1 To: git@vger.kernel.org
3 Subject: What's cooking in git.git (May 2015, #01; Mon, 4)
4 X-master-at: 3d4a3ffe64162b45ae7c991fc60623ecb4678cfd
5 X-next-at: 7a0482398323b9bd8dbb64ab9010e63a95b81bce
7 What's cooking in git.git (May 2015, #01; Mon, 4)
8 --------------------------------------------------
10 Here are the topics that have been cooking. Commits prefixed with
11 '-' are only in 'pu' (proposed updates) while commits prefixed with
14 Git 2.4 final has been tagged. We'll wait for a few days to see
15 if there are brown-paper-bag bugs, fix them if there are any, and
16 then rewind 'next' and rebuild it to kick off the next cycle.
18 You can find the changes described here in the integration branches
19 of the repositories listed at
21 http://git-blame.blogspot.com/p/git-public-repositories.html
23 --------------------------------------------------
26 * mh/ref-directory-file (2015-05-03) 18 commits
27 - reflog_expire(): integrate lock_ref_sha1_basic() errors into ours
28 - ref_transaction_commit(): delete extra "the" from error message
29 - ref_transaction_commit(): provide better error messages
30 - rename_ref(): integrate lock_ref_sha1_basic() errors into ours
31 - lock_ref_sha1_basic(): improve diagnostics for D/F conflicts
32 - lock_ref_sha1_basic(): report errors via a "struct strbuf *err"
33 - verify_refname_available(): report errors via a "struct strbuf *err"
34 - verify_refname_available(): rename function
35 - refs: check for D/F conflicts among refs processed in a transaction
36 - ref_transaction_commit(): use a string_list for detecting duplicates
37 - is_refname_available(): use dirname in first loop
38 - struct nonmatching_ref_data: store a refname instead of a ref_entry
39 - report_refname_conflict(): inline function
40 - entry_matches(): inline function
41 - is_refname_available(): convert local variable "dirname" to strbuf
42 - is_refname_available(): avoid shadowing "dir" variable
43 - is_refname_available(): explain the reason for an early exit
44 - t1404: new tests of D/F conflicts within ref transactions
45 (this branch uses mh/ref-lock-avoid-running-out-of-fds.)
47 Somehow this does not seem to pass its own test.
50 * nd/dwim-wildcards-as-pathspecs (2015-05-03) 1 commit
51 - pathspec: avoid the need of "--" when wildcard is used
53 A heuristic to help the "git <cmd> <revs> <pathspec>" command line
54 convention to catch mistyped paths is to make sure all the non-rev
55 parameters in the later part of the command line are names of the
56 files in the working tree, but that means "git grep $str -- \*.c"
57 must always be disambiguated with "--", because nobody sane will
58 create a file whose name literally is asterisk-dot-see. Loosen the
59 heuristic to declare that with a wildcard string the user likely
60 meant to give us a pathspec.
65 * pt/pull-tests (2015-05-03) 7 commits
66 - t5521: test --dry-run does not make any changes
67 - t5520: test --rebase failure on unborn branch with index
68 - t5520: test --rebase with multiple branches
69 - t5520: test work tree fast-forward when fetch updates head
70 - t5520: test for failure if index has unresolved entries
71 - t5520: implement tests for no merge candidates cases
72 - t5520: test pulling multiple branches into an empty repository
74 Add more test coverage to "git pull". This conflicts slightly with
75 a new test jc/merge topic adds, but resolution is fairly trivial.
80 * sg/complete-decorate-full-not-long (2015-05-03) 1 commit
81 - completion: fix and update 'git log --decorate=' options
83 The completion for "log --decorate=" parameter value was incorrect.
88 * sg/completion-no-redundant-all-command-list (2015-05-03) 1 commit
89 - completion: remove redundant __git_compute_all_commands() call
94 * jc/hash-object (2015-05-04) 4 commits
95 - write_sha1_file(): do not use a separate sha1[] array
96 - t1007: add hash-object --literally tests
97 - write_sha1_file_prepare: fix buffer overrun with extra-long object type
98 - git-hash-object.txt: document --literally option
100 "hash-object --literally" introduced in v2.2 days was not prepared
101 to take a really long object type name.
103 Will merge to 'next'.
106 * jn/clean-use-error-not-fprintf-on-stderr (2015-05-04) 1 commit
107 - config: use error() instead of fprintf(stderr, ...)
109 Some error messages in "git config" were emitted without calling
110 the usual error() facility.
112 Will merge to 'next'.
115 * jc/clone-bundle (2015-04-30) 1 commit
116 - repack: optionally create a clone.bundle
120 --------------------------------------------------
123 * kk/log-merges-config (2015-04-21) 5 commits
124 - bash-completion: add support for git-log --merges= and log.merges
125 - t4202-log: add tests for --merges=
126 - Documentation: add git-log --merges= option and log.merges config. var
127 - log: honor log.merges= option
128 - revision: add --merges={show|only|hide} option
130 "git log" (but not other commands in the "log" family) learned to
131 pay attention to the log.merges configuration variable that can be
132 set to "show" (the normal behaviour), "only" (hide non-merge
133 commits), or "hide" (hide merge commits). --merges=(show|only|hide)
134 can be used to override the setting from the command line.
