0b3e657f53 · matheusbsbs/matheus-bagliano

@@ -1,10 +1,10 @@

1 To: git@vger.kernel.org

2 Bcc: lwn@lwn.net

3 -Subject: What's cooking in git.git (Oct 2016, #08; Fri, 28)

4 -X-master-at: 7805bda2ac68c659b0042f45473723d9fdcece74

5 -X-next-at: 86d95836a33d12ccd9eed86457166d0222f77f17

3 +Subject: What's cooking in git.git (Oct 2016, #09; Mon, 31)

4 +X-master-at: 1fe8f2cf461179c41f64efbd1dc0a9fb3b7a0fb1

5 +X-next-at: 2c8536d0921902ede8654c83a512865ff5ecd153

6

7 -What's cooking in git.git (Oct 2016, #08; Fri, 28)

7 +What's cooking in git.git (Oct 2016, #09; Mon, 31)

8 --------------------------------------------------

9

10 Here are the topics that have been cooking. Commits prefixed with

@@ -13,8 +13,10 @@ Here are the topics that have been cooking. Commits prefixed with

13 the integration branches, but I am still holding onto them.

14

15 Git v2.10.2, the second maintenance release for 2.10.x track, has

16 -been tagged. On the master front, things are getting close for the

17 -preview snapshot v2.11.0-rc0.

16 +been tagged. On the master front, an early preview v2.11.0-rc0 has

17 +been tagged. An updated, slightly slipped from the original,

18 +schedule is found at http://tinyurl.com/gitCal and I am hoping that

19 +we can conclude this cycle before US Thanksgiving.

20

21 You can find the changes described here in the integration branches

22 of the repositories listed at

@@ -24,141 +26,33 @@ of the repositories listed at

26 --------------------------------------------------

27 [Graduated to "master"]

28

27 -* jc/abbrev-auto (2016-10-22) 4 commits

28 - (merged to 'next' on 2016-10-26 at 92fdb66807)

29 - + transport: compute summary-width dynamically

30 - + transport: allow summary-width to be computed dynamically

31 - + fetch: pass summary_width down the callchain

32 - + transport: pass summary_width down the callchain

33 - (this branch uses jk/abbrev-auto and lt/abbrev-auto.)

34 -

35 - "git push" and "git fetch" reports from what old object to what new

36 - object each ref was updated, using abbreviated refnames, and they

37 - attempt to align the columns for this and other pieces of

38 - information. The way these codepaths compute how many display

39 - columns to allocate for the object names portion of this output has

40 - been updated to match the recent "auto scale the default

41 - abbreviation length" change.

42 -

43 -

44 -* jk/abbrev-auto (2016-10-03) 1 commit

45 - (merged to 'next' on 2016-10-21 at 8aa3d760d8)

46 - + find_unique_abbrev: move logic out of get_short_sha1()

47 - (this branch is used by jc/abbrev-auto; uses lt/abbrev-auto.)

48 -

49 - Updates the way approximate count of total objects is computed

50 - while attempting to come up with a unique abbreviated object name,

51 - which in turn needs to estimate how many hexdigits are necessary to

52 - ensure uniqueness.

53 -

54 -

55 -* jk/daemon-path-ok-check-truncation (2016-10-24) 1 commit

56 - (merged to 'next' on 2016-10-26 at 70c08241f6)

57 - + daemon: detect and reject too-long paths

58 -

59 - "git daemon" used fixed-length buffers to turn URL to the

60 - repository the client asked for into the server side directory

61 - path, using snprintf() to avoid overflowing these buffers, but

62 - allowed possibly truncated paths to the directory. This has been

63 - tightened to reject such a request that causes overlong path to be

64 - required to serve.

65 -

66 -

67 -* jk/no-looking-at-dotgit-outside-repo (2016-10-26) 6 commits

68 - (merged to 'next' on 2016-10-26 at 4aa877b578)

69 - + diff: handle sha1 abbreviations outside of repository

70 - + diff_aligned_abbrev: use "struct oid"

71 - + diff_unique_abbrev: rename to diff_aligned_abbrev

72 - + find_unique_abbrev: use 4-buffer ring

73 - + test-*-cache-tree: setup git dir

74 - + read info/{attributes,exclude} only when in repository

75 - (this branch is used by jk/no-looking-at-dotgit-outside-repo-final.)

