5a10c05e99 · matheusbsbs/matheus-bagliano

@@ -1,16 +1,18 @@

1 To: git@vger.kernel.org

2 Bcc: lwn@lwn.net

3 -Subject: What's cooking in git.git (May 2015, #05; Tue, 19)

4 -X-master-at: 6c1249c503f39ca45ca5b183527b22192cdaf7a7

5 -X-next-at: 3179493ddf99afea85d647dbfa9ebe48a70b9cc0

3 +Subject: What's cooking in git.git (May 2015, #06; Fri, 22)

4 +X-master-at: 9532ead9875b7b92ff2007f9de61af5874e8839a

5 +X-next-at: b70f6470d4a42c2efce07be8bc1edac7c5743a77

6

7 -What's cooking in git.git (May 2015, #05; Tue, 19)

7 +What's cooking in git.git (May 2015, #06; Fri, 22)

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

9

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

11 '-' are only in 'pu' (proposed updates) while commits prefixed with

12 '+' are in 'next'.

13

14 +The fourth batch of topics have been merged to 'master'.

15 +

16 The "untracked cache" series is in 'next', to give it a wider

17 exposure. I do not use it personally, but it is meant to make life

18 easier for those with large amount of untracked cruft in their

@@ -25,169 +27,30 @@ of the repositories listed at

27 --------------------------------------------------

28 [Graduated to "master"]

29

28 -* bc/connect-plink (2015-04-28) 3 commits

29 - (merged to 'next' on 2015-05-05 at 9def2e1)

30 - + connect: improve check for plink to reduce false positives

31 - + t5601: fix quotation error leading to skipped tests

32 - + connect: simplify SSH connection code path

33 -

34 - The connection initiation code for "ssh" transport tried to absorb

35 - differences between the stock "ssh" and Putty-supplied "plink" and

36 - its derivatives, but the logic to tell that we are using "plink"

37 - variants were too loose and falsely triggered when "plink" appeared

38 - anywhere in the path (e.g. "/home/me/bin/uplink/ssh").

39 -

40 -

41 -* fg/document-commit-message-stripping (2015-04-27) 1 commit

42 - (merged to 'next' on 2015-05-05 at 1892a99)

43 - + Documentation: clarify how "git commit" cleans up the edited log message

44 -

45 -

46 -* jc/gitignore-precedence (2015-04-22) 1 commit

47 - (merged to 'next' on 2015-05-05 at 6ef85da)

48 - + ignore: info/exclude should trump core.excludesfile

49 -

50 - core.excludesfile (defaulting to $XDG_HOME/git/ignore) is supposed

51 - to be overridden by repository-specific .git/info/exclude file, but

52 - the order was swapped from the beginning. This belatedly fixes it.

53 -

54 -

55 -* jc/merge (2015-04-29) 15 commits

56 - (merged to 'next' on 2015-05-07 at 1c56512)

57 - + merge: deprecate 'git merge <message> HEAD <commit>' syntax

58 - + merge: handle FETCH_HEAD internally

59 - + merge: decide if we auto-generate the message early in collect_parents()

60 - + merge: make collect_parents() auto-generate the merge message

61 - + merge: extract prepare_merge_message() logic out

62 - + merge: narrow scope of merge_names

63 - + merge: split reduce_parents() out of collect_parents()

64 - + merge: clarify collect_parents() logic

65 - + merge: small leakfix and code simplification

66 - + merge: do not check argc to determine number of remote heads

67 - + merge: clarify "pulling into void" special case

68 - + t5520: test pulling an octopus into an unborn branch

69 - + t5520: style fixes

70 - + merge: simplify code flow

71 - + merge: test the top-level merge driver

72 - (this branch is used by jc/merge-drop-old-syntax and pt/pull-tests.)

73 -

74 - "git merge FETCH_HEAD" learned that the previous "git fetch" could

75 - be to create an Octopus merge, i.e. recording multiple branches

76 - that are not marked as "not-for-merge"; this allows us to lose an

77 - old style invocation "git merge <msg> HEAD $commits..." in the

78 - implementation of "git pull" script; the old style syntax can now

79 - be deprecated.

80 -

81 -

82 -* jc/test-prereq-validate (2015-04-28) 1 commit

83 - (merged to 'next' on 2015-05-05 at a30464c)

84 - + test: validate prerequistes syntax

85 -

86 - Help us to find broken test script that splits the body part of the

87 - test by mistaken use of wrong kind of quotes.

88 -

89 -

90 -* jk/git-no-more-argv0-path-munging (2015-04-22) 1 commit

91 - (merged to 'next' on 2015-05-05 at b3f9a45)

92 - + stop putting argv[0] dirname at front of PATH

93 -

94 - We have prepended $GIT_EXEC_PATH and the path "git" is installed in

95 - (typically "/usr/bin") to $PATH when invoking subprograms and hooks

96 - for almost eternity, but the original use case the latter tried to

97 - support was semi-bogus (i.e. install git to /opt/foo/git and run it

98 - without having /opt/foo on $PATH), and more importantly it has

99 - become less and less relevant as Git grew more mainstream (i.e. the

100 - users would _want_ to have it on their $PATH). Stop prepending the

101 - path in which "git" is installed to users' $PATH, as that would

102 - interfere the command search order people depend on (e.g. they may

103 - not like versions of programs that are unrelated to Git in /usr/bin

104 - and want to override them by having different ones in /usr/local/bin

105 - and have the latter directory earlier in their $PATH).

106 -

107 -

108 -* jk/stash-require-clean-index (2015-04-22) 3 commits

109 - (merged to 'next' on 2015-05-05 at b5f6c32)

110 - + stash: require a clean index to apply

111 - + t3903: avoid applying onto dirty index

112 - + t3903: stop hard-coding commit sha1s

113 -

114 - "git stash pop/apply" forgot to make sure that not just the working

115 - tree is clean but also the index is clean. The latter is important

116 - as a stash application can conflict and the index will be used for

117 - conflict resolution.