136 The documentation may need to be updated once more ($gmane/267250).
137 Waiting for a reroll.
140 * mh/fdopen-with-retry (2015-03-06) 6 commits
141 - buffer_fdinit(): use fdopen_with_retry()
142 - update_info_file(): use fdopen_with_retry()
143 - copy_to_log(): use fdopen_with_retry()
144 - fdopen_lock_file(): use fdopen_with_retry()
145 - SQUASH??? $gmane/264889
146 - xfdopen(): if first attempt fails, free memory and try again
148 Various parts of the code where they call fdopen() can fail when
149 they run out of memory; attempt to proceed by retrying the
150 operation after freeing some resource.
155 * mg/httpd-tests-update-for-apache-2.4 (2015-04-08) 2 commits
156 - t/lib-git-svn: check same httpd module dirs as lib-httpd
157 - t/lib-httpd: load mod_unixd
159 This is the first two commits in a three-patch series $gmane/266962
161 with updated log message ($gmane/268061).
164 * pt/xdg-config-path (2015-04-12) 7 commits
165 - path.c: remove home_config_paths()
166 - git-config: replace use of home_config_paths()
167 - git-commit: replace use of home_config_paths()
168 - credential-store.c: replace home_config_paths() with xdg_config_home()
169 - dir.c: replace home_config_paths() with xdg_config_home()
170 - attr.c: replace home_config_paths() with xdg_config_home()
171 - path.c: implement xdg_config_home()
172 (this branch uses pt/credential-xdg.)
174 Seen some discussions.
175 Waiting for a reroll ($gmane/267518).
178 * mh/numparse (2015-03-19) 14 commits
179 - diff_opt_parse(): use convert_i() when handling --abbrev=<num>
180 - diff_opt_parse(): use convert_i() when handling "-l<num>"
181 - opt_arg(): simplify pointer handling
182 - opt_arg(): report errors parsing option values
183 - opt_arg(): use convert_i() in implementation
184 - opt_arg(): val is always non-NULL
185 - builtin_diff(): detect errors when parsing --unified argument
186 - handle_revision_opt(): use convert_ui() when handling "--abbrev="
187 - strtoul_ui(), strtol_i(): remove functions
188 - handle_revision_opt(): use convert_i() when handling "-<digit>"
189 - handle_revision_opt(): use skip_prefix() in many places
190 - write_subdirectory(): use convert_ui() for parsing mode
191 - cacheinfo_callback(): use convert_ui() when handling "--cacheinfo"
192 - numparse: new module for parsing integral numbers
194 Many codepaths use unchecked use of strtol() and friends (or even
195 worse, atoi()). Introduce a set of wrappers that try to be more
202 * tf/gitweb-project-listing (2015-03-19) 5 commits
203 - gitweb: make category headings into links when they are directories
204 - gitweb: optionally set project category from its pathname
205 - gitweb: add a link under the search box to clear a project filter
206 - gitweb: if the PATH_INFO is incomplete, use it as a project_filter
207 - gitweb: fix typo in man page
209 Update gitweb to make it more pleasant to deal with a hierarchical
210 forest of repositories.
212 Any comments from those who use or have their own code in Gitweb?
215 * jc/a-lone-dash-stands-for-previous-branch (2015-03-16) 1 commit
216 - "-" and "@{-1}" on various programs
218 Lose special case code to make a lone dash "-" mean the previous
219 branch aka "@{-1}" from a handful subcommands, and instead support
220 the notation throughout the system by reimplementing it at the
223 Needs tests, documentation updates, etc. Also does only a half-way
224 job dealing with range notation, which needs to be fixed before the
225 series goes anywhere.
228 * nd/list-files (2015-02-09) 21 commits
229 - t3080: tests for git-list-files
230 - list-files: -M aka diff-cached
231 - list-files -F: show submodules with the new indicator '&'
232 - list-files: add -F/--classify
233 - list-files: show directories as well as files
234 - list-files: do not show duplicate cached entries
235 - list-files: sort output and remove duplicates
236 - list-files: add -t back
237 - list-files: add -1 short for --no-column
238 - list-files: add -R/--recursive short for --max-depth=-1
239 - list-files: -u does not imply showing stages
240 - list-files: make alias 'ls' default to 'list-files'
241 - list-files: a user friendly version of ls-files and more
242 - ls-files: support --max-depth
243 - ls-files: add --column
244 - ls-files: add --color to highlight file names
245 - ls-files: buffer full item in strbuf before printing
246 - ls_colors.c: highlight submodules like directories
247 - ls_colors.c: add a function to color a file name
248 - ls_colors.c: parse color.ls.* from config file
249 - ls_colors.c: add $LS_COLORS parsing code
251 A new "git list-files" Porcelain command, "ls-files" with bells and
254 Reroll to base on wt-status work ($gmane/265142) has seen some
255 positive discussions.
257 Waiting for a further polished reroll ($gmane/265534).
260 * js/fsck-opt (2015-01-21) 19 commits
261 - fsck: support ignoring objects in `git fsck` via fsck.skiplist
262 - fsck: git receive-pack: support excluding objects from fsck'ing
263 - fsck: introduce `git fsck --quick`
264 - fsck: support demoting errors to warnings
265 - fsck: document the new receive.fsck.* options
266 - fsck: allow upgrading fsck warnings to errors
267 - fsck: optionally ignore specific fsck issues completely
268 - fsck: disallow demoting grave fsck errors to warnings
269 - fsck: add a simple test for receive.fsck.*
270 - fsck: make fsck_tag() warn-friendly
271 - fsck: handle multiple authors in commits specially
272 - fsck: make fsck_commit() warn-friendly
273 - fsck: make fsck_ident() warn-friendly
274 - fsck: report the ID of the error/warning
275 - fsck: allow demoting errors to warnings via receive.fsck.warn = <key>
276 - fsck: offer a function to demote fsck errors to warnings
277 - fsck: provide a function to parse fsck message IDs
278 - fsck: introduce identifiers for fsck messages
279 - fsck: introduce fsck options
281 "fsck.warnings = <list of error tokens>" I suggested turned out to
282 be an unpopular choice (sorry Dscho).
