c8dfb872eb · 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 (Jul 2018, #03; Wed, 25)

4 -X-master-at: ffc6fa0e396238de3a30623912980263b4f283ab

5 -X-next-at: a71716f1adea8665ad1231d8d929021f562d287b

3 +Subject: What's cooking in git.git (Aug 2018, #01; Thu, 2)

4 +X-master-at: 1d89318c48d233d52f1db230cf622935ac3c69fa

5 +X-next-at: f7a957e2e7d6c0922e75e5f5e5e1a0bb01cfe2bb

6

7 -What's cooking in git.git (Jul 2018, #03; Wed, 25)

7 +What's cooking in git.git (Aug 2018, #01; Thu, 2)

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

9

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

@@ -13,14 +13,7 @@ Here are the topics that have been cooking. Commits prefixed with

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

14

15 Many topics have moved to 'master' and 'next' from 'next' to 'pu'

16 -respectively. As I needed to re-resolve semantic merge conflicts

17 -while reordering some topics (i.e. some that were merged earlier in

18 -'pu' are left in 'pu' while another topic that had interactions with

19 -them and required evil merge to resolve semantic conflicts

20 -leapfrogged to 'next'---the semantic conflict resolution then need

21 -to happen at a different merge than done earlier in 'pu' while

22 -rebuilding today's integration), I didn't have enough time to spend

23 -on new topics posted to the list in the past few days.

16 +respectively.

17

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

19 of the repositories listed at

@@ -30,648 +23,766 @@ of the repositories listed at

23 --------------------------------------------------

24 [Graduated to "master"]

25

33 -* ag/rebase-p (2018-07-06) 1 commit

34 - (merged to 'next' on 2018-07-18 at c36ebba99b)

35 - + git-rebase--preserve-merges: fix formatting of todo help message

36 -

37 - The help message shown in the editor to edit todo list in "rebase -p"

38 - has regressed recently, which has been corrected.

39 -

40 -

41 -* as/sequencer-customizable-comment-char (2018-07-16) 1 commit

42 - (merged to 'next' on 2018-07-18 at 4163e23f29)

43 - + sequencer: use configured comment character

44 -

45 - Honor core.commentchar when preparing the list of commits to replay

46 - in "rebase -i".

47 -

26 +* ab/checkout-default-remote (2018-06-11) 8 commits

27 + (merged to 'next' on 2018-07-24 at 6ef645f485)

28 + + checkout & worktree: introduce checkout.defaultRemote

29 + + checkout: add advice for ambiguous "checkout <branch>"

30 + + builtin/checkout.c: use "ret" variable for return

31 + + checkout: pass the "num_matches" up to callers

32 + + checkout.c: change "unique" member to "num_matches"

33 + + checkout.c: introduce an *_INIT macro

34 + + checkout.h: wrap the arguments to unique_tracking_name()

35 + + checkout tests: index should be clean after dwim checkout

36 + (this branch is used by ab/test-must-be-empty.)

37

49 -* bb/pedantic (2018-07-09) 8 commits

50 - (merged to 'next' on 2018-07-18 at e9d075e8ed)

51 - + utf8.c: avoid char overflow

52 - + string-list.c: avoid conversion from void * to function pointer

53 - + sequencer.c: avoid empty statements at top level

54 - + convert.c: replace "\e" escapes with "\033".

55 - + fixup! refs/refs-internal.h: avoid forward declaration of an enum

56 - + refs/refs-internal.h: avoid forward declaration of an enum

57 - + fixup! connect.h: avoid forward declaration of an enum

58 - + connect.h: avoid forward declaration of an enum

59 - (this branch is used by bb/make-developer-pedantic.)

38 + "git checkout" and "git worktree add" learned to honor

39 + checkout.defaultRemote when auto-vivifying a local branch out of a

40 + remote tracking branch in a repository with multiple remotes that

41 + have tracking branches that share the same names.

42

61 - The codebase has been updated to compile cleanly with -pedantic

62 - option.

43

44 +* bc/object-id (2018-07-16) 16 commits

45 + (merged to 'next' on 2018-07-24 at 23680778a9)

46 + + pretty: switch hard-coded constants to the_hash_algo

47 + + sha1-file: convert constants to uses of the_hash_algo

48 + + log-tree: switch GIT_SHA1_HEXSZ to the_hash_algo->hexsz

49 + + diff: switch GIT_SHA1_HEXSZ to use the_hash_algo

50 + + builtin/merge-recursive: make hash independent

51 + + builtin/merge: switch to use the_hash_algo

52 + + builtin/fmt-merge-msg: make hash independent

53 + + builtin/update-index: simplify parsing of cacheinfo

54 + + builtin/update-index: convert to using the_hash_algo

55 + + refs/files-backend: use the_hash_algo for writing refs

56 + + sha1-name: use the_hash_algo when parsing object names

57 + + strbuf: allocate space with GIT_MAX_HEXSZ

58 + + commit: express tree entry constants in terms of the_hash_algo

59 + + hex: switch to using the_hash_algo

60 + + tree-walk: replace hard-coded constants with the_hash_algo

61 + + cache: update object ID functions for the_hash_algo

62

65 -* bb/unicode-11-width (2018-07-09) 1 commit

66 - (merged to 'next' on 2018-07-18 at 075648ed37)

67 - + unicode: update the width tables to Unicode 11

63 + Conversion from uchar[40] to struct object_id continues.

64

69 - The character display width table has been updated to match the

70 - latest Unicode standard.

65

66 +* bc/sequencer-export-work-tree-as-well (2018-07-16) 1 commit

67 + (merged to 'next' on 2018-07-24 at 0b83ade721)

68 + + sequencer: pass absolute GIT_WORK_TREE to exec commands

69

73 -* bc/send-email-auto-cte (2018-07-09) 4 commits

74 - (merged to 'next' on 2018-07-18 at d16c2a301a)

75 - + docs: correct RFC specifying email line length

76 - + send-email: automatically determine transfer-encoding

77 - + send-email: accept long lines with suitable transfer encoding

78 - + send-email: add an auto option for transfer encoding

70 + "git rebase" started exporting GIT_DIR environment variable and

71 + exposing it to hook scripts when part of it got rewritten in C.

72 + Instead of matching the old scripted Porcelains' behaviour,

73 + compensate by also exporting GIT_WORK_TREE environment as well to

74 + lessen the damage. This can harm existing hooks that want to

75 + operate on different repository, but the current behaviour is

76 + already broken for them anyway.

77

80 - The content-transfer-encoding of the message "git send-email" sends

81 - out by default was 8bit, which can cause trouble when there is an

82 - overlong line to bust RFC 5322/2822 limit. A new option 'auto' to

83 - automatically switch to quoted-printable when there is such a line

84 - in the payload has been introduced and is made the default.

78

79 +* bp/test-drop-caches-for-windows (2018-07-12) 1 commit

80 + (merged to 'next' on 2018-07-24 at 257bb336c6)

81 + + handle lower case drive letters on Windows

82

87 -* bp/log-ref-write-fd-with-strbuf (2018-07-10) 1 commit

88 - (merged to 'next' on 2018-07-18 at 25a3a99528)

89 - + convert log_ref_write_fd() to use strbuf

83 + A test helper update for Windows.

84

91 - Code clean-up.

85

86 +* ds/commit-graph-fsck (2018-07-16) 23 commits

87 + (merged to 'next' on 2018-07-24 at 6a802adc7a)

88 + + coccinelle: update commit.cocci

89 + + commit-graph: update design document

90 + + gc: automatically write commit-graph files

91 + + commit-graph: add '--reachable' option

92 + + commit-graph: use string-list API for input

93 + + fsck: verify commit-graph

94 + + commit-graph: verify contents match checksum

95 + + commit-graph: test for corrupted octopus edge

96 + + commit-graph: verify commit date

97 + + commit-graph: verify generation number

98 + + commit-graph: verify parent list

99 + + commit-graph: verify root tree OIDs

100 + + commit-graph: verify objects exist

101 + + commit-graph: verify corrupt OID fanout and lookup

102 + + commit-graph: verify required chunks are present

103 + + commit-graph: verify catches corrupt signature

104 + + commit-graph: add 'verify' subcommand

105 + + commit-graph: load a root tree from specific graph

106 + + commit: force commit to parse from object database

107 + + commit-graph: parse commit from chosen graph

108 + + commit-graph: fix GRAPH_MIN_SIZE

109 + + commit-graph: UNLEAK before die()

110 + + t5318-commit-graph.sh: use core.commitGraph

111 + (this branch is used by ds/commit-graph-with-grafts, ds/reachable and jt/commit-graph-per-object-store.)

112

94 -* bw/ref-in-want (2018-06-28) 8 commits

95 - (merged to 'next' on 2018-07-18 at 7e9f8db37c)

96 - + fetch-pack: implement ref-in-want

97 - + fetch-pack: put shallow info in output parameter

98 - + fetch: refactor to make function args narrower

99 - + fetch: refactor fetch_refs into two functions

100 - + fetch: refactor the population of peer ref OIDs

101 - + upload-pack: test negotiation with changing repository

102 - + upload-pack: implement ref-in-want

103 - + test-pkt-line: add unpack-sideband subcommand

104 - (this branch is used by bw/fetch-pack-i18n, jt/connectivity-check-after-unshallow, jt/partial-clone-fsck-connectivity and jt/tags-to-promised-blobs-fix.)

113 + "git fsck" learns to make sure the optional commit-graph file is in

114 + a sane state.

115

106 - Protocol v2 has been updated to allow slightly out-of-sync set of

107 - servers to work together to serve a single client, which would be

108 - useful with load-balanced servers that talk smart HTTP transport.

116

117 +* en/dirty-merge-fixes (2018-07-11) 9 commits

118 + (merged to 'next' on 2018-07-24 at 7b6ca3507c)

119 + + merge: fix misleading pre-merge check documentation

120 + + merge-recursive: enforce rule that index matches head before merging

121 + + t6044: add more testcases with staged changes before a merge is invoked

122 + + merge-recursive: fix assumption that head tree being merged is HEAD

123 + + merge-recursive: make sure when we say we abort that we actually abort

124 + + t6044: add a testcase for index matching head, when head doesn't match HEAD

125 + + t6044: verify that merges expected to abort actually abort

126 + + index_has_changes(): avoid assuming operating on the_index

127 + + read-cache.c: move index_has_changes() from merge.c

128

111 -* en/apply-comment-fix (2018-06-28) 1 commit

112 - (merged to 'next' on 2018-07-18 at 31d818f17d)

113 - + apply: fix grammar error in comment

129 + The recursive merge strategy did not properly ensure there was no

130 + change between HEAD and the index before performing its operation,

131 + which has been corrected.