118 -

119 -

120 -* jk/test-chain-lint (2015-04-28) 2 commits

121 - (merged to 'next' on 2015-05-05 at e6f0290)

122 - + test-lib: turn on GIT_TEST_CHAIN_LINT by default

123 - + t7502-commit.sh: fix a broken and-chain

124 -

125 - Developer support to automatically detect broken &&-chain in the

126 - test scripts is now turned on by default.

127 -

128 -

129 -* kn/cat-file-literally (2015-05-06) 4 commits

130 - (merged to 'next' on 2015-05-12 at 04b29b5)

131 - + t1006: add tests for git cat-file --allow-unknown-type

132 - + cat-file: teach cat-file a '--allow-unknown-type' option

133 - + cat-file: make the options mutually exclusive

134 - + sha1_file: support reading from a loose object of unknown type

135 -

136 - Add the "--allow-unknown-type" option to "cat-file" to allow

137 - inspecting loose objects of an experimental or a broken type.

138 -

139 -

140 -* nd/diff-i-t-a (2015-03-23) 1 commit

141 - (merged to 'next' on 2015-05-05 at cba9cd9)

142 - + diff-lib.c: adjust position of i-t-a entries in diff

143 -

144 - Originally merged to 'next' on 2015-03-24

145 -

146 - After "git add -N", the path appeared in output of "git diff HEAD"

147 - and "git diff --cached HEAD", leading "git status" to classify it

148 - as "Changes to be committed". Such a path, however, is not yet to

149 - be scheduled to be committed. "git diff" showed the change to the

150 - path as modification, not as a "new file", in the header of its

151 - output.

152 -

153 - Treat such paths as "yet to be added to the index but Git already

154 - know about them"; "git diff HEAD" and "git diff --cached HEAD"

155 - should not talk about them, and "git diff" should show them as new

156 - files yet to be added to the index.

157 -

158 -

159 -* nd/dwim-wildcards-as-pathspecs (2015-05-03) 1 commit

160 - (merged to 'next' on 2015-05-07 at 8176de7)

161 - + pathspec: avoid the need of "--" when wildcard is used

30 +* dl/branch-error-message (2015-05-06) 1 commit

31 + (merged to 'next' on 2015-05-11 at ed947ab)

32 + + branch: do not call a "remote-tracking branch" a "remote branch"

33

163 - A heuristic to help the "git <cmd> <revs> <pathspec>" command line

164 - convention to catch mistyped paths is to make sure all the non-rev

165 - parameters in the later part of the command line are names of the

166 - files in the working tree, but that means "git grep $str -- \*.c"

167 - must always be disambiguated with "--", because nobody sane will

168 - create a file whose name literally is asterisk-dot-see. Loosen the

169 - heuristic to declare that with a wildcard string the user likely

170 - meant to give us a pathspec.

34 + Error messages from "git branch" called remote-tracking branches as

35 + "remote branches".

36

37

173 -* ph/rebase-i-redo (2015-04-29) 1 commit

174 - (merged to 'next' on 2015-05-07 at 8fc1f52)

175 - + rebase -i: redo tasks that die during cherry-pick

38 +* dl/subtree-avoid-tricky-echo (2015-05-08) 1 commit

39 + (merged to 'next' on 2015-05-11 at 36d4f0e)

40 + + contrib/subtree: portability fix for string printing

41

177 - "git rebase -i" moved the "current" command from "todo" to "done" a

178 - bit too prematurely, losing a step when a "pick" did not even start.

42 + "git subtree" script (in contrib/) used "echo -n" to produce

43 + progress messages in a non-portable way.

44

45

181 -* sb/ref-lock-lose-lock-fd (2015-05-10) 1 commit

182 - (merged to 'next' on 2015-05-12 at dde9be1)

183 - + refs.c: remove lock_fd from struct ref_lock

46 +* dl/subtree-push-no-squash (2015-05-07) 1 commit

47 + (merged to 'next' on 2015-05-11 at 74d07ca)

48 + + contrib/subtree: there's no push --squash

49

185 - The refs API uses ref_lock struct which had its own "int fd", even

186 - though the same file descriptor was in the lock struct it contains.

187 - Clean-up the code to lose this redundant field.

50 + "git subtree" script (in contrib/) does not have --squash option

51 + when pushing, but the documentation and help text pretended as if

52 + it did.

53

189 ---------------------------------------------------

190 -[New Topics]

54

55 * ja/tutorial-asciidoctor-fix (2015-05-12) 1 commit

56 (merged to 'next' on 2015-05-19 at f15d940)

@@ -197,33 +60,21 @@ of the repositories listed at

60 delimit it from the rest of the document, which choke GitHub's

61 AsciiDoc renderer.

62

200 - Will merge to 'master'.

201 -

202 -

203 -* jk/stripspace-asciidoctor-fix (2015-05-12) 1 commit

204 - (merged to 'next' on 2015-05-19 at 12f9059)

205 - + doc: fix unmatched code fences in git-stripspace

206 -

207 - A literal block in the tutorial had lines with unequal lengths to

208 - delimit it from the rest of the document, which choke GitHub's

209 - AsciiDoc renderer.

210 -

211 - Will merge to 'master'.

212 -

63

214 -* dt/cat-file-follow-symlinks (2015-05-14) 3 commits

215 - - cat-file: add --follow-symlinks to --batch

216 - - sha1_name: get_sha1_with_context learns to follow symlinks

217 - - tree-walk: learn get_tree_entry_follow_symlinks

64 +* jc/ignore-epipe-in-filter (2015-05-20) 2 commits

65 + (merged to 'next' on 2015-05-20 at 2b14ed7)

66 + + filter_buffer_or_fd(): ignore EPIPE

67 + + copy.c: make copy_fd() report its status silently

68

219 - "git cat-file --batch(-check)" learned the "--follow-symlinks"

220 - option that follows an in-tree symblic links when asked about an

221 - object via extended SHA-1 syntax, e.g. HEAD:RelNotes that points at

222 - Documentation/RelNotes/2.5.0.txt. With the new option, the command

223 - behaves as if HEAD:Documentation/RelNotes/2.5.0.txt was given as

224 - input instead.