287 * nd/untracked-cache (2015-03-12) 24 commits
288 - git-status.txt: advertisement for untracked cache
289 - untracked cache: guard and disable on system changes
290 - mingw32: add uname()
291 - t7063: tests for untracked cache
292 - update-index: test the system before enabling untracked cache
293 - update-index: manually enable or disable untracked cache
294 - status: enable untracked cache
295 - untracked-cache: temporarily disable with $GIT_DISABLE_UNTRACKED_CACHE
296 - untracked cache: mark index dirty if untracked cache is updated
297 - untracked cache: print stats with $GIT_TRACE_UNTRACKED_STATS
298 - untracked cache: avoid racy timestamps
299 - read-cache.c: split racy stat test to a separate function
300 - untracked cache: invalidate at index addition or removal
301 - untracked cache: load from UNTR index extension
302 - untracked cache: save to an index extension
303 - ewah: add convenient wrapper ewah_serialize_strbuf()
304 - untracked cache: don't open non-existent .gitignore
305 - untracked cache: mark what dirs should be recursed/saved
306 - untracked cache: record/validate dir mtime and reuse cached output
307 - untracked cache: make a wrapper around {open,read,close}dir()
308 - untracked cache: invalidate dirs recursively if .gitignore changes
309 - untracked cache: initial untracked cache validation
310 - untracked cache: record .gitignore information and dir hierarchy
311 - dir.c: optionally compute sha-1 of a .gitignore file
313 Need extra sets of eyes to review this.
316 * nd/pathspec-strip-fix (2015-04-18) 1 commit
317 - pathspec: adjust prefixlen after striping trailing slash
319 Does not quite fix ($gmane/267614).
322 * jc/diff-b-m (2015-02-23) 5 commits
325 - diffcore-rename: allow easier debugging
326 - diffcore-rename.c: add locate_rename_src()
327 - diffcore-break: allow debugging
329 "git diff -B -M" produced incorrect patch when the postimage of a
330 completely rewritten file is similar to the preimage of a removed
331 file; such a resulting file must not be expressed as a rename from
334 The fix in this patch is broken, unfortunately.
337 * pw/remote-set-url-fetch (2014-11-26) 1 commit
338 - remote: add --fetch and --both options to set-url
343 * tr/remerge-diff (2014-11-10) 9 commits
344 - t4213: avoid "|" in sed regexp
345 - log --remerge-diff: show what the conflict resolution changed
346 - name-hash: allow dir hashing even when !ignore_case
347 - merge-recursive: allow storing conflict hunks in index
348 - merge_diff_mode: fold all merge diff variants into an enum
349 - combine-diff: do not pass revs->dense_combined_merges redundantly
350 - merge-recursive: -Xindex-only to leave worktree unchanged
351 - merge-recursive: internal flag to avoid touching the worktree
352 - merge-recursive: remove dead conditional in update_stages()
354 "log -p" output learns a new way to let users inspect a merge
355 commit by showing the differences between the automerged result
356 with conflicts the person who recorded the merge would have seen
357 and the final conflict resolution that was recorded in the merge.
359 Waiting for a reroll ($gmane/256591).
362 * hv/submodule-config (2014-11-11) 4 commits
363 - do not die on error of parsing fetchrecursesubmodules option
364 - use new config API for worktree configurations of submodules
365 - extract functions for submodule config set and lookup
366 - implement submodule config cache for lookup of submodule names
368 Kicked back to 'pu' per request ($gmane/255610).
371 * tg/perf-lib-test-perf-cleanup (2013-09-19) 2 commits
372 - perf-lib: add test_perf_cleanup target
373 - perf-lib: split starting the test from the execution
375 Add test_perf_cleanup shell function to the perf suite, that allows
376 the script writers to define a test with a clean-up action.
381 * jc/show-branch (2014-03-24) 5 commits
382 - show-branch: use commit slab to represent bitflags of arbitrary width
383 - show-branch.c: remove "all_mask"
384 - show-branch.c: abstract out "flags" operation
385 - show-branch.c: lift all_mask/all_revs to a global static
386 - show-branch.c: update comment style
388 Waiting for the final step to lift the hard-limit.
390 --------------------------------------------------
393 * jk/at-push-sha1 (2015-05-03) 12 commits
394 - for-each-ref: accept "%(push)" format
395 - for-each-ref: use skip_prefix instead of starts_with
396 - sha1_name: implement @{push} shorthand
397 - sha1_name: refactor interpret_upstream_mark
398 - sha1_name: refactor upstream_mark
399 - remote.c: add branch_get_push
400 - remote.c: report specific errors from branch_get_upstream
401 - remote.c: introduce branch_get_upstream helper
402 - remote.c: provide per-branch pushremote name
403 - remote.c: hoist branch.*.remote lookup out of remote_get_1
404 - remote.c: drop "remote" pointer from "struct branch"