132

133

116 -* en/rebase-consistency (2018-06-27) 9 commits

117 - (merged to 'next' on 2018-07-18 at d597206c79)

118 - + git-rebase: make --allow-empty-message the default

119 - + t3401: add directory rename testcases for rebase and am

120 - + git-rebase.txt: document behavioral differences between modes

121 - + directory-rename-detection.txt: technical docs on abilities and limitations

122 - + git-rebase.txt: address confusion between --no-ff vs --force-rebase

123 - + git-rebase: error out when incompatible options passed

124 - + t3422: new testcases for checking when incompatible options passed

125 - + git-rebase.sh: update help messages a bit

126 - + git-rebase.txt: document incompatible options

134 +* en/t6036-merge-recursive-tests (2018-07-11) 6 commits

135 + (merged to 'next' on 2018-07-24 at 75055cb6e1)

136 + + t6036: add a failed conflict detection case: regular files, different modes

137 + + t6036: add a failed conflict detection case with conflicting types

138 + + t6036: add a failed conflict detection case with submodule add/add

139 + + t6036: add a failed conflict detection case with submodule modify/modify

140 + + t6036: add a failed conflict detection case with symlink add/add

141 + + t6036: add a failed conflict detection case with symlink modify/modify

142

128 - "git rebase" behaved slightly differently depending on which one of

129 - the three backends gets used; this has been documented and an

130 - effort to make them more uniform has begun.

143 + Tests to cover various conflicting cases have been added for

144 + merge-recursive.

145

146

133 -* en/t5407-rebase-m-fix (2018-06-28) 1 commit

134 - (merged to 'next' on 2018-07-18 at 459875daeb)

135 - + t5407: fix test to cover intended arguments

147 +* en/t6036-recursive-corner-cases (2018-07-12) 2 commits

148 + (merged to 'next' on 2018-07-24 at b7b3514ef4)

149 + + t6036: fix broken && chain in sub-shell

150 + + t6036: add lots of detail for directory/file conflicts in recursive case

151

152 + Tests to cover more D/F conflict cases have been added for

153 + merge-recursive.

154

138 -* es/test-lint-one-shot-export (2018-07-16) 5 commits

139 - (merged to 'next' on 2018-07-18 at 26a6124963)

140 - + t/check-non-portable-shell: detect "FOO=bar shell_func"

141 - + t/check-non-portable-shell: make error messages more compact

142 - + t/check-non-portable-shell: stop being so polite

143 - + t6046/t9833: fix use of "VAR=VAL cmd" with a shell function

144 - + Merge branch 'jc/t3404-one-shot-export-fix' into es/test-lint-one-shot-export

145 - (this branch uses jc/t3404-one-shot-export-fix.)

155

147 - Look for broken use of "VAR=VAL shell_func" in test scripts as part

148 - of test-lint.

156 +* en/t6042-insane-merge-rename-testcases (2018-07-03) 3 commits

157 + (merged to 'next' on 2018-07-24 at 65c80f72da)

158 + + t6042: add testcase covering long chains of rename conflicts

159 + + t6042: add testcase covering rename/rename(2to1)/delete/delete conflict

160 + + t6042: add testcase covering rename/add/delete conflict type

161

162 + Various glitches in the heuristics of merge-recursive strategy have

163 + been documented in new tests.

164

151 -* hs/push-cert-check-cleanup (2018-07-11) 2 commits

152 - (merged to 'next' on 2018-07-18 at 1ed25fbd77)

153 - + gpg-interface: make parse_gpg_output static and remove from interface header

154 - + builtin/receive-pack: use check_signature from gpg-interface

155 - (this branch is used by hs/gpgsm.)

165 + I am not sure if there is a single "correct" answer everybody can

166 + agree on for each of these "insane" cases, though.

167

157 - Code clean-up.

168

169 +* en/t7405-recursive-submodule-conflicts (2018-07-11) 3 commits

170 + (merged to 'next' on 2018-07-24 at 6cb7d02298)

171 + + t7405: verify 'merge --abort' works after submodule/path conflicts

172 + + t7405: add a directory/submodule conflict

173 + + t7405: add a file/submodule conflict

174

160 -* jc/t3404-one-shot-export-fix (2018-07-12) 1 commit

161 - (merged to 'next' on 2018-07-18 at e15a79dca7)

162 - + t3404: fix use of "VAR=VAL cmd" with a shell function

163 - (this branch is used by es/test-lint-one-shot-export.)

175 + Tests to cover conflict cases that involve submodules have been

176 + added for merge-recursive.

177

165 - Correct a broken use of "VAR=VAL shell_func" in a test.

178

179 +* es/chain-lint-in-subshell (2018-07-31) 11 commits

180 + (merged to 'next' on 2018-07-31 at 4ce2a8faa4)

181 + + t/chainlint.sed: drop extra spaces from regex character class

182 + (merged to 'next' on 2018-07-24 at 9370bbdfaf)

183 + + t/chainlint: add chainlint "specialized" test cases

184 + + t/chainlint: add chainlint "complex" test cases

185 + + t/chainlint: add chainlint "cuddled" test cases

186 + + t/chainlint: add chainlint "loop" and "conditional" test cases

187 + + t/chainlint: add chainlint "nested subshell" test cases

188 + + t/chainlint: add chainlint "one-liner" test cases

189 + + t/chainlint: add chainlint "whitespace" test cases

190 + + t/chainlint: add chainlint "basic" test cases

191 + + t/Makefile: add machinery to check correctness of chainlint.sed

192 + + t/test-lib: teach --chain-lint to detect broken &&-chains in subshells

193 + (this branch uses es/test-fixes.)

194

168 -* jk/empty-pick-fix (2018-07-11) 2 commits

169 - (merged to 'next' on 2018-07-18 at 43bfa862f2)

170 - + sequencer: don't say BUG on bogus input

171 - + sequencer: handle empty-set cases consistently

195 + Look for broken "&&" chains that are hidden in subshell, many of

196 + which have been found and corrected.

197

173 - Handling of an empty range by "git cherry-pick" was inconsistent

174 - depending on how the range ended up to be empty, which has been

175 - corrected.

198

199 +* es/test-fixes (2018-07-17) 26 commits

200 + (merged to 'next' on 2018-07-24 at fd6796a3ef)

201 + + t5608: fix broken &&-chain

202 + + t9119: fix broken &&-chains

203 + + t9000-t9999: fix broken &&-chains

204 + + t7000-t7999: fix broken &&-chains

205 + + t6000-t6999: fix broken &&-chains

206 + + t5000-t5999: fix broken &&-chains

207 + + t4000-t4999: fix broken &&-chains

208 + + t3030: fix broken &&-chains

209 + + t3000-t3999: fix broken &&-chains

210 + + t2000-t2999: fix broken &&-chains

211 + + t1000-t1999: fix broken &&-chains

212 + + t0000-t0999: fix broken &&-chains

213 + + t9814: simplify convoluted check that command correctly errors out

214 + + t9001: fix broken "invoke hook" test

215 + + t7810: use test_expect_code() instead of hand-rolled comparison

216 + + t7400: fix broken "submodule add/reconfigure --force" test

217 + + t7201: drop pointless "exit 0" at end of subshell

218 + + t6036: fix broken "merge fails but has appropriate contents" tests

219 + + t5505: modernize and simplify hard-to-digest test

220 + + t5406: use write_script() instead of birthing shell script manually

221 + + t5405: use test_must_fail() instead of checking exit code manually

222 + + t/lib-submodule-update: fix "absorbing" test

223 + + t: drop unnecessary terminating semicolon in subshell

224 + + t: use sane_unset() rather than 'unset' with broken &&-chain

225 + + t: use test_write_lines() instead of series of 'echo' commands

226 + + t: use test_might_fail() instead of manipulating exit code manually

227 + (this branch is used by es/chain-lint-in-subshell.)

228

178 -* jk/fetch-all-peeled-fix (2018-07-06) 1 commit

179 - (merged to 'next' on 2018-07-18 at d06c6f1665)

180 - + t5500: prettify non-commit tag tests

229 + Test clean-up and corrections.

230

182 - Test modernization.

231

232 +* is/parsing-line-range (2018-06-15) 2 commits

233 + (merged to 'next' on 2018-07-24 at a06b453f32)

234 + + log: prevent error if line range ends past end of file

235 + + blame: prevent error if range ends past end of file

236

185 -* jk/for-each-ref-icase (2018-07-03) 3 commits

186 - (merged to 'next' on 2018-07-18 at 4c86d62adb)

187 - + ref-filter: avoid backend filtering with --ignore-case

188 - + for-each-ref: consistently pass WM_IGNORECASE flag

189 - + t6300: add a test for --ignore-case

237 + Parsing of -L[<N>][,[<M>]] parameters "git blame" and "git log"

238 + take has been tweaked.

239

191 - The "--ignore-case" option of "git for-each-ref" (and its friends)

192 - did not work correctly, which has been fixed.

240

241 +* jk/fsck-gitmodules-gently (2018-07-16) 6 commits

242 + (merged to 'next' on 2018-07-24 at 5b15c800db)

243 + + fsck: downgrade gitmodulesParse default to "info"

244 + + fsck: split ".gitmodules too large" error from parse failure

245 + + fsck: silence stderr when parsing .gitmodules

246 + + config: add options parameter to git_config_from_mem

247 + + config: add CONFIG_ERROR_SILENT handler

248 + + config: turn die_on_error into caller-facing enum

249

195 -* jt/connectivity-check-after-unshallow (2018-07-03) 1 commit

196 - (merged to 'next' on 2018-07-18 at 8e7ee889c3)

197 - + fetch-pack: write shallow, then check connectivity

198 - (this branch is used by jt/partial-clone-fsck-connectivity and jt/tags-to-promised-blobs-fix; uses bw/ref-in-want; is tangled with bw/fetch-pack-i18n.)