69 + Filter scripts were run with SIGPIPE disabled on the Git side,

70 + expecting that they may not read what Git feeds them to filter.

71 + We however treated a filter that does not read its input fully

72 + before exiting as an error.

73

226 - Will merge to 'next'.

74 + This changes semantics, but arguably in a good way. If a filter

75 + can produce its output without consuming its input using whatever

76 + magic, we now let it do so, instead of diagnosing it as a

77 + programming error.

78

79

80 * jk/add-e-kill-editor (2015-05-12) 1 commit

@@ -233,8 +84,6 @@ of the repositories listed at

84 "git add -e" did not allow the user to abort the operation by

85 killing the editor.

86

236 - Will merge to 'master'.

237 -

87

88 * jk/asciidoc-markup-fix (2015-05-14) 9 commits

89 (merged to 'next' on 2015-05-19 at df0c63e)

@@ -252,7 +101,68 @@ of the repositories listed at

101 consistent in general and also make AsciiDoctor (an alternative

102 formatter) happier.

103

255 - Will merge to 'master'.

104 +

105 +* jk/skip-http-tests-under-no-curl (2015-05-07) 2 commits

106 + (merged to 'next' on 2015-05-11 at a52b711)

107 + + tests: skip dav http-push tests under NO_EXPAT=NoThanks

108 + + t/lib-httpd.sh: skip tests if NO_CURL is defined

109 +

110 + Test clean-up.

111 +

112 +

113 +* jk/stripspace-asciidoctor-fix (2015-05-12) 1 commit

114 + (merged to 'next' on 2015-05-19 at 12f9059)

115 + + doc: fix unmatched code fences in git-stripspace

116 +

117 + A literal block in the tutorial had lines with unequal lengths to

118 + delimit it from the rest of the document, which choke GitHub's

119 + AsciiDoc renderer.

120 +

121 +

122 +* lm/squelch-bg-progress (2015-05-19) 1 commit

123 + (merged to 'next' on 2015-05-20 at 60916e6)

124 + + progress: treat "no terminal" as being in the foreground

125 +

126 + The controlling tty-based heuristics to squelch progress output did

127 + not consider that the process may not be talking to a tty at all

128 + (e.g. sending the progress to sideband #2). This is a finishing

129 + touch to a topic that is already in 'master'.

130 +

131 +

132 +* ls/http-ssl-cipher-list (2015-05-08) 1 commit

133 + (merged to 'next' on 2015-05-11 at 55764ce)

134 + + http: add support for specifying an SSL cipher list

135 +

136 + Introduce http.<url>.SSLCipherList configuration variable to tweak

137 + the list of cipher suite to be used with libcURL when talking with

138 + https:// sites.

139 +

140 +

141 +* mg/log-decorate-HEAD (2015-05-13) 2 commits

142 + (merged to 'next' on 2015-05-19 at 009342b)

143 + + log: do not shorten decoration names too early

144 + + log: decorate HEAD with branch name under --decorate=full, too

145 +

146 + The "log --decorate" enhancement in Git 2.4 that shows the commit

147 + at the tip of the current branch e.g. "HEAD -> master", did not

148 + work with --decorate=full.

149 +

150 +

151 +* mh/clone-verbosity-fix (2015-05-19) 1 commit

152 + (merged to 'next' on 2015-05-20 at f613575)

153 + + clone: call transport_set_verbosity before anything else on the newly created transport

154 +

155 + Git 2.4 broke setting verbosity and progress levels on "git clone"

156 + with native transports.

157 +

158 +

159 +* mh/lockfile-retry (2015-05-14) 2 commits

160 + (merged to 'next' on 2015-05-20 at 7ad2e9d)

161 + + lock_packed_refs(): allow retries when acquiring the packed-refs lock

162 + + lockfile: allow file locking to be retried with a timeout

163 +

164 + Instead of dying immediately upon failing to obtain a lock, retry

165 + after a short while with backoff.

166

167

168 * mh/ref-directory-file (2015-05-11) 18 commits

@@ -280,196 +190,218 @@ of the repositories listed at

190 removed at the same time as 'refs/heads/xyzzy' is added (or vice

191 versa) very well.

192

283 - Will merge to 'master'.

284 -

193

286 -* dt/clean-pathspec-filter-then-lstat (2015-05-18) 1 commit

287 - - clean: only lstat files in pathspec

194 +* mh/write-refs-sooner-2.2 (2015-05-12) 8 commits

195 + + ref_transaction_commit(): fix atomicity and avoid fd exhaustion

196 + + ref_transaction_commit(): remove the local flags variable

197 + + ref_transaction_commit(): inline call to write_ref_sha1()

198 + + rename_ref(): inline calls to write_ref_sha1() from this function

199 + + commit_ref_update(): new function, extracted from write_ref_sha1()

200 + + write_ref_to_lockfile(): new function, extracted from write_ref_sha1()

201 + + t7004: rename ULIMIT test prerequisite to ULIMIT_STACK_SIZE

202 + + update-ref: test handling large transactions properly

203 + (this branch is used by mh/write-refs-sooner-2.3 and mh/write-refs-sooner-2.4.)

204

289 - "git clean pathspec..." tried to lstat(2) and complain even for

290 - paths outside the given pathspec.

205 + Multi-ref transaction support we merged a few releases ago

206 + unnecessarily kept many file descriptors open, risking to fail with

207 + resource exhaustion.

208

292 - Will merge to 'next'.

209

210 +* mh/write-refs-sooner-2.3 (2015-05-12) 1 commit

211 + + Merge branch 'mh/write-refs-sooner-2.2' into mh/write-refs-sooner-2.3

212 + (this branch is used by mh/write-refs-sooner-2.4; uses mh/write-refs-sooner-2.2.)

213

295 -* jh/filter-empty-contents (2015-05-18) 1 commit

296 - - sha1_file: pass empty buffer to index empty file

214 + Multi-ref transaction support we merged a few releases ago

215 + unnecessarily kept many file descriptors open, risking to fail with

216 + resource exhaustion. This is for 2.3.x track.

217

298 - The clean/smudge interface did not work well when filtering an

299 - empty contents (failed and then passed the empty input through).