405 - remote.c: drop default_remote_name variable
407 Introduce <branch>@{push} short-hand to denote the remote-tracking
408 branch that tracks the branch at the remote the <branch> would be
411 Rerolled (not reviewed yet).
414 * jk/rebase-quiet-noop (2015-04-28) 1 commit
415 - rebase: silence "git checkout" for noop rebase
417 "git rebase --quiet" was not quite quiet when there is nothing to
420 Will merge to 'next'.
423 * ld/p4-case-fold (2015-04-28) 1 commit
424 - git-p4: add failing tests for case-folding p4d
426 Will merge to 'next'.
429 * va/fix-git-p4-tests (2015-04-28) 1 commit
430 - git-p4: t9814: prevent --chain-lint failure
432 Will merge to 'next'.
435 * jk/filter-branch-use-of-sed-on-incomplete-line (2015-04-29) 1 commit
436 - filter-branch: avoid passing commit message through sed
438 "filter-branch" was broken by some "sed" implementations that
439 corrupt commit log message that ends with an incomplete line.
440 Work it around by avoiding to use "sed".
442 Will merge to 'next'.
445 * ph/rebase-i-redo (2015-04-29) 1 commit
446 - rebase -i: redo tasks that die during cherry-pick
448 "git rebase -i" moved the "current" command from "todo" to "done" a
449 bit too prematurely, losing a step when a "pick" did not even start.
451 Will merge to 'next'.
454 * ep/do-not-feed-a-pointer-to-array-size (2015-04-30) 2 commits
456 - git-compat-util.h: implement a different ARRAY_SIZE macro for for safely deriving the size of array
458 Catch programmer mistake to feed a pointer not an array to
459 ARRAY_SIZE() macro, by using a couple of GCC extensions.
461 With the fix-up, will merge to 'next'.
464 * tb/blame-resurrect-convert-to-git (2015-05-03) 1 commit
465 - blame: CRLF in the working tree and LF in the repo
467 Some time ago, "git blame" (incorrectly) lost the convert_to_git()
468 call when synthesizing a fake "tip" commit that represents the
469 state in the working tree, which broke folks who record the history
470 with LF line ending to make their project portabile across
471 platforms while terminating lines in their working tree files with
472 CRLF for their platform.
474 Will merge to 'next'.
477 * jc/epochtime-wo-tz (2015-04-15) 2 commits
478 (merged to 'next' on 2015-04-21 at b926f72)
479 + parse_date_basic(): let the system handle DST conversion
480 + parse_date_basic(): return early when given a bogus timestamp
482 "git commit --date=now" or anything that relies on approxidate lost
483 the daylight-saving-time offset.
485 Will merge to 'master' in the second batch of post v2.4 cycle.
488 * jc/plug-fmt-merge-msg-leak (2015-04-20) 1 commit
489 (merged to 'next' on 2015-04-21 at 443c728)
490 + fmt-merge-msg: plug small leak of commit buffer
492 Will merge to 'master' in the second batch of post v2.4 cycle.
495 * cn/bom-in-gitignore (2015-04-16) 5 commits
496 (merged to 'next' on 2015-04-21 at f7d56f1)
497 + attr: skip UTF8 BOM at the beginning of the input file
498 + config: use utf8_bom[] from utf.[ch] in git_parse_source()
499 + utf8-bom: introduce skip_utf8_bom() helper
500 + add_excludes_from_file: clarify the bom skipping logic
501 + dir: allow a BOM at the beginning of exclude files
503 Teach the codepaths that read .gitignore and .gitattributes files
504 that these files encoded in UTF-8 may have UTF-8 BOM marker at the
505 beginning; this makes it in line with what we do for configuration
508 Will merge to 'master' in the second batch of post v2.4 cycle.
511 * ee/clean-remove-dirs (2015-04-26) 5 commits
512 - clean: improve performance when removing lots of directories
513 - p7300: add performance tests for clean
514 - t7300: add tests to document behavior of clean and nested git
515 - setup: sanity check file size in read_gitfile_gently
516 - setup: add gentle version of read_gitfile
518 Replace "is this subdirectory a separate repository that should not
519 be touched?" check "git clean" does by checking if it has .git/HEAD
520 using the submodule-related code with a more optimized check.
522 Waiting for a reroll.
525 * ep/fix-test-lib-functions-report (2015-04-16) 1 commit
526 (merged to 'next' on 2015-04-21 at 23e584f)
527 + test-lib-functions.sh: fix the second argument to some helper functions
529 Will merge to 'master' in the second batch of post v2.4 cycle.
532 * jk/still-interesting (2015-04-17) 1 commit
533 (merged to 'next' on 2015-04-21 at 58b54af)
534 + limit_list: avoid quadratic behavior from still_interesting
536 "git rev-list --objects $old --not --all" to see if everything that
537 is reachable from $old is already connected to the existing refs
538 was very inefficient.
540 Will merge to 'master' in the second batch of post v2.4 cycle.
543 * jk/type-from-string-gently (2015-04-17) 1 commit
544 (merged to 'next' on 2015-04-20 at a97611f)
545 + type_from_string_gently: make sure length matches
547 "git cat-file bl $blob" failed to barf even though there is no
548 object type that is "bl".