250 + Recent "security fix" to pay attention to contents of ".gitmodules"

251 + while accepting "git push" was a bit overly strict than necessary,

252 + which has been adjusted.

253

200 - "git fetch" failed to correctly validate the set of objects it

201 - received when making a shallow history deeper, which has been

202 - corrected.

254

255 +* jk/has-uncommitted-changes-fix (2018-07-11) 1 commit

256 + (merged to 'next' on 2018-07-24 at 2ea14c0afb)

257 + + has_uncommitted_changes(): fall back to empty tree

258

205 -* jt/partial-clone-fsck-connectivity (2018-07-09) 2 commits

206 - (merged to 'next' on 2018-07-18 at 968fd9c9f0)

207 - + clone: check connectivity even if clone is partial

208 - + upload-pack: send refs' objects despite "filter"

209 - (this branch is used by jt/tags-to-promised-blobs-fix; uses bw/ref-in-want and jt/connectivity-check-after-unshallow; is tangled with bw/fetch-pack-i18n.)

259 + "git pull --rebase" on a corrupt HEAD caused a segfault. In

260 + general we substitute an empty tree object when running the in-core

261 + equivalent of the diff-index command, and the codepath has been

262 + corrected to do so as well to fix this issue.

263

211 - Partial clone support of "git clone" has been updated to correctly

212 - validate the objects it receives from the other side. The server

213 - side has been corrected to send objects that are directly

214 - requested, even if they may match the filtering criteria (e.g. when

215 - doing a "lazy blob" partial clone).

264

265 +* jm/cache-entry-from-mem-pool (2018-07-03) 8 commits

266 + (merged to 'next' on 2018-07-24 at 9be51a88dc)

267 + + block alloc: add validations around cache_entry lifecyle

268 + + block alloc: allocate cache entries from mem_pool

269 + + mem-pool: fill out functionality

270 + + mem-pool: add life cycle management functions

271 + + mem-pool: only search head block for available space

272 + + block alloc: add lifecycle APIs for cache_entry structs

273 + + read-cache: teach make_cache_entry to take object_id

274 + + read-cache: teach refresh_cache_entry to take istate

275

218 -* kn/userdiff-php (2018-07-06) 2 commits

219 - (merged to 'next' on 2018-07-18 at 9a533dc33a)

220 - + userdiff: support new keywords in PHP hunk header

221 - + t4018: add missing test cases for PHP

276 + For a large tree, the index needs to hold many cache entries

277 + allocated on heap. These cache entries are now allocated out of a

278 + dedicated memory pool to amortize malloc(3) overhead.

279

223 - The userdiff pattern for .php has been updated.

280 + This makes each cache-entry larger by either 4 or 8 bytes, which is

281 + a bit sad, though.

282

283

226 -* mh/fast-import-no-diff-delta-empty (2018-07-06) 1 commit

227 - (merged to 'next' on 2018-07-18 at eb393871f4)

228 - + fast-import: do not call diff_delta() with empty buffer

284 +* jm/send-email-tls-auth-on-batch (2018-07-16) 1 commit

285 + (merged to 'next' on 2018-07-24 at fb3e653f44)

286 + + send-email: fix tls AUTH when sending batch

287

230 - "git fast-import" has been updated to avoid attempting to create

231 - delta against a zero-byte-long string, which is pointless.

288 + "git send-email" when using in a batched mode that limits the

289 + number of messages sent in a single SMTP session lost the contents

290 + of the variable used to choose between tls/ssl, unable to send the

291 + second and later batches, which has been fixed.

292

293 + This is marked to be merged to 'next' already, but I do not mind

294 + getting an updated version with an improved log message before that

295 + happens.

296

234 -* mk/merge-in-sparse-checkout (2018-07-11) 1 commit

235 - (merged to 'next' on 2018-07-18 at d2a6d2684d)

236 - + unpack-trees: do not fail reset because of unmerged skipped entry

297

238 - "git reset --merge" (hence "git merge ---abort") and "git reset --hard"

239 - had trouble working correctly in a sparsely checked out working

240 - tree after a conflict, which has been corrected.

298 +* js/rebase-merge-octopus (2018-07-11) 3 commits

299 + (merged to 'next' on 2018-07-24 at 14ad8699de)

300 + + rebase --rebase-merges: adjust man page for octopus support

301 + + rebase --rebase-merges: add support for octopus merges

302 + + merge: allow reading the merge commit message from a file

303

304 + "git rebase --rebase-merges" mode now handles octopus merges as

305 + well.

306

243 -* nd/command-list (2018-07-16) 1 commit

244 - (merged to 'next' on 2018-07-18 at 77ed2a3914)

245 - + vcbuild/README: update to accommodate for missing common-cmds.h

307

247 - Build doc update for Windows.

308 +* jt/commit-graph-per-object-store (2018-07-17) 7 commits

309 + (merged to 'next' on 2018-07-24 at 090d1a4d59)

310 + + commit-graph: add repo arg to graph readers

311 + + commit-graph: store graph in struct object_store

312 + + commit-graph: add free_commit_graph

313 + + commit-graph: add missing forward declaration

314 + + object-store: add missing include

315 + + commit-graph: refactor preparing commit graph

316 + + Merge branch 'ds/commit-graph-fsck' into jt/commit-graph-per-object-store

317 + (this branch is used by ds/commit-graph-with-grafts and ds/reachable; uses ds/commit-graph-fsck and sb/object-store-lookup.)

318

319 + The singleton commit-graph in-core instance is made per in-core

320 + repository instance.

321

250 -* rj/submodule-fsck-skip (2018-07-03) 1 commit

251 - (merged to 'next' on 2018-07-11 at 985f88cf7e)

252 - + fsck: check skiplist for object in fsck_blob()

322

254 - "fsck.skipList" did not prevent a blob object listed there from

255 - being inspected for is contents (e.g. we recently started to

256 - inspect the contents of ".gitmodules" for certain malicious

257 - patterns), which has been corrected.

323 +* jt/fetch-nego-tip (2018-07-03) 1 commit

324 + (merged to 'next' on 2018-07-24 at a9e299006d)

325 + + fetch-pack: support negotiation tip whitelist

326 + (this branch is used by ab/fetch-nego; uses jt/fetch-pack-negotiator; is tangled with jt/fetch-negotiator-skipping.)

327

328 + "git fetch" learned a new option "--negotiation-tip" to limit the

329 + set of commits it tells the other end as "have", to reduce wasted

330 + bandwidth and cycles, which would be helpful when the receiving

331 + repository has a lot of refs that have little to do with the

332 + history at the remote it is fetching from.

333

260 -* sb/blame-color (2018-07-16) 1 commit

261 - (merged to 'next' on 2018-07-18 at c319268502)

262 - + blame: prefer xsnprintf to strcpy for colors

263 - (this branch is used by jk/banned-function.)

334

265 - Code clean-up.

335 +* jt/fetch-negotiator-skipping (2018-07-16) 1 commit

336 + (merged to 'next' on 2018-07-24 at 8e25a49405)

337 + + negotiator/skipping: skip commits during fetch

338 + (this branch is used by ab/fetch-nego; uses jt/fetch-pack-negotiator; is tangled with jt/fetch-nego-tip.)

339

340 + Add a server-side knob to skip commits in exponential/fibbonacci

341 + stride in an attempt to cover wider swath of history with a smaller

342 + number of iterations, potentially accepting a larger packfile

343 + transfer, instead of going back one commit a time during common

344 + ancestor discovery during the "git fetch" transaction.

345

268 -* sb/submodule-move-head-error-msg (2018-06-25) 1 commit

269 - (merged to 'next' on 2018-07-18 at 9e213ad1aa)

270 - + submodule.c: report the submodule that an error occurs in

346

272 - "git checkout --recurse-submodules another-branch" did not report

273 - in which submodule it failed to update the working tree, which

274 - resulted in an unhelpful error message.

347 +* jt/fetch-pack-negotiator (2018-06-15) 7 commits

348 + (merged to 'next' on 2018-07-24 at 438efcd6b1)

349 + + fetch-pack: introduce negotiator API

350 + + fetch-pack: move common check and marking together

351 + + fetch-pack: make negotiation-related vars local

352 + + fetch-pack: use ref adv. to prune "have" sent

353 + + fetch-pack: directly end negotiation if ACK ready

354 + + fetch-pack: clear marks before re-marking

355 + + fetch-pack: split up everything_local()

356 + (this branch is used by ab/fetch-nego, jt/fetch-nego-tip and jt/fetch-negotiator-skipping.)

357

358 + Code restructuring and a small fix to transport protocol v2 during

359 + fetching.

360

277 -* tb/config-default (2018-07-06) 1 commit

278 - (merged to 'next' on 2018-07-18 at 7994476f6f)

279 - + builtin/config: work around an unsized array forward declaration

361

281 - Compilation fix.

362 +* jt/tags-to-promised-blobs-fix (2018-07-16) 2 commits

363 + (merged to 'next' on 2018-07-24 at 8d7e78a671)

364 + + tag: don't warn if target is missing but promised

365 + + revision: tolerate promised targets of tags

366

367 + The lazy clone support had a few places where missing but promised

368 + objects were not correctly tolerated, which have been fixed.

369

284 -* wc/find-commit-with-pattern-on-detached-head (2018-07-12) 1 commit

285 - (merged to 'next' on 2018-07-18 at 334d2420c0)

286 - + sha1-name.c: for ":/", find detached HEAD commits

370

288 - "git rev-parse ':/substring'" did not consider the history leading

289 - only to HEAD when looking for a commit with the given substring,

290 - when the HEAD is detached. This has been fixed.

371 +* kg/gc-auto-windows-workaround (2018-07-09) 1 commit

372 + (merged to 'next' on 2018-07-24 at 71c05d27b6)

373 + + gc --auto: release pack files before auto packing

374

292 ---------------------------------------------------

293 -[New Topics]

375 + "git gc --auto" opens file descriptors for the packfiles before

376 + spawning "git repack/prune", which would upset Windows that does

377 + not want a process to work on a file that is open by another

378 + process. The issue has been worked around.