300 - It can be argued that a filter that produces anything but empty for

301 - an empty input is nonsense, but if the user wants to do strange

302 - things, then why not?

218

304 - Will merge to 'next'.

219 +* mh/write-refs-sooner-2.4 (2015-05-12) 9 commits

220 + (merged to 'next' on 2015-05-19 at f7be074)

221 + + Merge branch 'mh/write-refs-sooner-2.3' into mh/write-refs-sooner-2.4

222 + + ref_transaction_commit(): fix atomicity and avoid fd exhaustion

223 + + ref_transaction_commit(): remove the local flags variable

224 + + ref_transaction_commit(): inline call to write_ref_sha1()

225 + + rename_ref(): inline calls to write_ref_sha1() from this function

226 + + commit_ref_update(): new function, extracted from write_ref_sha1()

227 + + write_ref_to_lockfile(): new function, extracted from write_ref_sha1()

228 + + t7004: rename ULIMIT test prerequisite to ULIMIT_STACK_SIZE

229 + + update-ref: test handling large transactions properly

230 + (this branch uses mh/write-refs-sooner-2.2 and mh/write-refs-sooner-2.3.)

231

232 + Multi-ref transaction support we merged a few releases ago

233 + unnecessarily kept many file descriptors open, risking to fail with

234 + resource exhaustion. This is for 2.4.x track.

235

307 -* jk/http-backend-deadlock (2015-05-15) 2 commits

308 - - http-backend: spool ref negotiation requests to buffer

309 - - http-backend: fix die recursion with custom handler

236

311 - Communication between the HTTP server and http_backend process can

312 - lead to a dead-lock when relaying a large ref negotiation request.

313 - Diagnose the situation better, and mitigate it by reading such a

314 - request first into core (to a reasonable limit).

237 +* ps/bundle-verify-arg (2015-05-08) 1 commit

238 + (merged to 'next' on 2015-05-11 at 9f1b1ae)

239 + + bundle: verify arguments more strictly

240

316 - Will merge to 'next'.

241 + "git bundle verify" did not diagnose extra parameters on the

242 + command line.

243

244

319 -* jk/rerere-forget-check-enabled (2015-05-14) 1 commit

320 - (merged to 'next' on 2015-05-19 at bfe67dc)

321 - + rerere: exit silently on "forget" when rerere is disabled

245 +* ps/doc-packfile-vs-pack-file (2015-05-17) 3 commits

246 + (merged to 'next' on 2015-05-20 at 7302174)

247 + + pack-protocol.txt: fix insconsistent spelling of "packfile"

248 + + git-unpack-objects.txt: fix inconsistent spelling of "packfile"

249 + + git-verify-pack.txt: fix inconsistent spelling of "packfile"

250

323 - "git rerere forget" in a repository without rerere enabled gave a

324 - cryptic error message; it should be a silent no-op instead.

251 + Doc consistency updates.

252

326 - Will merge to 'master'.

253

254 +* pt/pull-tags-error-diag (2015-05-14) 1 commit

255 + (merged to 'next' on 2015-05-20 at 6e1a5f1)

256 + + pull: remove --tags error in no merge candidates case

257

329 -* lm/squelch-bg-progress (2015-05-19) 1 commit

330 - - progress: treat "no terminal" as being in the foreground

258 + There was a dead code that used to handle "git pull --tags" and

259 + show special-cased error message, which was made irrelevant when

260 + the semantics of the option changed back in Git 1.9 days.

261

332 - The controlling tty-based heuristics to squelch progress output did

333 - not consider that the process may not be talking to a tty at all

334 - (e.g. sending the progress to sideband #2). This is a finishing

335 - touch to a topic that is already in 'master'.

262

337 - Will merge to 'next'.

263 +* sb/t1020-cleanup (2015-05-18) 1 commit

264 + (merged to 'next' on 2015-05-20 at fa74b14)

265 + + subdirectory tests: code cleanup, uncomment test

266

267 + There was a commented-out (instead of being marked to expect

268 + failure) test that documented a breakage that was fixed since the

269 + test was written; turn it into a proper test.

270

340 -* mg/log-decorate-HEAD (2015-05-13) 2 commits

341 - (merged to 'next' on 2015-05-19 at 009342b)

342 - + log: do not shorten decoration names too early

343 - + log: decorate HEAD with branch name under --decorate=full, too

271

345 - The "log --decorate" enhancement in Git 2.4 that shows the commit

346 - at the tip of the current branch e.g. "HEAD -> master", did not

347 - work with --decorate=full.

272 +* sg/completion-config (2015-05-12) 2 commits

273 + (merged to 'next' on 2015-05-19 at c83fde1)

274 + + completion: simplify query for config variables

275 + + completion: add a helper function to get config variables

276

349 - Will merge to 'master'.

277 + Code clean-up for completion script (in contrib/).

278

279

352 -* mh/clone-verbosity-fix (2015-05-19) 1 commit

353 - - clone: call transport_set_verbosity before anything else on the newly created transport

280 +* sg/completion-omit-credential-helpers (2015-05-06) 1 commit

281 + (merged to 'next' on 2015-05-11 at aa7e554)

282 + + completion: remove credential helpers from porcelain commands

283

355 - Git 2.4 broke setting verbosity and progress levels on "git clone"

356 - with native transports.

284 + The Git subcommand completion (in contrib/) listed credential

285 + helpers among candidates, which is not something the end user would

286 + invoke interatively.

287

358 - Will merge to 'next'.

288

289 +* sg/help-subcommands (2015-05-08) 1 commit

290 + (merged to 'next' on 2015-05-11 at 91e4f9e)

291 + + command-list.txt: fix whitespace inconsistency

292 + (this branch is used by sg/help-group.)

293

361 -* mh/lockfile-retry (2015-05-14) 2 commits

362 - - lock_packed_refs(): allow retries when acquiring the packed-refs lock

363 - - lockfile: allow file locking to be retried with a timeout

294 + A preparatory clean-up step.

295

365 - Instead of dying immediately upon failing to obtain a lock, retry

366 - after a short while with backoff.