550 Will merge to 'master' in the first batch of post v2.4 cycle.
553 * ls/p4-changes-block-size (2015-04-20) 1 commit
554 (merged to 'next' on 2015-04-21 at 830eeed)
555 + git-p4: use -m when running p4 changes
557 "git p4" learned "--changes-block-size <n>" to read the changes in
558 chunks from Perforce, instead of making one call to "p4 changes"
559 that may trigger "too many rows scanned" error from Perforce.
561 Will merge to 'master' in the second batch of post v2.4 cycle.
564 * mm/add-p-split-error (2015-04-16) 5 commits
565 (merged to 'next' on 2015-04-23 at b1bd21a)
566 + stash -p: demonstrate failure of split with mixed y/n
567 + t3904-stash-patch: factor PERL prereq at the top of the file
568 + t3904-stash-patch: fix test description
569 + add -p: demonstrate failure when running 'edit' after a split
570 + t3701-add-interactive: simplify code
572 When "add--interactive" splits a hunk into two overlapping hunks
573 and then let the user choose only one, it sometimes feeds an
574 incorrect patch text to "git apply". Add tests to demonstrate
577 I have a slight suspicion that this may be $gmane/87202 coming back
578 and biting us (I seem to have said "let's run with this and see
579 what happens" back then).
581 Will merge to 'master' in the third batch of post v2.4 cycle.
584 * mm/usage-log-l-can-take-regex (2015-04-20) 2 commits
585 (merged to 'next' on 2015-04-21 at 03e951c)
586 + log -L: improve error message on malformed argument
587 + Documentation: change -L:<regex> to -L:<funcname>
589 Will merge to 'master' in the second batch of post v2.4 cycle.
592 * nd/t1509-chroot-test (2015-04-18) 1 commit
593 (merged to 'next' on 2015-04-21 at 60d35d3)
594 + t1509: update prepare script to be able to run t1509 in chroot again
598 Will merge to 'master' in the first batch of post v2.4 cycle.
601 * oh/fix-config-default-user-name-section (2015-04-17) 1 commit
602 (merged to 'next' on 2015-04-20 at b74b914)
603 + config: fix settings in default_user_config template
605 The default $HOME/.gitconfig file created upon "git config --global"
606 that edits it had incorrectly spelled user.name and user.email
609 Will merge to 'master' in the first batch of post v2.4 cycle.
612 * tb/t0027-crlf (2015-04-25) 3 commits
613 - t0027: Add repoMIX and LF_nul
614 (merged to 'next' on 2015-04-21 at 142cb99)
615 + t0027: support NATIVE_CRLF platforms
616 + t0027: cleanup: rename functions; avoid non-leading TABs
618 Will merge to 'next'.
621 * jk/prune-mtime (2015-04-20) 3 commits
622 (merged to 'next' on 2015-04-21 at 9990d41)
623 + sha1_file: only freshen packs once per run
624 + sha1_file: freshen pack objects before loose
625 + reachable: only mark local objects as recent
627 Access to objects in repositories that borrow from another one on a
628 slow NFS server unnecessarily got more expensive due to recent code
629 becoming more cautious in a naive way not to lose objects to pruning.
631 Will merge to 'master' in the second batch of post v2.4 cycle.
634 * jc/gitignore-precedence (2015-04-22) 1 commit
635 - ignore: info/exclude should trump core.excludesfile
637 core.excludesfile (defaulting to $XDG_HOME/git/ignore) is supposed
638 to be overridden by repository-specific .git/info/exclude file, but
639 the order was swapped from the beginning. This belatedly fixes it.
641 Will merge to 'next'.
644 * jk/git-no-more-argv0-path-munging (2015-04-22) 1 commit
645 - stop putting argv[0] dirname at front of PATH
647 We have prepended $GIT_EXEC_PATH and the path "git" is installed in
648 (typically "/usr/bin") to $PATH when invoking subprograms and hooks
649 for almost eternity, but the original use case the latter tried to
650 support was semi-bogus (i.e. install git to /opt/foo/git and run it
651 without having /opt/foo on $PATH), and more importantly it has
652 become less and less relevant as Git grew more mainstream (i.e. the
653 users would _want_ to have it on their $PATH). Stop prepending the
654 path in which "git" is installed to users' $PATH, as that would
655 interfere the command search order people depend on (e.g. they may
656 not like versions of programs that are unrelated to Git in /usr/bin
657 and want to override them by having different ones in /usr/local/bin
658 and have the latter directory earlier in their $PATH).
660 Will merge to 'next'.
663 * jk/stash-require-clean-index (2015-04-22) 3 commits
664 - stash: require a clean index to apply
665 - t3903: avoid applying onto dirty index
666 - t3903: stop hard-coding commit sha1s
668 "git stash pop/apply" forgot to make sure that not just the working
669 tree is clean but also the index is clean. The latter is important
670 as a stash application can conflict and the index will be used for
673 Will merge to 'next'.
676 * jc/merge (2015-04-29) 15 commits
677 - merge: deprecate 'git merge <message> HEAD <commit>' syntax
678 - merge: handle FETCH_HEAD internally
679 - merge: decide if we auto-generate the message early in collect_parents()
680 - merge: make collect_parents() auto-generate the merge message
681 - merge: extract prepare_merge_message() logic out
682 - merge: narrow scope of merge_names
683 - merge: split reduce_parents() out of collect_parents()
684 - merge: clarify collect_parents() logic
685 - merge: small leakfix and code simplification
686 - merge: do not check argc to determine number of remote heads
687 - merge: clarify "pulling into void" special case
688 - t5520: test pulling an octopus into an unborn branch
690 - merge: simplify code flow
691 - merge: test the top-level merge driver
692 (this branch is used by jc/merge-drop-old-syntax.)