379

295 -* ds/commit-graph-with-grafts (2018-07-19) 8 commits

296 - - commit-graph: close_commit_graph before shallow walk

297 - - commit-graph: not compatible with uninitialized repo

298 - - commit-graph: not compatible with grafts

299 - - commit-graph: not compatible with replace objects

300 - - test-repository: properly init repo

301 - - commit-graph: update design document

302 - - refs.c: upgrade for_each_replace_ref to be a each_repo_ref_fn callback

303 - - refs.c: migrate internal ref iteration to pass thru repository argument

304 - (this branch uses ds/commit-graph-fsck, jt/commit-graph-per-object-store and sb/object-store-lookup; is tangled with ds/reachable.)

380

306 - The recently introduced commit-graph auxiliary data is incompatible

307 - with mechanisms such as replace & grafts that "breaks" immutable

308 - nature of the object reference relationship. Disable optimizations

309 - based on its use (and updating existing commit-graph) when these

310 - incompatible features are in use in the repository.

381 +* sb/diff-color-move-more (2018-07-19) 10 commits

382 + (merged to 'next' on 2018-07-24 at 89c893cab2)

383 + + diff.c: offer config option to control ws handling in move detection

384 + + diff.c: add white space mode to move detection that allows indent changes

385 + + diff.c: factor advance_or_nullify out of mark_color_as_moved

386 + + diff.c: decouple white space treatment from move detection algorithm

387 + + diff.c: add a blocks mode for moved code detection

388 + + diff.c: adjust hash function signature to match hashmap expectation

389 + + diff.c: do not pass diff options as keydata to hashmap

390 + + t4015: avoid git as a pipe input

391 + + xdiff/xdiffi.c: remove unneeded function declarations

392 + + xdiff/xdiff.h: remove unused flags

393

312 - Will merge to 'next'.

394 + "git diff --color-moved" feature has further been tweaked.

395

396

315 -* jk/core-use-replace-refs (2018-07-18) 3 commits

316 - - add core.usereplacerefs config option

317 - - check_replace_refs: rename to read_replace_refs

318 - - check_replace_refs: fix outdated comment

397 +* sb/object-store-lookup (2018-06-29) 33 commits

398 + (merged to 'next' on 2018-07-24 at dd96e29376)

399 + + commit.c: allow lookup_commit_reference to handle arbitrary repositories

400 + + commit.c: allow lookup_commit_reference_gently to handle arbitrary repositories

401 + + tag.c: allow deref_tag to handle arbitrary repositories

402 + + object.c: allow parse_object to handle arbitrary repositories

403 + + object.c: allow parse_object_buffer to handle arbitrary repositories

404 + + commit.c: allow get_cached_commit_buffer to handle arbitrary repositories

405 + + commit.c: allow set_commit_buffer to handle arbitrary repositories

406 + + commit.c: migrate the commit buffer to the parsed object store

407 + + commit-slabs: remove realloc counter outside of slab struct

408 + + commit.c: allow parse_commit_buffer to handle arbitrary repositories

409 + + tag: allow parse_tag_buffer to handle arbitrary repositories

410 + + tag: allow lookup_tag to handle arbitrary repositories

411 + + commit: allow lookup_commit to handle arbitrary repositories

412 + + tree: allow lookup_tree to handle arbitrary repositories

413 + + blob: allow lookup_blob to handle arbitrary repositories

414 + + object: allow lookup_object to handle arbitrary repositories

415 + + object: allow object_as_type to handle arbitrary repositories

416 + + tag: add repository argument to deref_tag

417 + + tag: add repository argument to parse_tag_buffer

418 + + tag: add repository argument to lookup_tag

419 + + commit: add repository argument to get_cached_commit_buffer

420 + + commit: add repository argument to set_commit_buffer

421 + + commit: add repository argument to parse_commit_buffer

422 + + commit: add repository argument to lookup_commit

423 + + commit: add repository argument to lookup_commit_reference

424 + + commit: add repository argument to lookup_commit_reference_gently

425 + + tree: add repository argument to lookup_tree

426 + + blob: add repository argument to lookup_blob

427 + + object: add repository argument to object_as_type

428 + + object: add repository argument to parse_object_buffer

429 + + object: add repository argument to lookup_object

430 + + object: add repository argument to parse_object

431 + + Merge branch 'sb/object-store-grafts' into sb/object-store-lookup

432 + (this branch is used by ds/commit-graph-with-grafts, ds/reachable and jt/commit-graph-per-object-store.)

433

320 - A new configuration variable core.usereplacerefs has been added,

321 - primarily to help server installations that want to ignore the

322 - replace mechanism altogether.

434 + lookup_commit_reference() and friends have been updated to find

435 + in-core object for a specific in-core repository instance.

436

324 - Will merge to 'next'.

437

438 +* sg/httpd-test-unflake (2018-07-12) 3 commits

439 + (merged to 'next' on 2018-07-24 at b7df820256)

440 + + t/lib-httpd: avoid occasional failures when checking access.log

441 + + t/lib-httpd: add the strip_access_log() helper function

442 + + t5541: clean up truncating access log

443

327 -* nd/i18n (2018-07-23) 23 commits

328 - - transport-helper.c: mark more strings for translation

329 - - transport.c: mark more strings for translation

330 - - sha1-file.c: mark more strings for translation

331 - - sequencer.c: mark more strings for translation

332 - - replace-object.c: mark more strings for translation

333 - - refspec.c: mark more strings for translation

334 - - refs.c: mark more strings for translation

335 - - pkt-line.c: mark more strings for translation

336 - - object.c: mark more strings for translation

337 - - exec-cmd.c: mark more strings for translation

338 - - environment.c: mark more strings for translation

339 - - dir.c: mark more strings for translation

340 - - convert.c: mark more strings for translation

341 - - connect.c: mark more strings for translation

342 - - config.c: mark more strings for translation

343 - - commit-graph.c: mark more strings for translation

344 - - builtin/replace.c: mark more strings for translation

345 - - builtin/pack-objects.c: mark more strings for translation

346 - - builtin/grep.c: mark strings for translation

347 - - builtin/config.c: mark more strings for translation

348 - - archive-zip.c: mark more strings for translation

349 - - archive-tar.c: mark more strings for translation

350 - - Update messages in preparation for i18n

444 + httpd tests saw occasional breakage due to the way its access log

445 + gets inspected by the tests, which has been updated to make them

446 + less flaky.

447

352 - Many more strings are prepared for l10n.

448 +

449 +* tb/grep-only-matching (2018-07-09) 2 commits

450 + (merged to 'next' on 2018-07-24 at 7e878b9d95)

451 + + grep.c: teach 'git grep --only-matching'

452 + + grep.c: extract show_line_header()

453 +

454 + "git grep" learned the "--only-matching" option.

455 +

456 +--------------------------------------------------

457 +[New Topics]

458 +

459 +* ab/fsck-transfer-updates (2018-07-27) 10 commits

460 + - fsck: test and document unknown fsck.<msg-id> values

461 + - fsck: add stress tests for fsck.skipList

462 + - fsck: test & document {fetch,receive}.fsck.* config fallback

463 + - fetch: implement fetch.fsck.*

464 + - transfer.fsckObjects tests: untangle confusing setup

465 + - config doc: elaborate on fetch.fsckObjects security

466 + - config doc: elaborate on what transfer.fsckObjects does

467 + - config doc: unify the description of fsck.* and receive.fsck.*

468 + - config doc: don't describe *.fetchObjects twice

469 + - receive.fsck.<msg-id> tests: remove dead code

470 +

471 + The test performed at the receiving end of "git push" to prevent

472 + bad objects from entering repository can be customized via

473 + receive.fsck.* configuration variables; we now have gained a

474 + counterpart to do the same on the "git fetch" side, with

475 + fetch.fsck.* configuration variables.

476

477 Will merge to 'next'.

478

479

357 -* sb/histogram-less-memory (2018-07-23) 4 commits

358 - - xdiff/histogram: remove tail recursion

359 - - xdiff/xhistogram: move index allocation into find_lcs

360 - - xdiff/xhistogram: factor out memory cleanup into free_index()

361 - - xdiff/xhistogram: pass arguments directly to fall_back_to_classic_diff

480 +* ab/test-must-be-empty (2018-07-30) 1 commit

481 + - tests: make use of the test_must_be_empty function

482

363 - "git diff --histogram" had a bad memory usage pattern, which has

364 - been rearranged to reduce the peak usage.

483 + Test updates.

484

485 Will merge to 'next'.

486

487

369 -* bb/make-developer-pedantic (2018-07-25) 1 commit

370 - - Makefile: add a DEVOPTS flag to get pedantic compilation

488 +* ab/test-must-be-empty-for-master (2018-07-30) 1 commit

489 + - tests: make use of the test_must_be_empty function

490

372 - "make DEVELOPER=1 DEVOPTS=pedantic" allows developers to compile

373 - with -pedantic option, which may catch more problematic program

374 - constructs and potential bugs.

491 + Test updates.

492

376 - Will merge to 'next' and then to 'master'.

493 + Did anybody spot incorrect conversion in this yet?

494

495

379 -* bw/clone-ref-prefixes (2018-07-20) 1 commit

380 - - clone: send ref-prefixes when using protocol v2

496 +* cb/p4-pre-submit-hook (2018-08-01) 1 commit

497 + - git-p4: add the `p4-pre-submit` hook

498

382 - The wire-protocol v2 relies on the client to send "ref prefixes" to

383 - limit the bandwidth spent on the initial ref advertisement. "git

384 - clone" when learned to speak v2 forgot to do so, which has been

385 - corrected.

499 + "git p4 submit" learns to ask its own pre-submit hook if it should

500 + continue with submitting.

501

502 Will merge to 'next'.

503

504

390 -* bw/fetch-pack-i18n (2018-07-23) 1 commit

391 - - fetch-pack: mark die strings for translation

392 -

393 - i18n updates.

505 +* es/rebase-i-author-script-fix (2018-07-31) 4 commits

506 + - sequencer: don't die() on bogus user-edited timestamp

507 + - sequencer: fix "rebase -i --root" corrupting author header timestamp

508 + - sequencer: fix "rebase -i --root" corrupting author header timezone

509 + - sequencer: fix "rebase -i --root" corrupting author header

510 + (this branch is used by pw/rebase-i-author-script-fix.)

511

395 - Will merge to 'next' and then to 'master'.