296 +--------------------------------------------------

297 +[New Topics]

298

368 - Will merge to 'next'.

299 +* da/mergetool-winmerge (2015-05-20) 2 commits

300 + - mergetools: add winmerge as a builtin tool

301 + - mergetool--lib: set IFS for difftool and mergetool

302

303 + "git mergetool" learned to drive WinMerge as a backend.

304

371 -* mm/log-format-raw-doc (2015-05-18) 3 commits

372 - - SQUASH

373 - - Documentation/log: clarify sha1 non-abbreviation in log --raw

374 - - Documentation/log: clarify what --raw means

305 + Will merge to 'next'.

306

376 - Clarify that "log --raw" and "log --format=raw" are unrelated

377 - concepts.

307

379 - Will merge to 'next' after squashing the fixup.

308 +* jk/http-backend-deadlock-2.2 (2015-05-20) 3 commits

309 + - http-backend: spool ref negotiation requests to buffer

310 + - t5551: factor out tag creation

311 + - http-backend: fix die recursion with custom handler

312 + (this branch is used by jk/http-backend-deadlock and jk/http-backend-deadlock-2.3.)

313

314 + Communication between the HTTP server and http_backend process can

315 + lead to a dead-lock when relaying a large ref negotiation request.

316 + Diagnose the situation better, and mitigate it by reading such a

317 + request first into core (to a reasonable limit).

318

382 -* ps/doc-packfile-vs-pack-file (2015-05-17) 3 commits

383 - - pack-protocol.txt: fix insconsistent spelling of "packfile"

384 - - git-unpack-objects.txt: fix inconsistent spelling of "packfile"

385 - - git-verify-pack.txt: fix inconsistent spelling of "packfile"

319 + This was wiggled back to apply to the 2.2 maintenance track

320 + (original was for 2.4).

321

387 - Doc consistency updates. I am not sure if this is unifying in the

388 - right direction, though.

322 + Will merge to 'next'.

323

324

391 -* pt/pull-ff-vs-merge-ff (2015-05-18) 2 commits

392 - - pull: parse pull.ff as a bool or string

393 - - pull: make pull.ff=true override merge.ff

325 +* jk/http-backend-deadlock-2.3 (2015-05-20) 1 commit

326 + - Merge branch 'jk/http-backend-deadlock-2.2' into jk/http-backend-deadlock-2.3

327 + (this branch is used by jk/http-backend-deadlock; uses jk/http-backend-deadlock-2.2.)

328

395 - The pull.ff configuration was supposed to override the merge.ff

396 - configuration, but it didn't.

329 + Same for 2.3 maintenance track.

330

331 Will merge to 'next'.

332

333

401 -* pt/pull-log-n (2015-05-18) 1 commit

402 - - pull: handle --log=<n>

334 +* jk/stash-options (2015-05-20) 2 commits

335 + - stash: recognize "--help" for subcommands

336 + - stash: complain about unknown flags

337

404 - "git pull --log" and "git pull --no-log" worked as expected, but

405 - "git pull --log=20" did not.

338 + Make "git stash something --help" error out, so that users can

339 + safely say "git stash drop --help".

340

341 Will merge to 'next'.

342

343

410 -* pt/pull-tags-error-diag (2015-05-14) 1 commit

411 - - pull: remove --tags error in no merge candidates case

344 +* mc/commit-doc-grammofix (2015-05-19) 1 commit

345 + - Documentation/git-commit: grammofix

346

413 - There was a dead code that used to handle "git pull --tags" and

414 - show special-cased error message, which was made irrelevant when

415 - the semantics of the option changed back in Git 1.9 days.

347 + Will merge to 'next'.

348 +

349 +

350 +* rs/janitorial (2015-05-20) 3 commits

351 + - dir: remove unused variable sb

352 + - clean: remove unused variable buf

353 + - use file_exists() to check if a file exists in the worktree

354

355 Will merge to 'next'.

356

357

420 -* rs/plug-leak-in-pack-bitmaps (2015-05-19) 1 commit

421 - - pack-bitmaps: plug memory leak, fix allocation size for recent_bitmaps

358 +* ah/send-email-postfix-alias (2015-05-21) 1 commit

359 + - git-send-email.perl: Add sendmail aliases support

360

423 - The code to memoise recently used pack bitmaps were totally broken.

361

425 - Will merge to 'next'.

362 +* fm/fetch-raw-sha1 (2015-05-21) 5 commits

363 + - SQUASH

364 + - upload-pack: optionally allow fetching reachable sha1

365 + - SQUASH

366 + - upload-pack: prepare to extend allow-tip-sha1-in-want

367 + - config.txt: clarify allowTipSHA1InWant with camelCase

368

369 + "git upload-pack" that serves "git fetch" can be told to serve

370 + commits that are not at the tip of any ref as long as they are

371 + reachable from a ref with uploadpack.allowReachableSHA1InWant

372 + configuration variable.

373

428 -* sb/t1020-cleanup (2015-05-18) 1 commit

429 - - subdirectory tests: code cleanup, uncomment test

374 + Will squash the fixes in and then merge to 'next'.

375

431 - There was a commented-out (instead of being marked to expect

432 - failure) test that documented a breakage that was fixed since the

433 - test was written; turn it into a proper test.

376

435 - Will merge to 'next'.

377 +* jc/commit-slab (2015-05-22) 1 commit

378 + - commit-slab: introduce slabname##_peek() function

379

380 + Memory use reduction when commit-slab facility is used to annotate

381 + sparsely (which is not recommended in the first place).

382

438 -* sg/completion-config (2015-05-12) 2 commits

439 - (merged to 'next' on 2015-05-19 at c83fde1)

440 - + completion: simplify query for config variables

441 - + completion: add a helper function to get config variables

383

443 - Code clean-up for completion script (in contrib/).

384 +* mm/rebase-i-post-rewrite-exec (2015-05-22) 3 commits

385 + - t5407: use <<- to align the expected output

386 + - rebase -i: fix post-rewrite hook with failed exec command

387 + - rebase -i: demonstrate incorrect behavior of post-rewrite

388

445 - Will merge to 'master'.