694 "git merge FETCH_HEAD" learned that the previous "git fetch" could
695 be to create an Octopus merge, i.e. recording multiple branches
696 that are not marked as "not-for-merge"; this allows us to lose an
697 old style invocation "git merge <msg> HEAD $commits..." in the
698 implementation of "git pull" script; the old style syntax can now
701 Will merge to 'next'.
704 * jc/merge-drop-old-syntax (2015-04-29) 1 commit
705 - merge: drop 'git merge <message> HEAD <commit>' syntax
706 (this branch uses jc/merge.)
708 Stop supporting "git merge <messsage> HEAD <commit>" syntax that
709 has been deprecated since October 2007.
711 Will merge to 'next'.
714 * jk/test-chain-lint (2015-04-28) 2 commits
715 - test-lib: turn on GIT_TEST_CHAIN_LINT by default
716 - t7502-commit.sh: fix a broken and-chain
718 Will merge to 'next'.
721 * mh/ref-lock-avoid-running-out-of-fds (2015-04-25) 8 commits
722 - ref_transaction_commit(): only keep one lockfile open at a time
723 - ref_transaction_commit(): remove the local flags variables
724 - write_ref_sha1(): inline function at callers
725 - commit_ref_update(): new function, extracted from write_ref_sha1()
726 - write_ref_to_lockfile(): new function, extracted from write_ref_sha1()
727 - refs.c: remove lock_fd from struct ref_lock
728 - t7004: rename ULIMIT test prerequisite to ULIMIT_STACK_SIZE
729 - update-ref: test handling large transactions properly
730 (this branch is used by mh/ref-directory-file.)
732 "git update-ref --stdin" was converted to be "transactional" some
733 time ago, but the ref transaction machinery was not prepared to
734 handle many refs, primarily because it tried to keep the loose ref
735 files open at the same time, which would cause us to hit the ENFILE
736 limit. Fix the ref_transaction_commit() code not to keep the file
737 descriptors open unnecessarily.
739 We would want to apply the fix to older codebase, but this fix is
740 unfortunately done to apply only on v2.4.x track.
742 Expecting a rebase to older codebase.
745 * bc/connect-plink (2015-04-28) 3 commits
746 - connect: improve check for plink to reduce false positives
747 - t5601: fix quotation error leading to skipped tests
748 - connect: simplify SSH connection code path
750 The connection initiation code for "ssh" transport tried to absorb
751 differences between the stock "ssh" and Putty-supplied "plink" and
752 its derivatives, but the logic to tell that we are using "plink"
753 variants were too loose and falsely triggered when "plink" appeared
754 anywhere in the path (e.g. "/home/me/bin/uplink/ssh").
756 Will merge to 'next'.
759 * jc/test-prereq-validate (2015-04-28) 1 commit
760 - test: validate prerequistes syntax
762 Help us to find broken test script that splits the body part of the
763 test by mistaken use of wrong kind of quotes.
765 Will merge to 'next'.
768 * fg/document-commit-message-stripping (2015-04-27) 1 commit
769 - Documentation: clarify how "git commit" cleans up the edited log message
771 Will merge to 'next'.
774 * ah/usage-strings (2015-05-03) 2 commits
775 - branch: fix funny-sounding error message
776 - blame, log: format usage strings similarly to those in documentation
778 A few usage string updates. The tip one still needs work.
781 * jk/reading-packed-refs (2015-04-16) 9 commits
782 (merged to 'next' on 2015-04-21 at c9bce73)
783 + t1430: add another refs-escape test
784 + read_packed_refs: avoid double-checking sane refs
785 + strbuf_getwholeline: use getdelim if it is available
786 + strbuf_getwholeline: avoid calling strbuf_grow
787 + strbuf_addch: avoid calling strbuf_grow
788 + config: use getc_unlocked when reading from file
789 + strbuf_getwholeline: use getc_unlocked
790 + git-compat-util: add fallbacks for unlocked stdio
791 + strbuf_getwholeline: use getc macro
793 An earlier rewrite to use strbuf_getwholeline() instead of fgets(3)
794 to read packed-refs file revealed that the former is unacceptably
797 Will merge to 'master' in the second batch of post v2.4 cycle.
800 * kn/cat-file-literally (2015-05-04) 5 commits
801 - fixup! sha1_file: support reading from a loose object of unknown type
802 - t1006: add tests for git cat-file --allow-unknown-type
803 - cat-file: teach cat-file a '--allow-unknown-type' option
804 - cat-file: make the options mutually exclusive
805 - sha1_file: support reading from a loose object of unknown type
807 Add the "--allow-unknown-type" option to "cat-file" to allow
808 inspecting loose objects of an experimental or a broken type.
810 This is v10 $gmane/267960; I think with fixup! squashed in the
811 series is in a good shape (the test may fail without Eric's fix
812 to "hash-object --literally" elsewhere, though).