512 + The "author-script" file "git rebase -i" creates got broken when

513 + we started to move the command away from shell script, which is

514 + getting fixed now.

515

516 + Will merge to 'next'.

517

398 -* bw/protocol-v2 (2018-07-24) 1 commit

399 - - pack-protocol: mention and point to docs for protocol v2

518

401 - Doc update.

519 +* hn/highlight-sideband-keywords (2018-07-31) 1 commit

520 + - sideband: highlight keywords in remote output

521

403 - Will merge to 'next' and then to 'master'.

522 + The sideband code learned to optionally paint selected keywords at

523 + the beginning of incoming lines on the receiving end.

524

525

406 -* ds/reachable (2018-07-20) 18 commits

407 - - commit-reach: use can_all_from_reach

408 - - commit-reach: make can_all_from_reach... linear

409 - - commit-reach: replace ref_newer logic

410 - - test-reach: test commit_contains

411 - - test-reach: test can_all_from_reach_with_flags

412 - - test-reach: test reduce_heads

413 - - test-reach: test get_merge_bases_many

414 - - test-reach: test is_descendant_of

415 - - test-reach: test in_merge_bases

416 - - test-reach: create new test tool for ref_newer

417 - - commit-reach: move can_all_from_reach_with_flags

418 - - upload-pack: generalize commit date cutoff

419 - - upload-pack: refactor ok_to_give_up()

420 - - upload-pack: make reachable() more generic

421 - - commit-reach: move commit_contains from ref-filter

422 - - commit-reach: move ref_newer from remote.c

423 - - commit.h: remove method declarations

424 - - commit-reach: move walk methods from commit.c

425 - (this branch uses ds/commit-graph-fsck, jt/commit-graph-per-object-store and sb/object-store-lookup; is tangled with ds/commit-graph-with-grafts.)

526 +* jn/subtree-test-fixes (2018-07-30) 2 commits

527 + - subtree test: simplify preparation of expected results

528 + - subtree test: add missing && to &&-chain

529

427 - The code for computing history reachability has been shuffled,

428 - obtained a bunch of new tests to cover them, and then being

429 - improved.

530 + Test fix.

531

431 - Stuck in review?

432 - cf. <20180723203500.231932-1-jonathantanmy@google.com>

433 - cf. <20180723204112.233274-1-jonathantanmy@google.com>

434 - cf. <CAGZ79kb7tWV=cmboA+nsChAUaiC+fVVM-GBCuWfsypC+-wyaVg@mail.gmail.com>

532 + Will merge to 'next'.

533

534

437 -* en/merge-recursive-skip-fix (2018-07-23) 2 commits

438 - - merge-recursive: preserve skip_worktree bit when necessary

439 - - t3507: add a testcase showing failure with sparse checkout

535 +* ms/http-proto-doc (2018-07-30) 1 commit

536 + - doc: fix want-capability separator

537

441 - When the sparse checkout feature is in use, "git cherry-pick" and

442 - other mergy operations lost the skip_worktree bit when a path that

443 - is excluded from checkout requires content level merge, which is

444 - resolved as the same as the HEAD version, without materializing the

445 - merge result in the working tree, which made the path appear as

446 - deleted. This has been corrected by preserving the skip_worktree

447 - bit (and not materializing the file in the working tree).

538 + Doc fix.

539

449 - Stuck in review?

450 - cf. <75aa297e-4857-d92a-7041-618ff3b0b77a@gmail.com>

540 + Will merge to 'next'.

541

542

453 -* es/format-patch-interdiff (2018-07-23) 6 commits

454 - - format-patch: allow --interdiff to apply to a lone-patch

455 - - log-tree: show_log: make commentary block delimiting reusable

456 - - interdiff: teach show_interdiff() to indent interdiff

457 - - format-patch: teach --interdiff to respect -v/--reroll-count

458 - - format-patch: add --interdiff option to embed diff in cover letter

459 - - format-patch: allow additional generated content in make_cover_letter()

460 - (this branch is used by es/format-patch-rangediff.)

543 +* nd/pack-objects-threading-doc (2018-07-30) 1 commit

544 + - pack-objects: document about thread synchronization

545

462 - "git format-patch" learned a new "--interdiff" option to explain

463 - the difference between this version and the previous atttempt in

464 - the cover letter (or after the tree-dashes as a comment).

546 + Doc fix.

547

466 - Stuck in review?

467 - cf. <CAPig+cSuYUYSPTuKx08wcmQM-G12_-W2T4BS07fA=6grM1b8Gw@mail.gmail.com>

548 + Will merge to 'next'.

549

550

470 -* es/format-patch-rangediff (2018-07-25) 10 commits

471 - - format-patch: allow --range-diff to apply to a lone-patch

472 - - format-patch: add --creation-factor tweak for --range-diff

473 - - format-patch: teach --range-diff to respect -v/--reroll-count

474 - - format-patch: extend --range-diff to accept revision range

475 - - format-patch: add --range-diff option to embed diff in cover letter

476 - - range-diff: relieve callers of low-level configuration burden

477 - - range-diff: publish default creation factor

478 - - range-diff: respect diff_option.file rather than assuming 'stdout'

479 - - Merge branch 'es/format-patch-interdiff' into es/format-patch-rangediff

480 - - Merge branch 'js/range-diff' into es/format-patch-rangediff

481 - (this branch uses es/format-patch-interdiff and js/range-diff.)

551 +* sb/indent-heuristic-optim (2018-08-01) 1 commit

552 + - xdiff: reduce indent heuristic overhead

553

483 - "git format-patch" learned a new "--range-diff" option to explain

484 - the difference between this version and the previous atttempt in

485 - the cover letter (or after the tree-dashes as a comment).

554 + "git diff --indent-heuristic" had a bad corner case performance.

555

487 - Need to wait for the prereq topics to solidify a bit more.

556 + Will merge to 'next'.

557

558

490 -* jk/banned-function (2018-07-24) 5 commits

491 - - banned.h: mark strncpy() as banned

492 - - banned.h: mark sprintf() as banned

493 - - banned.h: mark strcat() as banned

494 - - automatically ban strcpy()

495 - - Merge branch 'sb/blame-color' into jk/banned-function

559 +* ab/fetch-nego (2018-08-01) 3 commits

560 + - fetch doc: cross-link two new negotiation options

561 + - negotiator: unknown fetch.negotiationAlgorithm should error out

562 + - Merge branch 'jt/fetch-nego-tip' into ab/fetch-nego

563

497 - It is too easy to misuse system API functions such as strcat();

498 - these selected functions are now forbidden in this codebase and

499 - will cause a compilation failure.

564 + Update to a few other topics.

565

566 Will merge to 'next'.

567

568

504 -* jk/size-t (2018-07-24) 6 commits

505 - - strbuf_humanise: use unsigned variables

506 - - pass st.st_size as hint for strbuf_readlink()

507 - - strbuf_readlink: use ssize_t

508 - - strbuf: use size_t for length in intermediate variables

509 - - reencode_string: use size_t for string lengths

510 - - reencode_string: use st_add/st_mult helpers

569 +* ab/fetch-tags-noclobber (2018-07-31) 10 commits

570 + - fetch: stop clobbering existing tags without --force

571 + - pull doc: fix a long-standing grammar error

572 + - fetch tests: add a test clobbering tag behavior

573 + - fetch tests: correct a comment "remove it" -> "remove them"

574 + - push doc: correct lies about how push refspecs work

575 + - push tests: assert re-pushing annotated tags

576 + - push tests: add more testing for forced tag pushing

577 + - push tests: fix logic error in "push" test assertion

578 + - push tests: remove redundant 'git push' invocation

579 + - fetch tests: change "Tag" test tag to "testTag"

580 +

581 + "git fetch" used to apply the same "fast-forward" rule and allow

582 + tags to move without "--force" option, which made little sense,

583 + which has been corrected.

584

512 - Code clean-up to use size_t/ssize_t when they are the right type.

585 + Expecting a reroll.

586 + cf. <xmqq4lgfcn5a.fsf@gitster-ct.c.googlers.com>

587 + cf. <xmqqzhy7b7v9.fsf@gitster-ct.c.googlers.com>

588

514 - Will merge to 'next'.

589

590 +* bp/checkout-new-branch-optim (2018-07-31) 1 commit

591 + - checkout: optimize "git checkout -b <new_branch>"

592

517 -* js/t7406-recursive-submodule-update-order-fix (2018-07-23) 1 commit

518 - - t7406: avoid failures solely due to timing issues

593 + "git checkout -b newbranch [HEAD]" should not have to do as much as

594 + checking out a commit different from HEAD. An attempt is made to

595 + optimize this special case.

596 +

597 + Waiting for review comments to be responded.

598 + cf. <CACsJy8DMEMsDnKZc65K-0EJcm2udXZ7OKY=xoFmX4COM0dSH=g@mail.gmail.com>

599 +

600 +

601 +* es/mw-to-git-chain-fix (2018-07-31) 1 commit

602 + - mw-to-git/t9360: fix broken &&-chain

603

604 Test fix.

605

522 - Will merge to 'next' and then to 'master'.

606 + Will merge to 'next'.

607

608

525 -* js/vscode (2018-07-23) 9 commits

526 - - vscode: let cSpell work on commit messages, too

527 - - vscode: add a dictionary for cSpell

528 - - vscode: use 8-space tabs, no trailing ws, etc for Git's source code

529 - - vscode: wrap commit messages at column 72 by default

530 - - vscode: only overwrite C/C++ settings

531 - - mingw: define WIN32 explicitly

532 - - cache.h: extract enum declaration from inside a struct declaration

533 - - vscode: hard-code a couple defines

534 - - contrib: add a script to initialize VS Code configuration

609 +* jk/merge-subtree-heuristics (2018-08-02) 1 commit

610 + - score_trees(): fix iteration over trees with missing entries

611

536 - Add a script (in contrib/) to help users of VSCode work better with

537 - our codebase.

612 + The automatic tree-matching in "git merge -s subtree" was broken 5

613 + years ago and nobody has noticed since then, which is now fixed.

614

539 - Stuck in review?

540 - cf. <20180723165719.GA16420@aiede.svl.corp.google.com>

541 - cf. <20180723174108.GA9285@aiede.svl.corp.google.com>

615 + Will merge to 'next'.