76 -

77 - Update "git diff --no-index" codepath not to try to peek into .git/

78 - directory that happens to be under the current directory, when we

79 - know we are operating outside any repository.

80 -

81 -

82 -* js/prepare-sequencer (2016-10-21) 27 commits

83 - (merged to 'next' on 2016-10-26 at 12be8ebe90)

84 - + sequencer: mark all error messages for translation

85 - + sequencer: start error messages consistently with lower case

86 - + sequencer: quote filenames in error messages

87 - + sequencer: mark action_name() for translation

88 - + sequencer: remove overzealous assumption in rebase -i mode

89 - + sequencer: teach write_message() to append an optional LF

90 - + sequencer: refactor write_message() to take a pointer/length

91 - + sequencer: roll back lock file if write_message() failed

92 - + sequencer: stop releasing the strbuf in write_message()

93 - + sequencer: left-trim lines read from the script

94 - + sequencer: support cleaning up commit messages

95 - + sequencer: support amending commits

96 - + sequencer: allow editing the commit message on a case-by-case basis

97 - + sequencer: introduce a helper to read files written by scripts

98 - + sequencer: prepare for rebase -i's commit functionality

99 - + sequencer: remember the onelines when parsing the todo file

100 - + sequencer: get rid of the subcommand field

101 - + sequencer: avoid completely different messages for different actions

102 - + sequencer: strip CR from the todo script

103 - + sequencer: completely revamp the "todo" script parsing

104 - + sequencer: refactor the code to obtain a short commit name

105 - + sequencer: future-proof read_populate_todo()

106 - + sequencer: plug memory leaks for the option values

107 - + sequencer: future-proof remove_sequencer_state()

108 - + sequencer: avoid unnecessary indirection

109 - + sequencer: use memoized sequencer directory path

110 - + sequencer: use static initializers for replay_opts

111 -

112 - Update of the sequencer codebase to make it reusable to reimplement

113 - "rebase -i" continues.

114 -

115 -

116 -* lt/abbrev-auto (2016-10-03) 3 commits

117 - (merged to 'next' on 2016-10-03 at bb188d00f7)

118 - + abbrev: auto size the default abbreviation

119 - + abbrev: prepare for new world order

120 - + abbrev: add FALLBACK_DEFAULT_ABBREV to prepare for auto sizing

121 - (this branch is used by jc/abbrev-auto and jk/abbrev-auto.)

122 -

123 - Allow the default abbreviation length, which has historically been

124 - 7, to scale as the repository grows. The logic suggests to use 12

125 - hexdigits for the Linux kernel, and 9 to 10 for Git itself.

126 -

127 -

128 -* nd/ita-empty-commit (2016-10-24) 4 commits

129 - (merged to 'next' on 2016-10-26 at fb007cdae1)

130 - + commit: don't be fooled by ita entries when creating initial commit

131 - + commit: fix empty commit creation when there's no changes but ita entries

132 - + diff: add --ita-[in]visible-in-index

133 - + diff-lib: allow ita entries treated as "not yet exist in index"

134 -

135 - When new paths were added by "git add -N" to the index, it was

136 - enough to circumvent the check by "git commit" to refrain from

137 - making an empty commit without "--allow-empty". The same logic

138 - prevented "git status" to show such a path as "new file" in the

139 - "Changes not staged for commit" section.

140 -

141 -

142 -* rs/ring-buffer-wraparound (2016-10-26) 1 commit

143 - (merged to 'next' on 2016-10-26 at d2da68a14a)

144 - + hex: make wraparound of the index into ring-buffer explicit

145 -

146 - The code that we have used for the past 10+ years to cycle

147 - 4-element ring buffers turns out to be not quite portable in

148 - theoretical world.

149 -

150 -

151 -* sb/submodule-ignore-trailing-slash (2016-10-25) 3 commits

152 - (merged to 'next' on 2016-10-26 at e56a8ebb38)

153 - + t0060: sidestep surprising path mangling results on Windows

154 - (merged to 'next' on 2016-10-11 at e37425ed17)

155 - + submodule: ignore trailing slash in relative url

156 - + submodule: ignore trailing slash on superproject URL

157 -

158 - A minor regression fix for "git submodule".

29 +* ak/pre-receive-hook-template-modefix (2016-10-28) 1 commit

30 + (merged to 'next' on 2016-10-28 at 86d95836a3)

31 + + pre-receive.sample: mark it executable

32 +

33 + A trivial clean-up to a recently graduated topic.