814 Will merge to 'next'.
817 * ld/p4-filetype-detection (2015-04-04) 3 commits
818 (merged to 'next' on 2015-04-08 at da735b4)
819 + git-p4: fix filetype detection on files opened exclusively
820 + git-p4: small fix for locked-file-move-test
821 + git-p4: fix small bug in locked test scripts
823 Will merge to 'master' in the first batch of post v2.4 cycle.
826 * lm/squelch-bg-progress (2015-04-15) 2 commits
827 (merged to 'next' on 2015-04-21 at 9a8b1e9)
828 + compat/mingw: stubs for getpgid() and tcgetpgrp()
829 + progress: no progress in background
831 Many long-running operations show progress eye-candy, even when
832 they are later backgrounded. Hide the eye-candy when the process
833 is sent to the background instead.
835 Will merge to 'master' in the second batch of post v2.4 cycle.
838 * sb/test-bitmap-free-at-end (2015-04-12) 1 commit
839 (merged to 'next' on 2015-04-14 at 0ae4759)
840 + pack-bitmap.c: fix a memleak
842 Will merge to 'master' in the first batch of post v2.4 cycle.
845 * va/p4-client-path (2015-04-23) 2 commits
846 (merged to 'next' on 2015-04-23 at e0d2065)
847 + git-p4: improve client path detection when branches are used
848 + t9801: check git-p4's branch detection with client spec enabled
850 git p4 attempts to better handle branches in Perforce.
852 Will merge to 'master' in the third batch of post v2.4 cycle.
855 * jk/sha1-file-reduce-useless-warnings (2015-03-30) 1 commit
856 (merged to 'next' on 2015-04-21 at a8bd995)
857 + sha1_file: squelch "packfile cannot be accessed" warnings
859 Will merge to 'master' in the second batch of post v2.4 cycle.
862 * mh/show-branch-topic (2015-03-31) 1 commit
863 (merged to 'next' on 2015-04-08 at 628ada5)
864 + show-branch: show all local heads when only giving one rev along --topics
866 "git show-branch --topics HEAD" (with no other arguments) did not
867 do anything interesting. Instead, contrast the given revision
868 against all the local branches by default.
870 Will merge to 'master' in the first batch of post v2.4 cycle.
873 * sb/line-log-plug-pairdiff-leak (2015-03-30) 1 commit
874 (merged to 'next' on 2015-04-08 at 6ea8b71)
875 + line-log.c: fix a memleak
877 Will merge to 'master' in the first batch of post v2.4 cycle.
880 * jk/init-core-worktree-at-root (2015-04-02) 1 commit
881 (merged to 'next' on 2015-04-08 at 92a942b)
882 + init: don't set core.worktree when initializing /.git
884 We avoid setting core.worktree when the repository location is the
885 ".git" directory directly at the top level of the working tree, but
886 the code misdetected the case in which the working tree is at the
887 root level of the filesystem (which arguably is a silly thing to
888 do, but still valid).
890 Will merge to 'master' in the first batch of post v2.4 cycle.
893 * ts/checkout-advice-plural (2015-04-02) 1 commit
894 (merged to 'next' on 2015-04-08 at dd5c436)
895 + checkout: call a single commit "it" intead of "them"
897 Will merge to 'master' in the first batch of post v2.4 cycle.
900 * pt/credential-xdg (2015-03-25) 4 commits
901 (merged to 'next' on 2015-03-25 at 765128e)
902 + t0302: "unreadable" test needs POSIXPERM
903 (merged to 'next' on 2015-03-24 at 9a3706e)
904 + t0302: test credential-store support for XDG_CONFIG_HOME
905 + git-credential-store: support XDG_CONFIG_HOME
906 + git-credential-store: support multiple credential files
907 (this branch is used by pt/xdg-config-path.)
909 Tweak the sample "store" backend of the credential helper to honor
910 XDG configuration file locations when specified.
912 As this may see further updates, and also this is not an urgent
913 issue anyway, I'll give it a bit more time for it to simmer.
918 * jc/diff-no-index-d-f (2015-03-26) 2 commits
919 (merged to 'next' on 2015-03-27 at 1f270f9)
920 + diff-no-index: align D/F handling with that of normal Git
921 + diff-no-index: DWIM "diff D F" into "diff D/F F"
923 The usual "git diff" when seeing a file turning into a directory
924 showed a patchset to remove the file and create all files in the
925 directory, but "git diff --no-index" simply refused to work. Also,
926 when asked to compare a file and a directory, imitate POSIX "diff"
927 and compare the file with the file with the same name in the
928 directory, instead of refusing to run.
930 Will merge to 'master' in the first batch of post v2.4 cycle.
933 * nd/diff-i-t-a (2015-03-23) 1 commit
934 (merged to 'next' on 2015-03-24 at a6be89f)
935 + diff-lib.c: adjust position of i-t-a entries in diff
937 After "git add -N", the path appeared in output of "git diff HEAD"
938 and "git diff --cached HEAD", leading "git status" to classify it
939 as "Changes to be committed". Such a path, however, is not yet to
940 be scheduled to be committed. "git diff" showed the change to the
941 path as modification, not as a "new file", in the header of its
944 Treat such paths as "yet to be added to the index but Git already
945 know about them"; "git diff HEAD" and "git diff --cached HEAD"
946 should not talk about them, and "git diff" should show them as new
947 files yet to be added to the index.