616

617

544 -* jt/tag-following-with-proto-v2-fix (2018-07-24) 2 commits

545 - - fetch: send "refs/tags/" prefix upon CLI refspecs

546 - - t5702: test fetch with multiple refspecs at a time

618 +* jt/connectivity-check-after-unshallow (2018-08-01) 1 commit

619 + - fetch-pack: unify ref in and out param

620

548 - The wire-protocol v2 relies on the client to send "ref prefixes" to

549 - limit the bandwidth spent on the initial ref advertisement. "git

550 - fetch $remote branch:branch" that asks tags that point into the

551 - history leading to the "branch" automatically followed sent to

552 - narrow prefix and broke the tag following, which has been fixed.

621 + Recent update to the transport layer broke ref updates after "git

622 + fetch", which is now fixed.

623

624 Will merge to 'next'.

625

626

557 -* nd/pack-deltify-regression-fix (2018-07-23) 1 commit

558 - - pack-objects: fix performance issues on packing large deltas

627 +* jt/refspec-dwim-precedence-fix (2018-08-02) 1 commit

628 + - remote: make refspec follow the same disambiguation rule as local refs

629

560 - In a recent update in 2.18 era, "git pack-objects" started

561 - producing a larger than necessary packfiles by missing

562 - opportunities to use large deltas.

630 + "git fetch $there refs/heads/s" ought to fetch the tip of the

631 + branch 's', but when "refs/heads/refs/heads/s", i.e. a branch whose

632 + name is "refs/heads/s" exists at the same time, fetched that one

633 + instead by mistake. This has been corrected to honor the usual

634 + disambiguation rules for abbreviated refnames.

635

564 - Will merge to and cook in 'next'.

636 + Will merge to 'next'.

637

638

567 -* sb/trailers-docfix (2018-07-20) 1 commit

568 - - Documentation/git-interpret-trailers: explain possible values

639 +* nd/clone-case-smashing-warning (2018-07-31) 1 commit

640 + - clone: report duplicate entries on case-insensitive filesystems

641

570 - Doc update.

642 + Running "git clone" against a project that contain two files with

643 + pathnames that differ only in cases on a case insensitive

644 + filesystem would result in one of the files lost because the

645 + underlying filesystem is incapable of holding both at the same

646 + time. An attempt is made to detect such a case and warn.

647

572 - Will merge to 'next' and then to 'master'.

648 + Discussion getting petered out.

649 + Doing this portably and extending it to UTF-8 normalization issue

650 + HFS+ has would be costly.

651

652 + cf. <20180728095659.GA21450@sigill.intra.peff.net>

653 + cf. <xmqq1sbh7phx.fsf@gitster-ct.c.googlers.com>

654

575 -* sg/coccicheck-updates (2018-07-23) 5 commits

576 - - coccinelle: extract dedicated make target to clean Coccinelle's results

577 - - coccinelle: put sane filenames into output patches

578 - - coccinelle: exclude sha1dc source files from static analysis

579 - - coccinelle: use $(addsuffix) in 'coccicheck' make target

580 - - coccinelle: mark the 'coccicheck' make target as .PHONY

655

582 - Update the way we use Coccinelle to find out-of-style code that

583 - need to be modernised.

656 +* nd/unpack-trees-with-cache-tree (2018-07-31) 4 commits

657 + - unpack-trees: cheaper index update when walking by cache-tree

658 + - unpack-trees: reduce malloc in cache-tree walk

659 + - unpack-trees: optimize walking same trees with cache-tree

660 + - unpack-trees.c: add performance tracing

661

585 - Will merge to 'next'.

662 + The unpack_trees() API used in checking out a branch and merging

663 + walks one or more trees along with the index. When the cache-tree

664 + in the index tells us that we are walking a tree whose flattened

665 + contents is known (i.e. matches a span in the index), as linearly

666 + scanning a span in the index is much more efficient than having to

667 + open tree objects recursively and listing their entries, the walk

668 + can be optimized, which is done in this topic.

669

670

588 -* sg/fast-import-dump-refs-on-checkpoint-fix (2018-07-20) 1 commit

589 - - t9300: wait for background fast-import process to die after killing it

671 +* rs/remote-mv-leakfix (2018-08-01) 1 commit

672 + - remote: clear string_list after use in mv()

673

591 - Test update.

674 + Leakfix.

675

593 - Will merge to 'next' and then to 'master'.

676 + Will merge to 'next'.

677

678

596 -* sg/travis-cocci-diagnose-failure (2018-07-23) 2 commits

597 - - travis-ci: fail if Coccinelle static analysis found something to transform

598 - - travis-ci: run Coccinelle static analysis with two parallel jobs

679 +* sb/config-write-fix (2018-08-01) 3 commits

680 + - git-config: document accidental multi-line setting in deprecated syntax

681 + - config: fix case sensitive subsection names on writing

682 + - t1300: document current behavior of setting options

683

600 - Update the way we run static analysis tool at TravisCI to make it

601 - easier to use its findings.

684 + Recent update to "git config" broke updating variable in a

685 + subsection, which has been corrected.

686

603 - Will merge to 'next' and then to 'master'.

687 + Not quite?

688 + cf. <xmqq4lgc1rbv.fsf@gitster-ct.c.googlers.com>

689

690

606 -* ab/newhash-is-sha256 (2018-07-25) 1 commit

607 - - doc hash-function-transition: note the lack of a changelog

691 +* sb/range-diff-colors (2018-08-01) 9 commits

692 + - fixup! t3206: add color test for range-diff --dual-color

693 + - diff.c: rewrite emit_line_0 more understandably

694 + - diff.c: compute reverse locally in emit_line_0

695 + - diff: use emit_line_0 once per line

696 + - diff.c: add set_sign to emit_line_0

697 + - diff.c: reorder arguments for emit_line_ws_markup

698 + - diff.c: simplify caller of emit_line_0

699 + - t3206: add color test for range-diff --dual-color

700 + - test_decode_color: understand FAINT and ITALIC

701 + (this branch uses js/range-diff; is tangled with es/format-patch-rangediff.)

702

609 - Documentation update.

703

611 - Waiting for another attempt for the second part.

704 +* sg/t1404-update-ref-test-timeout (2018-08-01) 1 commit

705 + - t1404: increase core.packedRefsTimeout to avoid occasional test failure

706

707 + An attempt to unflake a test a bit.

708

614 -* bb/redecl-enum-fix (2018-07-25) 1 commit

615 - - packfile: drop a repeated enum declaration

709

617 - Compilation fix.

710 +* sg/travis-retrieve-trash-upon-failure (2018-08-01) 1 commit

711 + - travis-ci: include the trash directories of failed tests in the trace log

712

713 + The Travis CI scripts were taught to ship back the test data from

714 + failed tests.

715

620 -* es/diff-color-move-fix (2018-07-25) 1 commit

621 - - diff: --color-moved: rename "dimmed_zebra" to "dimmed-zebra"

716 + Will merge to 'next'.

717

623 - One of the "diff --color-moved" mode "dimmed_zebra" that was named

624 - in an unusual way has been deprecated and replaced by

625 - "dimmed-zebra".

718

627 - Will merge to 'next' and then to 'master'.

719 +* jt/fetch-follow-fix (2018-08-01) 1 commit

720 + - fetch-pack: unify ref in and out param

721

722 + "git fetch" sometimes failed to update the remote-tracking refs,

723 + which has been corrected.

724

630 -* jh/structured-logging (2018-07-25) 25 commits

631 - - structured-logging: add config data facility

632 - - structured-logging: t0420 tests for interacitve child_summary

633 - - structured-logging: t0420 tests for child process detail events

634 - - structured-logging: add child process classification

635 - - structured-logging: add detail-events for child processes

636 - - structured-logging: add structured logging to remote-curl

637 - - structured-logging: t0420 tests for aux-data

638 - - structured-logging: add aux-data for size of sparse-checkout file

639 - - structured-logging: add aux-data for index size

640 - - structured-logging: add aux-data facility

641 - - structured-logging: t0420 tests for timers

642 - - structured-logging: add timer around preload_index

643 - - structured-logging: add timer around wt-status functions

644 - - structured-logging: add timer around do_write_index

645 - - structured-logging: add timer around do_read_index

646 - - structured-logging: add timer facility

647 - - structured-logging: add detail-event for lazy_init_name_hash

648 - - structured-logging: add detail-event facility

649 - - structured-logging: t0420 basic tests

650 - - structured-logging: set sub_command field for checkout command

651 - - structured-logging: set sub_command field for branch command

652 - - structured-logging: add session-id to log events

653 - - structured-logging: add structured logging framework

654 - - structured-logging: add STRUCTURED_LOGGING=1 to Makefile

655 - - structured-logging: design document

656 - (this branch uses jh/json-writer.)

725 + Will merge to 'next'.

726 +

727 +

728 +* ab/sha1dc (2018-08-02) 1 commit

729 + - sha1dc: update from upstream

730 +

731 + AIX portability update for SHADC hash, imported from upstream.

732

658 - X-Gah.

733 + Will merge to 'next'.

734 +

735 +

736 +* es/want-color-fd-defensive (2018-08-02) 1 commit

737 + - color: protect against out-of-bounds array access/assignment

738 +

739 + Futureproofing a helper function that can easily misused.

740 +

741 + Will merge to 'next'.

742 +

743 +

744 +* pw/rebase-i-author-script-fix (2018-08-02) 2 commits

745 + - sequencer: fix quoting in write_author_script

746 + - sequencer: handle errors in read_author_ident()

747 + (this branch uses es/rebase-i-author-script-fix.)

748 +

749 + Recent "git rebase -i" update started to write bogusly formatted

750 + author-script, with a matching broken reading code. These are

751 + being fixed.

752 +

753 + Undecided.

754 + Is it the list consensus to favor this "with extra code, read the

755 + script written by bad writer" approach?

756 +

757 +

758 +* rs/parse-opt-lithelp (2018-08-02) 6 commits

759 + - parse-options: automatically infer PARSE_OPT_LITERAL_ARGHELP

760 + - shortlog: correct option help for -w

761 + - send-pack: specify --force-with-lease argument help explicitly

762 + - pack-objects: specify --index-version argument help explicitly

763 + - difftool: remove angular brackets from argument help

764 + - add, update-index: fix --chmod argument help

765 +

766 + The parse-options machinery learned to refrain from enclosing

767 + placeholder string inside a "<bra" and "ket>" pair automatically

768 + without PARSE_OPT_LITERAL_ARGHELP. Existing help text for option

769 + arguments that are not formatted correctly have been identified and

770 + fixed.