34 +

35 +

36 +* ak/sh-setup-dot-source-i18n-fix (2016-10-30) 1 commit

37 + (merged to 'next' on 2016-10-31 at 538a72700e)

38 + + git-sh-setup: be explicit where to dot-source git-sh-i18n from.

39 +

40 + Recent update to git-sh-setup (a library of shell functions that

41 + are used by our in-tree scripted Porcelain commands) included

42 + another shell library git-sh-i18n without specifying where it is,

43 + relying on the $PATH. This has been fixed to be more explicit by

44 + prefixing $(git --exec-path) output in front.

45 +

46 +

47 +* aw/numbered-stash (2016-10-26) 1 commit

48 + (merged to 'next' on 2016-10-26 at 8d9325fa3a)

49 + + stash: allow stashes to be referenced by index only

50 +

51 + The user always has to say "stash@{$N}" when naming a single

52 + element in the default location of the stash, i.e. reflogs in

53 + refs/stash. The "git stash" command learned to accept "git stash

54 + apply 4" as a short-hand for "git stash apply stash@{4}".

55

160 ---------------------------------------------------

161 -[New Topics]

56

57 * jk/common-main (2016-10-27) 1 commit

58 (merged to 'next' on 2016-10-28 at fcdd4f8a26)

@@ -166,35 +60,117 @@ of the repositories listed at

60

61 A trivial clean-up to a recently graduated topic.

62

169 - Will merge to 'master'.

63

64 +* jk/rebase-config-insn-fmt-docfix (2016-10-30) 1 commit

65 + (merged to 'next' on 2016-10-31 at d48e20ffa2)

66 + + doc: fix missing "::" in config list

67

172 -* ak/pre-receive-hook-template-modefix (2016-10-28) 1 commit

173 - (merged to 'next' on 2016-10-28 at 86d95836a3)

174 - + pre-receive.sample: mark it executable

68 + Documentation fix.

69

176 - A trivial clean-up to a recently graduated topic.

70

178 - Will merge to 'master'.

71 +* jt/trailer-with-cruft (2016-10-21) 8 commits

72 + (merged to 'next' on 2016-10-27 at b5d1a21811)

73 + + trailer: support values folded to multiple lines

74 + + trailer: forbid leading whitespace in trailers

75 + + trailer: allow non-trailers in trailer block

76 + + trailer: clarify failure modes in parse_trailer

77 + + trailer: make args have their own struct

78 + + trailer: streamline trailer item create and add

79 + + trailer: use list.h for doubly-linked list

80 + + trailer: improve const correctness

81 + (this branch is used by jt/use-trailer-api-in-commands.)

82

83 + Update "interpret-trailers" machinery and teaches it that people in

84 + real world write all sorts of crufts in the "trailer" that was

85 + originally designed to have the neat-o "Mail-Header: like thing"

86 + and nothing else.

87

181 -* nd/rebase-forget (2016-10-28) 1 commit

182 - - rebase: add --forget to cleanup rebase, leave HEAD untouched

88

184 - "git rebase" learned "--forget" option, which allows a user to

185 - remove the metadata left by an earlier "git rebase" that was

186 - manually aborted without using "git rebase --abort".

89 +* ls/filter-process (2016-10-17) 14 commits

90 + (merged to 'next' on 2016-10-19 at ffd0de042c)

91 + + contrib/long-running-filter: add long running filter example

92 + + convert: add filter.<driver>.process option

93 + + convert: prepare filter.<driver>.process option

94 + + convert: make apply_filter() adhere to standard Git error handling

95 + + pkt-line: add functions to read/write flush terminated packet streams

96 + + pkt-line: add packet_write_gently()

97 + + pkt-line: add packet_flush_gently()

98 + + pkt-line: add packet_write_fmt_gently()

99 + + pkt-line: extract set_packet_header()

100 + + pkt-line: rename packet_write() to packet_write_fmt()

101 + + run-command: add clean_on_exit_handler

102 + + run-command: move check_pipe() from write_or_die to run_command

103 + + convert: modernize tests

104 + + convert: quote filter names in error messages

105

106 + The smudge/clean filter API expect an external process is spawned

107 + to filter the contents for each path that has a filter defined. A

108 + new type of "process" filter API has been added to allow the first

109 + request to run the filter for a path to spawn a single process, and

110 + all filtering need is served by this single process for multiple

111 + paths, reducing the process creation overhead.