949 Will cook in 'next', as this brings in a new world order.
952 * bc/object-id (2015-03-13) 10 commits
953 (merged to 'next' on 2015-03-24 at 3ec4f83)
954 + apply: convert threeway_stage to object_id
955 + patch-id: convert to use struct object_id
956 + commit: convert parts to struct object_id
957 + diff: convert struct combine_diff_path to object_id
958 + bulk-checkin.c: convert to use struct object_id
959 + zip: use GIT_SHA1_HEXSZ for trailers
960 + archive.c: convert to use struct object_id
961 + bisect.c: convert leaf functions to use struct object_id
962 + define utility functions for object IDs
963 + define a structure for object IDs
965 Identify parts of the code that knows that we use SHA-1 hash to
966 name our objects too much, and use (1) symbolic constants instead
967 of hardcoded 20 as byte count and/or (2) use struct object_id
968 instead of unsigned char [20] for object names.
970 Will merge to 'master' in the first batch of post v2.4 cycle.
973 * nd/slim-index-pack-memory-usage (2015-04-18) 2 commits
974 (merged to 'next' on 2015-04-21 at 919b80d)
975 + index-pack: kill union delta_base to save memory
976 + index-pack: reduce object_entry size to save memory
978 Memory usage of "git index-pack" has been trimmed by tens of
981 Will merge to 'master' in the second batch of post v2.4 cycle.
984 * nd/multiple-work-trees (2015-03-31) 41 commits
985 (merged to 'next' on 2015-04-02 at 7ea51b4)
986 + prune --worktrees: fix expire vs worktree existence condition
987 (merged to 'next' on 2015-03-24 at 58b5a60)
988 + t1501: fix test with split index
989 (merged to 'next' on 2015-03-20 at cc98ed0)
990 + t2026: fix broken &&-chain
991 (merged to 'next' on 2015-02-18 at b51f696)
992 + t2026 needs procondition SANITY
993 + git-checkout.txt: a note about multiple checkout support for submodules
994 + checkout: add --ignore-other-wortrees
995 + checkout: pass whole struct to parse_branchname_arg instead of individual flags
996 + git-common-dir: make "modules/" per-working-directory directory
997 + checkout: do not fail if target is an empty directory
998 + t2025: add a test to make sure grafts is working from a linked checkout
999 + checkout: don't require a work tree when checking out into a new one
1000 + git_path(): keep "info/sparse-checkout" per work-tree
1001 + count-objects: report unused files in $GIT_DIR/worktrees/...
1002 + gc: support prune --worktrees
1003 + gc: factor out gc.pruneexpire parsing code
1004 + gc: style change -- no SP before closing parenthesis
1005 + checkout: clean up half-prepared directories in --to mode
1006 + checkout: reject if the branch is already checked out elsewhere
1007 + prune: strategies for linked checkouts
1008 + checkout: support checking out into a new working directory
1009 + use new wrapper write_file() for simple file writing
1010 + wrapper.c: wrapper to open a file, fprintf then close
1011 + setup.c: support multi-checkout repo setup
1012 + setup.c: detect $GIT_COMMON_DIR check_repository_format_gently()
1013 + setup.c: convert check_repository_format_gently to use strbuf
1014 + setup.c: detect $GIT_COMMON_DIR in is_git_directory()
1015 + setup.c: convert is_git_directory() to use strbuf
1016 + git-stash: avoid hardcoding $GIT_DIR/logs/....
1017 + *.sh: avoid hardcoding $GIT_DIR/hooks/...
1018 + git-sh-setup.sh: use rev-parse --git-path to get $GIT_DIR/objects
1019 + $GIT_COMMON_DIR: a new environment variable
1020 + commit: use SEQ_DIR instead of hardcoding "sequencer"
1021 + fast-import: use git_path() for accessing .git dir instead of get_git_dir()
1022 + reflog: avoid constructing .lock path with git_path
1023 + *.sh: respect $GIT_INDEX_FILE
1024 + git_path(): be aware of file relocation in $GIT_DIR
1025 + path.c: group git_path(), git_pathdup() and strbuf_git_path() together
1026 + path.c: rename vsnpath() to do_git_path()
1027 + git_snpath(): retire and replace with strbuf_git_path()
1028 + path.c: make get_pathname() call sites return const char *
1029 + path.c: make get_pathname() return strbuf instead of static buffer
1031 A replacement for contrib/workdir/git-new-workdir that does not
1032 rely on symbolic links and make sharing of objects and refs safer
1033 by making the borrowee and borrowers aware of each other.
1035 Will cook in 'next'.
1037 --------------------------------------------------
1040 * as/userdiff-sh (2015-03-13) 1 commit
1041 . userdiff: funcname and word patterns for sh
1043 Add a built-in "userdiff" patterns to word-split and identify
1044 notable lines in shell scripts to help presentation of diff and
1048 * sb/ref-lock-avoid-running-out-of-fds (2015-04-23) 5 commits
1049 . refs.c: enable large transactions
1050 . sha1_file.c: move get_max_fd_limit(void) to wrapper.c
1051 - refs.c: remove lock_fd from struct ref_lock
1052 - t7004: rename ULIMIT test prerequisite to ULIMIT_STACK_SIZE
1053 - update-ref: test handling large transactions properly
1056 * jc/merge-deprecate-old-syntax (2015-03-26) 1 commit
1057 . merge: deprecate 'git merge <message> HEAD <commit>' syntax
1059 The first step to really start the process of removing the ancient
1060 syntax to invoke a two-way merge, which has been deprecated since
1063 This regresses "git pull" in various ways; do not merge ($gmane/267432).