771 +

772 + Will merge to 'next'.

773

774 --------------------------------------------------

775 [Stalled]

776

663 -* pw/add-p-select (2018-03-16) 3 commits

664 - - add -p: optimize line selection for short hunks

665 - - add -p: allow line selection to be inverted

666 - - add -p: select individual hunk lines

667 -

668 - "git add -p" interactive interface learned to let users choose

669 - individual added/removed lines to be used in the operation, instead

670 - of accepting or rejecting a whole hunk.

777 +* ma/wrapped-info (2018-05-28) 2 commits

778 + - usage: prefix all lines in `vreportf()`, not just the first

779 + - usage: extract `prefix_suffix_lines()` from `advise()`

780

672 - Expecting a reroll to reignite the discussion.

673 - cf. <9895c7b7-eac4-28c1-90c6-443acd1131b7@talktalk.net>

781 + An attempt to help making multi-line messages fed to warning(),

782 + error(), and friends more easily translatable.

783

784 + Will discard and wait for a cleaned-up rewrite.

785 + cf. <20180529213957.GF7964@sigill.intra.peff.net>

786

787 * hn/bisect-first-parent (2018-04-21) 1 commit

788 - bisect: create 'bisect_flags' parameter in find_bisection()

@@ -718,17 +829,6 @@ of the repositories listed at

829 in the thread above---we are still waiting for a reroll.

830

831

721 -* mk/http-backend-content-length (2018-06-11) 3 commits

722 - - http-backend: respect CONTENT_LENGTH for receive-pack

723 - - http-backend: respect CONTENT_LENGTH as specified by rfc3875

724 - - http-backend: cleanup writing to child process

725 -

726 - The http-backend (used for smart-http transport) used to slurp the

727 - whole input until EOF, without paying attention to CONTENT_LENGTH

728 - that is supplied in the environment and instead expecting the Web

729 - server to close the input stream. This has been fixed.

730 -

731 -

832 * jk/drop-ancient-curl (2017-08-09) 5 commits

833 - http: #error on too-old curl

834 - curl: remove ifdef'd code never used with curl >=7.19.4

@@ -754,457 +854,552 @@ of the repositories listed at

854 --------------------------------------------------

855 [Cooking]

856

757 -* am/sequencer-author-script-fix (2018-07-18) 1 commit

758 - - sequencer.c: terminate the last line of author-script properly

759 -

760 - The author-script that records the author information created by

761 - the sequencer machinery lacked the closing single quote on the last

762 - entry.

763 -

764 - Fixing this alone may or may not break the reader that may have

765 - been compensating for this bogus writer. I think I saw another fix

766 - to the same source file posted today---if we are fixing, we should

767 - fix them all at the same time to keep the reader and the writer in

768 - sync.

769 -

770 -

771 -* bc/sequencer-export-work-tree-as-well (2018-07-16) 1 commit

772 - (merged to 'next' on 2018-07-24 at 0b83ade721)

773 - + sequencer: pass absolute GIT_WORK_TREE to exec commands

857 +* es/diff-color-moved-fix (2018-07-25) 1 commit

858 + (merged to 'next' on 2018-08-02 at 233bccfbfb)

859 + + diff: --color-moved: rename "dimmed_zebra" to "dimmed-zebra"

860

775 - "git rebase" started exporting GIT_DIR environment variable and

776 - exposing it to hook scripts when part of it got rewritten in C.

777 - Instead of matching the old scripted Porcelains' behaviour,

778 - compensate by also exporting GIT_WORK_TREE environment as well to

779 - lessen the damage. This can harm existing hooks that want to

780 - operate on different repository, but the current behaviour is

781 - already broken for them anyway.

861 + One of the "diff --color-moved" mode "dimmed_zebra" that was named

862 + in an unusual way has been deprecated and replaced by

863 + "dimmed-zebra".

864

865 Will merge to 'master'.

866

867

786 -* bp/test-drop-caches-for-windows (2018-07-12) 1 commit

787 - (merged to 'next' on 2018-07-24 at 257bb336c6)

788 - + handle lower case drive letters on Windows

868 +* pw/add-p-select (2018-07-26) 4 commits

869 + - add -p: optimize line selection for short hunks

870 + - add -p: allow line selection to be inverted

871 + - add -p: select modified lines correctly

872 + - add -p: select individual hunk lines

873

790 - A test helper update for Windows.

874 + "git add -p" interactive interface learned to let users choose

875 + individual added/removed lines to be used in the operation, instead

876 + of accepting or rejecting a whole hunk.

877

792 - Will merge to 'master'.

878 + Will merge to and cook in 'next'.

879

880 + I found the feature to be hard to explain, and may result in more

881 + end-user complaints, but let's see.

882

795 -* en/abort-df-conflict-fixes (2018-07-16) 2 commits

796 - - read-cache: fix directory/file conflict handling in read_index_unmerged()

797 - - t1015: demonstrate directory/file conflict recovery failures

883

799 - "git merge --abort" etc. did not clean things up properly when

800 - there were conflicted entries in certain order that are involved

801 - in D/F conflicts. This has been corrected.

884 +* mk/http-backend-content-length (2018-07-30) 4 commits

885 + - t5562: avoid non-portable "export FOO=bar" construct

886 + - http-backend: respect CONTENT_LENGTH for receive-pack

887 + - http-backend: respect CONTENT_LENGTH as specified by rfc3875

888 + - http-backend: cleanup writing to child process

889

803 - This may have to be rebased on an older maintenance track before

804 - moving forward.

890 + The http-backend (used for smart-http transport) used to slurp the

891 + whole input until EOF, without paying attention to CONTENT_LENGTH

892 + that is supplied in the environment and instead expecting the Web

893 + server to close the input stream. This has been fixed.

894

895 + Will merge to 'next'.

896

807 -* es/chain-lint-in-subshell (2018-07-17) 10 commits

808 - (merged to 'next' on 2018-07-24 at 9370bbdfaf)

809 - + t/chainlint: add chainlint "specialized" test cases

810 - + t/chainlint: add chainlint "complex" test cases

811 - + t/chainlint: add chainlint "cuddled" test cases

812 - + t/chainlint: add chainlint "loop" and "conditional" test cases

813 - + t/chainlint: add chainlint "nested subshell" test cases

814 - + t/chainlint: add chainlint "one-liner" test cases

815 - + t/chainlint: add chainlint "whitespace" test cases

816 - + t/chainlint: add chainlint "basic" test cases

817 - + t/Makefile: add machinery to check correctness of chainlint.sed

818 - + t/test-lib: teach --chain-lint to detect broken &&-chains in subshells

819 - (this branch uses es/test-fixes.)

897

821 - Look for broken "&&" chains that are hidden in subshell, many of

822 - which have been found and corrected.

898 +* ds/commit-graph-with-grafts (2018-07-19) 8 commits

899 + (merged to 'next' on 2018-08-02 at 0ee624e329)

900 + + commit-graph: close_commit_graph before shallow walk

901 + + commit-graph: not compatible with uninitialized repo

902 + + commit-graph: not compatible with grafts

903 + + commit-graph: not compatible with replace objects

904 + + test-repository: properly init repo

905 + + commit-graph: update design document

906 + + refs.c: upgrade for_each_replace_ref to be a each_repo_ref_fn callback

907 + + refs.c: migrate internal ref iteration to pass thru repository argument

908 +

909 + The recently introduced commit-graph auxiliary data is incompatible

910 + with mechanisms such as replace & grafts that "breaks" immutable

911 + nature of the object reference relationship. Disable optimizations

912 + based on its use (and updating existing commit-graph) when these

913 + incompatible features are in use in the repository.

914

915 Will merge to 'master'.

916

917

827 -* hs/gpgsm (2018-07-20) 7 commits

828 - - gpg-interface t: extend the existing GPG tests with GPGSM

829 - - gpg-interface: introduce new signature format "x509" using gpgsm

830 - - gpg-interface: introduce new config to select per gpg format program

831 - - gpg-interface: do not hardcode the key string len anymore

832 - - gpg-interface: introduce an abstraction for multiple gpg formats

833 - - t/t7510: check the validation of the new config gpg.format

834 - - gpg-interface: add new config to select how to sign a commit

918 +* jk/core-use-replace-refs (2018-07-18) 3 commits

919 + (merged to 'next' on 2018-08-02 at 90fb6b1056)

920 + + add core.usereplacerefs config option

921 + + check_replace_refs: rename to read_replace_refs

922 + + check_replace_refs: fix outdated comment

923

836 - Teach "git tag -s" etc. a few configuration varaibles (gpg.format

837 - that can be set to "openpgp" or "x509", and gpg.<format>.program

838 - that is used to specify what program to use to deal with the format)

839 - to allow x.509 certs with CMS via "gpgsm" to be used instead of

840 - openpgp via "gnupg".

924 + A new configuration variable core.usereplacerefs has been added,

925 + primarily to help server installations that want to ignore the

926 + replace mechanism altogether.

927

842 - Will merge to 'next'.

928 + Will merge to 'master'.

929

930

845 -* jk/has-uncommitted-changes-fix (2018-07-11) 1 commit

846 - (merged to 'next' on 2018-07-24 at 2ea14c0afb)

847 - + has_uncommitted_changes(): fall back to empty tree

931 +* nd/i18n (2018-07-23) 23 commits

932 + (merged to 'next' on 2018-08-02 at 904a22a5d1)

933 + + transport-helper.c: mark more strings for translation

934 + + transport.c: mark more strings for translation

935 + + sha1-file.c: mark more strings for translation

936 + + sequencer.c: mark more strings for translation

937 + + replace-object.c: mark more strings for translation

938 + + refspec.c: mark more strings for translation

939 + + refs.c: mark more strings for translation

940 + + pkt-line.c: mark more strings for translation

941 + + object.c: mark more strings for translation

942 + + exec-cmd.c: mark more strings for translation

943 + + environment.c: mark more strings for translation

944 + + dir.c: mark more strings for translation

945 + + convert.c: mark more strings for translation

946 + + connect.c: mark more strings for translation

947 + + config.c: mark more strings for translation

948 + + commit-graph.c: mark more strings for translation

949 + + builtin/replace.c: mark more strings for translation

950 + + builtin/pack-objects.c: mark more strings for translation

951 + + builtin/grep.c: mark strings for translation

952 + + builtin/config.c: mark more strings for translation

953 + + archive-zip.c: mark more strings for translation

954 + + archive-tar.c: mark more strings for translation

955 + + Update messages in preparation for i18n

956

849 - "git pull --rebase" on a corrupt HEAD caused a segfault. In

850 - general we substitute an empty tree object when running the in-core

851 - equivalent of the diff-index command, and the codepath has been

852 - corrected to do so as well to fix this issue.