112

189 -* jc/git-open-cloexec (2016-10-28) 2 commits

190 - - sha1_file: stop opening files with O_NOATIME

191 - - git_open(): untangle possible NOATIME and CLOEXEC interactions

192 - (this branch uses ls/git-open-cloexec.)

113

194 - The codeflow of setting NOATIME and CLOEXEC on file descriptors Git

195 - opens has been simplified.

114 +* ls/git-open-cloexec (2016-10-25) 3 commits

115 + (merged to 'next' on 2016-10-26 at f7259cbddb)

116 + + read-cache: make sure file handles are not inherited by child processes

117 + + sha1_file: open window into packfiles with O_CLOEXEC

118 + + sha1_file: rename git_open_noatime() to git_open()

119 + (this branch is used by jc/git-open-cloexec.)

120 +

121 + Git generally does not explicitly close file descriptors that were

122 + open in the parent process when spawning a child process, but most

123 + of the time the child does not want to access them. As Windows does

124 + not allow removing or renaming a file that has a file descriptor

125 + open, a slow-to-exit child can even break the parent process by

126 + holding onto them. Use O_CLOEXEC flag to open files in various

127 + codepaths.

128 +

129 +

130 +* nd/test-helpers (2016-10-27) 1 commit

131 + (merged to 'next' on 2016-10-31 at 9780fe6d90)

132 + + valgrind: support test helpers

133 +

134 + Update to the test framework made in 2.9 timeframe broke running

135 + the tests under valgrind, which has been fixed.

136 +

137 +

138 +* rs/commit-pptr-simplify (2016-10-30) 1 commit

139 + (merged to 'next' on 2016-10-31 at ddeed2e66a)

140 + + commit: simplify building parents list

141 +

142 + Code simplification.

143 +

144 +

145 +* sc/fmt-merge-msg-doc-markup-fix (2016-10-28) 1 commit

146 + (merged to 'next' on 2016-10-31 at 198c259bb2)

147 + + Documentation/fmt-merge-msg: fix markup in example

148

197 - We probably want to drop the tip one.

149 + Documentation fix.

150 +

151 +--------------------------------------------------

152 +[New Topics]

153 +

154 +* jc/push-default-explicit (2016-10-28) 1 commit

155 + - push: do not use potentially ambiguous default refspec

156 +

157 + A lazy "git push" without refspec did not internally use a fully

158 + specified refspec to perform 'current', 'simple', or 'upstream'

159 + push, causing unnecessary "ambiguous ref" errors.

160 +

161 + Needs updates to the tests.

162 +

163 +

164 +* jt/use-trailer-api-in-commands (2016-10-28) 4 commits

165 + - sequencer: use trailer's trailer layout

166 + - trailer: have function to describe trailer layout

167 + - trailer: avoid unnecessary splitting on lines

168 + - commit: make ignore_non_trailer take buf/len

169 +

170 + Commands that operate on a log message and add lines to the trailer

171 + blocks, such as "format-patch -s", "cherry-pick (-x|-s)", and

172 + "commit -s", have been taught to use the logic of and share the

173 + code with "git interpret-trailer".

174

175 --------------------------------------------------

176 [Stalled]

@@ -337,16 +313,25 @@ of the repositories listed at

313 --------------------------------------------------

314 [Cooking]

315

340 -* aw/numbered-stash (2016-10-26) 1 commit

341 - (merged to 'next' on 2016-10-26 at 8d9325fa3a)

342 - + stash: allow stashes to be referenced by index only

316 +* nd/rebase-forget (2016-10-28) 1 commit

317 + - rebase: add --forget to cleanup rebase, leave HEAD untouched

318

344 - The user always has to say "stash@{$N}" when naming a single

345 - element in the default location of the stash, i.e. reflogs in

346 - refs/stash. The "git stash" command learned to accept "git stash

347 - apply 4" as a short-hand for "git stash apply stash@{4}".

319 + "git rebase" learned "--forget" option, which allows a user to

320 + remove the metadata left by an earlier "git rebase" that was

321 + manually aborted without using "git rebase --abort".

322 +

323 + Waiting for a reroll.

324

349 - Will merge to 'master'.