389 + "git rebase -i" fired post-rewrite hook when it shouldn't (namely,

390 + when it was told to stop sequencing with 'exec' insn).

391

392 + Will merge to 'next'.

393

448 -* sg/help-group (2015-05-18) 5 commits

449 - - help: respect new common command grouping

450 - - command-list.txt: drop the "common" tag

451 - - generate-cmdlist: parse common group commands

452 - - command-list.txt: add the common groups block

453 - - command-list: prepare machinery for upcoming "common groups" section

454 - (this branch uses sg/help-subcommands.)

394

456 - Group list of commands shown by "git help" along the workflow

457 - elements to help early learners.

395 +* sb/submodule-doc-intro (2015-05-22) 1 commit

396 + - submodule documentation: reorder introductory paragraphs

397

398 + Will merge to 'next'.

399

460 -* jc/ignore-epipe-in-filter (2015-05-19) 2 commits

461 - - filter_buffer_or_fd(): ignore EPIPE

462 - - copy.c: make copy_fd() report its status silently

400

464 - Filter scripts were run with SIGPIPE disabled on the Git side,

465 - expecting that they may not read what Git feeds them to filter.

466 - We however treated a filter that does not read its input fully

467 - before exiting as an error.

401 +* sb/test-bitmap-free-at-end (2015-05-22) 1 commit

402 + - test_bitmap_walk: free bitmap with bitmap_free

403

469 - This changes semantics, but arguably in a good way. If a filter

470 - can produce its output without consuming its input using whatever

471 - magic, we now let it do so, instead of diagnosing it as a

472 - programming error.

404 + Will merge to 'next'.

405

406 --------------------------------------------------

407 [Stalled]

@@ -687,175 +619,166 @@ of the repositories listed at

619 --------------------------------------------------

620 [Cooking]

621

690 -* nd/untracked-cache (2015-03-12) 24 commits

691 - (merged to 'next' on 2015-05-19 at 26e619b)

692 - + git-status.txt: advertisement for untracked cache

693 - + untracked cache: guard and disable on system changes

694 - + mingw32: add uname()

695 - + t7063: tests for untracked cache

696 - + update-index: test the system before enabling untracked cache

697 - + update-index: manually enable or disable untracked cache

698 - + status: enable untracked cache

699 - + untracked-cache: temporarily disable with $GIT_DISABLE_UNTRACKED_CACHE

700 - + untracked cache: mark index dirty if untracked cache is updated

701 - + untracked cache: print stats with $GIT_TRACE_UNTRACKED_STATS

702 - + untracked cache: avoid racy timestamps

703 - + read-cache.c: split racy stat test to a separate function

704 - + untracked cache: invalidate at index addition or removal

705 - + untracked cache: load from UNTR index extension

706 - + untracked cache: save to an index extension

707 - + ewah: add convenient wrapper ewah_serialize_strbuf()

708 - + untracked cache: don't open non-existent .gitignore

709 - + untracked cache: mark what dirs should be recursed/saved

710 - + untracked cache: record/validate dir mtime and reuse cached output

711 - + untracked cache: make a wrapper around {open,read,close}dir()

712 - + untracked cache: invalidate dirs recursively if .gitignore changes

713 - + untracked cache: initial untracked cache validation

714 - + untracked cache: record .gitignore information and dir hierarchy

715 - + dir.c: optionally compute sha-1 of a .gitignore file

716 -

717 - Teach the index to optionally remember already seen untracked files

718 - to speed up "git status" in a working tree with tons of cruft.

622 +* dt/cat-file-follow-symlinks (2015-05-20) 3 commits

623 + - cat-file: add --follow-symlinks to --batch

624 + - sha1_name: get_sha1_with_context learns to follow symlinks

625 + - tree-walk: learn get_tree_entry_follow_symlinks

626

720 - Will cook in 'next'.

627 + "git cat-file --batch(-check)" learned the "--follow-symlinks"

628 + option that follows an in-tree symblic links when asked about an

629 + object via extended SHA-1 syntax, e.g. HEAD:RelNotes that points at

630 + Documentation/RelNotes/2.5.0.txt. With the new option, the command

631 + behaves as if HEAD:Documentation/RelNotes/2.5.0.txt was given as

632 + input instead.

633

634 + Will merge to 'next'.

635

723 -* dl/branch-error-message (2015-05-06) 1 commit

724 - (merged to 'next' on 2015-05-11 at ed947ab)

725 - + branch: do not call a "remote-tracking branch" a "remote branch"

636

727 - Error messages from "git branch" called remote-tracking branches as

728 - "remote branches".

637 +* dt/clean-pathspec-filter-then-lstat (2015-05-18) 1 commit

638 + - clean: only lstat files in pathspec

639

730 - Will merge to 'master' in the fourth batch.

640 + "git clean pathspec..." tried to lstat(2) and complain even for

641 + paths outside the given pathspec.

642

643 + Will merge to 'next'.

644

733 -* dl/subtree-push-no-squash (2015-05-07) 1 commit

734 - (merged to 'next' on 2015-05-11 at 74d07ca)

735 - + contrib/subtree: there's no push --squash

645

737 - "git subtree" script (in contrib/) does not have --squash option

738 - when pushing, but the documentation and help text pretended as if

739 - it did.

646 +* jh/filter-empty-contents (2015-05-18) 1 commit

647 + - sha1_file: pass empty buffer to index empty file

648

741 - Will merge to 'master' in the fourth batch.

649 + The clean/smudge interface did not work well when filtering an

650 + empty contents (failed and then passed the empty input through).

651 + It can be argued that a filter that produces anything but empty for

652 + an empty input is nonsense, but if the user wants to do strange

653 + things, then why not?

654

655 + Will merge to 'next'.

656

744 -* ld/p4-editor-multi-words (2015-05-19) 3 commits

745 - - git-p4: fix handling of multi-word P4EDITOR

746 - - SQUASH???

747 - - git-p4: add failing test for P4EDITOR handling

657

749 - Unlike "$EDITOR" and "$GIT_EDITOR" that can hold the path to the

750 - command and initial options (e.g. "/path/to/emacs -nw"), 'git p4'

751 - did not let the shell interpolate the contents of the environment

752 - variable that name the editor "$P4EDITOR" (and "$EDITOR", too).

