint).
 371   (merge 5554451de6 mg/timestamp-t-fix later to maint).
 372   (merge 276d0e35c0 ma/split-symref-update-fix later to maint).
 373   (merge 3bc4b8f7c7 bb/doc-eol-dirty later to maint).
 374   (merge c1bb33c99c jk/system-path-cleanup later to maint).
 375   (merge ab46e6fc72 cc/subprocess-handshake-missing-capabilities later to maint).
 376   (merge f7a32dd97f kd/doc-for-each-ref later to maint).
 377   (merge be94568bc7 ez/doc-duplicated-words-fix later to maint).
 378   (merge 01e4be6c3d ks/test-readme-phrasofix later to maint).