325 +

326 +* jc/git-open-cloexec (2016-10-31) 3 commits

327 + - sha1_file: stop opening files with O_NOATIME

328 + - git_open_cloexec(): use fcntl(2) w/ FD_CLOEXEC fallback

329 + - git_open(): untangle possible NOATIME and CLOEXEC interactions

330 +

331 + The codeflow of setting NOATIME and CLOEXEC on file descriptors Git

332 + opens has been simplified.

333 +

334 + We may want to drop the tip one.

335

336

337 * jk/no-looking-at-dotgit-outside-repo-final (2016-10-26) 1 commit

@@ -370,24 +355,6 @@ of the repositories listed at

355 "--unmerge" option to recover from this situation.

356

357

373 -* ls/git-open-cloexec (2016-10-25) 3 commits

374 - (merged to 'next' on 2016-10-26 at f7259cbddb)

375 - + read-cache: make sure file handles are not inherited by child processes

376 - + sha1_file: open window into packfiles with O_CLOEXEC

377 - + sha1_file: rename git_open_noatime() to git_open()

378 - (this branch is used by jc/git-open-cloexec.)

379 -

380 - Git generally does not explicitly close file descriptors that were

381 - open in the parent process when spawning a child process, but most

382 - of the time the child does not want to access them. As Windows does

383 - not allow removing or renaming a file that has a file descriptor

384 - open, a slow-to-exit child can even break the parent process by

385 - holding onto them. Use O_CLOEXEC flag to open files in various

386 - codepaths.

387 -

388 - Will merge to 'master'.

389 -

390 -

358 * jc/merge-base-fp-only (2016-10-19) 8 commits

359 . merge-base: fp experiment

360 - merge: allow to use only the fp-only merge bases

@@ -419,25 +386,6 @@ of the repositories listed at

386 cf. <xmqqd1il5w4e.fsf@gitster.mtv.corp.google.com>

387

388

422 -* jt/trailer-with-cruft (2016-10-21) 8 commits

423 - (merged to 'next' on 2016-10-27 at b5d1a21811)

424 - + trailer: support values folded to multiple lines

425 - + trailer: forbid leading whitespace in trailers

426 - + trailer: allow non-trailers in trailer block

427 - + trailer: clarify failure modes in parse_trailer

428 - + trailer: make args have their own struct

429 - + trailer: streamline trailer item create and add

430 - + trailer: use list.h for doubly-linked list

431 - + trailer: improve const correctness

432 -

433 - Update "interpret-trailers" machinery and teaches it that people in

434 - real world write all sorts of crufts in the "trailer" that was

435 - originally designed to have the neat-o "Mail-Header: like thing"

436 - and nothing else.

437 -

438 - Will merge to 'master'.

439 -

440 -

389 * pb/bisect (2016-10-18) 27 commits

390 - bisect--helper: remove the dequote in bisect_start()

391 - bisect--helper: retire `--bisect-auto-next` subcommand

@@ -569,33 +517,6 @@ of the repositories listed at

517 Will hold to see if people scream.

518

519

572 -* ls/filter-process (2016-10-17) 14 commits

573 - (merged to 'next' on 2016-10-19 at ffd0de042c)

574 - + contrib/long-running-filter: add long running filter example

575 - + convert: add filter.<driver>.process option

576 - + convert: prepare filter.<driver>.process option

577 - + convert: make apply_filter() adhere to standard Git error handling

578 - + pkt-line: add functions to read/write flush terminated packet streams

579 - + pkt-line: add packet_write_gently()

580 - + pkt-line: add packet_flush_gently()

581 - + pkt-line: add packet_write_fmt_gently()

582 - + pkt-line: extract set_packet_header()

583 - + pkt-line: rename packet_write() to packet_write_fmt()

584 - + run-command: add clean_on_exit_handler

585 - + run-command: move check_pipe() from write_or_die to run_command

586 - + convert: modernize tests

587 - + convert: quote filter names in error messages

588 -

589 - The smudge/clean filter API expect an external process is spawned

590 - to filter the contents for each path that has a filter defined. A

591 - new type of "process" filter API has been added to allow the first

592 - request to run the filter for a path to spawn a single process, and

593 - all filtering need is served by this single process for multiple

594 - paths, reducing the process creation overhead.

595 -

596 - Will merge to 'master'.

597 -

598 -

520 * sg/fix-versioncmp-with-common-suffix (2016-09-08) 5 commits

521 - versioncmp: cope with common leading parts in versionsort.prereleaseSuffix

522 - versioncmp: pass full tagnames to swap_prereleases()

What's cooking (2016/10 #09)