753 - Make it in line with the rest of Git, as well as with Perforce.

658 +* jk/http-backend-deadlock (2015-05-20) 1 commit

659 + - Merge branch 'jk/http-backend-deadlock-2.3' into jk/http-backend-deadlock

660 + (this branch uses jk/http-backend-deadlock-2.2 and jk/http-backend-deadlock-2.3.)

661

662 + Communication between the HTTP server and http_backend process can

663 + lead to a dead-lock when relaying a large ref negotiation request.

664 + Diagnose the situation better, and mitigate it by reading such a

665 + request first into core (to a reasonable limit).

666

756 -* dl/subtree-avoid-tricky-echo (2015-05-08) 1 commit

757 - (merged to 'next' on 2015-05-11 at 36d4f0e)

758 - + contrib/subtree: portability fix for string printing

667 + Will merge to 'next'.

668

760 - "git subtree" script (in contrib/) used "echo -n" to produce

761 - progress messages in a non-portable way.

669

763 - Will merge to 'master' in the fourth batch.

670 +* jk/rerere-forget-check-enabled (2015-05-14) 1 commit

671 + (merged to 'next' on 2015-05-19 at bfe67dc)

672 + + rerere: exit silently on "forget" when rerere is disabled

673

674 + "git rerere forget" in a repository without rerere enabled gave a

675 + cryptic error message; it should be a silent no-op instead.

676

766 -* ls/http-ssl-cipher-list (2015-05-08) 1 commit

767 - (merged to 'next' on 2015-05-11 at 55764ce)

768 - + http: add support for specifying an SSL cipher list

677 + Will merge to 'master'.

678

770 - Introduce http.<url>.SSLCipherList configuration variable to tweak

771 - the list of cipher suite to be used with libcURL when talking with

772 - https:// sites.

679

774 - Will merge to 'master' in the fourth batch.

680 +* mm/log-format-raw-doc (2015-05-20) 2 commits

681 + - Documentation/log: clarify sha1 non-abbreviation in log --raw

682 + - Documentation/log: clarify what --raw means

683

684 + Clarify that "log --raw" and "log --format=raw" are unrelated

685 + concepts.

686

777 -* ps/bundle-verify-arg (2015-05-08) 1 commit

778 - (merged to 'next' on 2015-05-11 at 9f1b1ae)

779 - + bundle: verify arguments more strictly

687 + Will merge to 'next'.

688

781 - "git bundle verify" did not diagnose extra parameters on the

782 - command line.

689

784 - Will merge to 'master' in the fourth batch.

690 +* pt/pull-ff-vs-merge-ff (2015-05-18) 2 commits

691 + (merged to 'next' on 2015-05-20 at 064a082)

692 + + pull: parse pull.ff as a bool or string

693 + + pull: make pull.ff=true override merge.ff

694

695 + The pull.ff configuration was supposed to override the merge.ff

696 + configuration, but it didn't.

697

787 -* sg/help-subcommands (2015-05-08) 1 commit

788 - (merged to 'next' on 2015-05-11 at 91e4f9e)

789 - + command-list.txt: fix whitespace inconsistency

790 - (this branch is used by sg/help-group.)

698 + Will merge to 'master'.

699

792 - A preparatory clean-up step.

700

794 - Will merge to 'master' in the fourth batch.

701 +* pt/pull-log-n (2015-05-18) 1 commit

702 + (merged to 'next' on 2015-05-20 at db6ce78)

703 + + pull: handle --log=<n>

704

705 + "git pull --log" and "git pull --no-log" worked as expected, but

706 + "git pull --log=20" did not.

707

797 -* mh/write-refs-sooner-2.2 (2015-05-12) 8 commits

798 - + ref_transaction_commit(): fix atomicity and avoid fd exhaustion

799 - + ref_transaction_commit(): remove the local flags variable

800 - + ref_transaction_commit(): inline call to write_ref_sha1()

801 - + rename_ref(): inline calls to write_ref_sha1() from this function

802 - + commit_ref_update(): new function, extracted from write_ref_sha1()

803 - + write_ref_to_lockfile(): new function, extracted from write_ref_sha1()

804 - + t7004: rename ULIMIT test prerequisite to ULIMIT_STACK_SIZE

805 - + update-ref: test handling large transactions properly

806 - (this branch is used by mh/write-refs-sooner-2.3 and mh/write-refs-sooner-2.4.)

708 + Will merge to 'master'.

709

808 - Multi-ref transaction support we merged a few releases ago

809 - unnecessarily kept many file descriptors open, risking to fail with

810 - resource exhaustion.

710

711 +* rs/plug-leak-in-pack-bitmaps (2015-05-19) 1 commit

712 + (merged to 'next' on 2015-05-20 at b70f647)

713 + + pack-bitmaps: plug memory leak, fix allocation size for recent_bitmaps

714

813 -* mh/write-refs-sooner-2.3 (2015-05-12) 1 commit

814 - + Merge branch 'mh/write-refs-sooner-2.2' into mh/write-refs-sooner-2.3

815 - (this branch is used by mh/write-refs-sooner-2.4; uses mh/write-refs-sooner-2.2.)

715 + The code to read pack-bitmap wanted to allocate a few hundred

716 + pointers to a structure, but by mistake allocated and leaked memory

717 + enough to hold that many actual structures. Correct the allocation

718 + size and also have it on stack, as it is small enough.

719

817 - Multi-ref transaction support we merged a few releases ago

818 - unnecessarily kept many file descriptors open, risking to fail with

819 - resource exhaustion. This is for 2.3.x track.

720 + Will merge to 'master'.

721

722

822 -* mh/write-refs-sooner-2.4 (2015-05-12) 9 commits

823 - (merged to 'next' on 2015-05-19 at f7be074)

824 - + Merge branch 'mh/write-refs-sooner-2.3' into mh/write-refs-sooner-2.4

825 - + ref_transaction_commit(): fix atomicity and avoid fd exhaustion

826 - + ref_transaction_commit(): remove the local flags variable

827 - + ref_transaction_commit(): inline call to write_ref_sha1()

828 - + rename_ref(): inline calls to write_ref_sha1() from this function

829 - + commit_ref_update(): new function, extracted from write_ref_sha1()

830 - + write_ref_to_lockfile(): new function, extracted from write_ref_sha1()

831 - + t7004: rename ULIMIT test prerequisite to ULIMIT_STACK_SIZE

832 - + update-ref: test handling large transactions properly

833 - (this branch uses mh/write-refs-sooner-2.2 and mh/write-refs-sooner-2.3.)

723 +* sg/help-group (2015-05-21) 5 commits

724 + - help: respect new common command grouping

725 + - command-list.txt: drop the "common" tag

726 + - generate-cmdlist: parse common group commands

727 + - command-list.txt: add the common groups block

728 + - command-list: prepare machinery for upcoming "common groups" section

729

835 - Multi-ref transaction support we merged a few releases ago

836 - unnecessarily kept many file descriptors open, risking to fail with

837 - resource exhaustion. This is for 2.4.x track.

730 + Group list of commands shown by "git help" along the workflow

731 + elements to help early learners.

732

733 + Will merge to 'next'.

734

840 -* sg/completion-omit-credential-helpers (2015-05-06) 1 commit

841 - (merged to 'next' on 2015-05-11 at aa7e554)

842 - + completion: remove credential helpers from porcelain commands

735

844 - The Git subcommand completion (in contrib/) listed credential

845 - helpers among candidates, which is not something the end user would

846 - invoke interatively.

736 +* nd/untracked-cache (2015-03-12) 24 commits

737 + (merged to 'next' on 2015-05-19 at 26e619b)

738 + + git-status.txt: advertisement for untracked cache

739 + + untracked cache: guard and disable on system changes

740 + + mingw32: add uname()

741 + + t7063: tests for untracked cache

742 + + update-index: test the system before enabling untracked cache

743 + + update-index: manually enable or disable untracked cache

744 + + status: enable untracked cache

745 + + untracked-cache: temporarily disable with $GIT_DISABLE_UNTRACKED_CACHE

746 + + untracked cache: mark index dirty if untracked cache is updated

747 + + untracked cache: print stats with $GIT_TRACE_UNTRACKED_STATS

748 + + untracked cache: avoid racy timestamps

749 + + read-cache.c: split racy stat test to a separate function

750 + + untracked cache: invalidate at index addition or removal

751 + + untracked cache: load from UNTR index extension

752 + + untracked cache: save to an index extension

753 + + ewah: add convenient wrapper ewah_serialize_strbuf()

754 + + untracked cache: don't open non-existent .gitignore

755 + + untracked cache: mark what dirs should be recursed/saved

756 + + untracked cache: record/validate dir mtime and reuse cached output

757 + + untracked cache: make a wrapper around {open,read,close}dir()

758 + + untracked cache: invalidate dirs recursively if .gitignore changes

759 + + untracked cache: initial untracked cache validation

760 + + untracked cache: record .gitignore information and dir hierarchy

761 + + dir.c: optionally compute sha-1 of a .gitignore file

762

848 - Will merge to 'master' in the fourth batch.

763 + Teach the index to optionally remember already seen untracked files

764 + to speed up "git status" in a working tree with tons of cruft.

765

766 + Will cook in 'next'.

767

851 -* jk/skip-http-tests-under-no-curl (2015-05-07) 2 commits

852 - (merged to 'next' on 2015-05-11 at a52b711)

853 - + tests: skip dav http-push tests under NO_EXPAT=NoThanks

854 - + t/lib-httpd.sh: skip tests if NO_CURL is defined

768

856 - Test clean-up.

769 +* ld/p4-editor-multi-words (2015-05-20) 6 commits

770 + - SQUASH

771 + - git-p4: tests: use test-chmtime in place of touch

772 + - SQUASH

773 + - git-p4: fix handling of multi-word P4EDITOR

774 + - SQUASH

775 + - git-p4: add failing test for P4EDITOR handling

776

858 - Will merge to 'master' in the fourth batch.

777 + Unlike "$EDITOR" and "$GIT_EDITOR" that can hold the path to the

778 + command and initial options (e.g. "/path/to/emacs -nw"), 'git p4'

779 + did not let the shell interpolate the contents of the environment

780 + variable that name the editor "$P4EDITOR" (and "$EDITOR", too).

781 + Make it in line with the rest of Git, as well as with Perforce.

782

783

784 * pt/pull-tests (2015-05-18) 8 commits

@@ -877,25 +800,29 @@ of the repositories listed at

800 Still an early WIP

801

802

880 -* jk/at-push-sha1 (2015-05-03) 12 commits

803 +* jk/at-push-sha1 (2015-05-22) 16 commits

804 - for-each-ref: accept "%(push)" format

805 - for-each-ref: use skip_prefix instead of starts_with

806 - sha1_name: implement @{push} shorthand

807 - sha1_name: refactor interpret_upstream_mark

808 - sha1_name: refactor upstream_mark

809 - remote.c: add branch_get_push

810 + - remote.c: return upstream name from stat_tracking_info

811 + - remote.c: untangle error logic in branch_get_upstream

812 - remote.c: report specific errors from branch_get_upstream

813 - remote.c: introduce branch_get_upstream helper

814 + - remote.c: hoist read_config into remote_get_1

815 - remote.c: provide per-branch pushremote name

816 - remote.c: hoist branch.*.remote lookup out of remote_get_1

817 - remote.c: drop "remote" pointer from "struct branch"

818 + - remote.c: refactor setup of branch->merge list

819 - remote.c: drop default_remote_name variable

820

821 Introduce <branch>@{push} short-hand to denote the remote-tracking

822 branch that tracks the branch at the remote the <branch> would be

823 pushed to.

824

898 - Expecting a reroll ($gmane/268419, etc.).

825 + Will merge to 'next'.

826

827

828 * ee/clean-remove-dirs (2015-04-26) 5 commits

What's cooking (2015/05 #06)