957 + Many more strings are prepared for l10n.

958

959 Will merge to 'master'.

960

961

857 -* jm/send-email-tls-auth-on-batch (2018-07-16) 1 commit

858 - (merged to 'next' on 2018-07-24 at fb3e653f44)

859 - + send-email: fix tls AUTH when sending batch

962 +* sb/histogram-less-memory (2018-07-23) 4 commits

963 + (merged to 'next' on 2018-08-02 at cfb02aa3b5)

964 + + xdiff/histogram: remove tail recursion

965 + + xdiff/xhistogram: move index allocation into find_lcs

966 + + xdiff/xhistogram: factor out memory cleanup into free_index()

967 + + xdiff/xhistogram: pass arguments directly to fall_back_to_classic_diff

968

861 - "git send-email" when using in a batched mode that limits the

862 - number of messages sent in a single SMTP session lost the contents

863 - of the variable used to choose between tls/ssl, unable to send the

864 - second and later batches, which has been fixed.

969 + "git diff --histogram" had a bad memory usage pattern, which has

970 + been rearranged to reduce the peak usage.

971

972 Will merge to 'master'.

973

868 - This is marked to be merged to 'next' already, but I do not mind

869 - getting an updated version with an improved log message before that

870 - happens.

871 -

974

873 -* jn/gc-auto (2018-07-17) 3 commits

874 - - gc: do not return error for prior errors in daemonized mode

875 - - gc: exit with status 128 on failure

876 - - gc: improve handling of errors reading gc.log

975 +* bb/make-developer-pedantic (2018-07-25) 1 commit

976 + (merged to 'next' on 2018-08-02 at c738a84b7e)

977 + + Makefile: add a DEVOPTS flag to get pedantic compilation

978

878 - "gc --auto" ended up calling exit(-1) upon error, which has been

879 - corrected to use exit(1). Also the error reporting behaviour when

880 - daemonized has been updated to exit with zero status when stopping

881 - due to a previously discovered error (which implies there is no

882 - point running gc to improve the situation); we used to exit with

883 - failure in such a case.

979 + "make DEVELOPER=1 DEVOPTS=pedantic" allows developers to compile

980 + with -pedantic option, which may catch more problematic program

981 + constructs and potential bugs.

982

885 - Stuck in review?

886 - cf. <20180717201348.GD26218@sigill.intra.peff.net>

983 + Will merge to 'master'.

984

985

889 -* js/rebase-merge-octopus (2018-07-11) 3 commits

890 - (merged to 'next' on 2018-07-24 at 14ad8699de)

891 - + rebase --rebase-merges: adjust man page for octopus support

892 - + rebase --rebase-merges: add support for octopus merges

893 - + merge: allow reading the merge commit message from a file

986 +* bw/clone-ref-prefixes (2018-07-20) 1 commit

987 + (merged to 'next' on 2018-08-02 at c8ad140ab0)

988 + + clone: send ref-prefixes when using protocol v2

989

895 - "git rebase --rebase-merges" mode now handles octopus merges as

896 - well.

990 + The wire-protocol v2 relies on the client to send "ref prefixes" to

991 + limit the bandwidth spent on the initial ref advertisement. "git

992 + clone" when learned to speak v2 forgot to do so, which has been

993 + corrected.

994

995 Will merge to 'master'.

996

997

901 -* jt/fetch-negotiator-skipping (2018-07-16) 1 commit

902 - (merged to 'next' on 2018-07-24 at 8e25a49405)

903 - + negotiator/skipping: skip commits during fetch

904 - (this branch uses jt/fetch-pack-negotiator; is tangled with jt/fetch-nego-tip.)

998 +* bw/fetch-pack-i18n (2018-07-23) 1 commit

999 + (merged to 'next' on 2018-08-02 at df72001755)

1000 + + fetch-pack: mark die strings for translation

1001

906 - Add a server-side knob to skip commits in exponential/fibbonacci

907 - stride in an attempt to cover wider swath of history with a smaller

908 - number of iterations, potentially accepting a larger packfile

909 - transfer, instead of going back one commit a time during common

910 - ancestor discovery during the "git fetch" transaction.

1002 + i18n updates.

1003

1004 Will merge to 'master'.

1005

1006

915 -* jt/tags-to-promised-blobs-fix (2018-07-16) 2 commits

916 - (merged to 'next' on 2018-07-24 at 8d7e78a671)

917 - + tag: don't warn if target is missing but promised

918 - + revision: tolerate promised targets of tags

1007 +* bw/protocol-v2 (2018-07-24) 1 commit

1008 + (merged to 'next' on 2018-08-02 at f4076b3e94)

1009 + + pack-protocol: mention and point to docs for protocol v2

1010

920 - The lazy clone support had a few places where missing but promised

921 - objects were not correctly tolerated, which have been fixed.

1011 + Doc update.

1012

1013 Will merge to 'master'.

1014

1015

926 -* sb/submodule-update-in-c (2018-07-18) 6 commits

927 - - submodule--helper: introduce new update-module-mode helper

928 - - builtin/submodule--helper: factor out method to update a single submodule

929 - - builtin/submodule--helper: store update_clone information in a struct

930 - - builtin/submodule--helper: factor out submodule updating

931 - - git-submodule.sh: rename unused variables

932 - - git-submodule.sh: align error reporting for update mode to use path

1016 +* ds/reachable (2018-07-20) 18 commits

1017 + - commit-reach: use can_all_from_reach

1018 + - commit-reach: make can_all_from_reach... linear

1019 + - commit-reach: replace ref_newer logic

1020 + - test-reach: test commit_contains

1021 + - test-reach: test can_all_from_reach_with_flags

1022 + - test-reach: test reduce_heads

1023 + - test-reach: test get_merge_bases_many

1024 + - test-reach: test is_descendant_of

1025 + - test-reach: test in_merge_bases

1026 + - test-reach: create new test tool for ref_newer

1027 + - commit-reach: move can_all_from_reach_with_flags

1028 + - upload-pack: generalize commit date cutoff

1029 + - upload-pack: refactor ok_to_give_up()

1030 + - upload-pack: make reachable() more generic

1031 + - commit-reach: move commit_contains from ref-filter

1032 + - commit-reach: move ref_newer from remote.c

1033 + - commit.h: remove method declarations

1034 + - commit-reach: move walk methods from commit.c

1035

934 - "git submodule update" is getting rewritten piece-by-piece into C.

1036 + The code for computing history reachability has been shuffled,

1037 + obtained a bunch of new tests to cover them, and then being

1038 + improved.

1039

936 - It seems to pass its own self-tests standalone, but seems to break

937 - horribly when merged to 'pu'.

1040 + Will merge to and cook in 'next'.

1041

1042

940 -* sg/httpd-test-unflake (2018-07-12) 3 commits

941 - (merged to 'next' on 2018-07-24 at b7df820256)

942 - + t/lib-httpd: avoid occasional failures when checking access.log

943 - + t/lib-httpd: add the strip_access_log() helper function

944 - + t5541: clean up truncating access log

1043 +* en/merge-recursive-skip-fix (2018-07-27) 2 commits

1044 + - merge-recursive: preserve skip_worktree bit when necessary

1045 + - t3507: add a testcase showing failure with sparse checkout

1046

946 - httpd tests saw occasional breakage due to the way its access log

947 - gets inspected by the tests, which has been updated to make them

948 - less flaky.

1047 + When the sparse checkout feature is in use, "git cherry-pick" and

1048 + other mergy operations lost the skip_worktree bit when a path that

1049 + is excluded from checkout requires content level merge, which is

1050 + resolved as the same as the HEAD version, without materializing the

1051 + merge result in the working tree, which made the path appear as

1052 + deleted. This has been corrected by preserving the skip_worktree

1053 + bit (and not materializing the file in the working tree).

1054

950 - Will merge to 'master'.

1055 + Will merge to 'next'.

1056

1057

953 -* sl/commit-dry-run-with-short-output-fix (2018-07-17) 3 commits

954 - - commit: fix exit code for --short/--porcelain

955 - - wt-status: teach wt_status_collect about merges in progress

956 - - t7501: add merge conflict tests for dry run

1058 +* es/format-patch-interdiff (2018-07-23) 6 commits

1059 + - format-patch: allow --interdiff to apply to a lone-patch

1060 + - log-tree: show_log: make commentary block delimiting reusable

1061 + - interdiff: teach show_interdiff() to indent interdiff

1062 + - format-patch: teach --interdiff to respect -v/--reroll-count

1063 + - format-patch: add --interdiff option to embed diff in cover letter

1064 + - format-patch: allow additional generated content in make_cover_letter()

1065 + (this branch is used by es/format-patch-rangediff.)

1066 +

1067 + "git format-patch" learned a new "--interdiff" option to explain

1068 + the difference between this version and the previous atttempt in

1069 + the cover letter (or after the tree-dashes as a comment).

1070 +

1071 + Stuck in review?

1072 + cf. <CAPig+cSuYUYSPTuKx08wcmQM-G12_-W2T4BS07fA=6grM1b8Gw@mail.gmail.com>

1073

958 - "git commit --dry-run" gave a correct exit status even during a

959 - conflict resolution toward a merge, but it did not with the

960 - "--short" option, which has been corrected.

1074 +

1075 +* es/format-patch-rangediff (2018-07-30) 10 commits

1076 + - format-patch: allow --range-diff to apply to a lone-patch

1077 + - format-patch: add --creation-factor tweak for --range-diff

1078 + - format-patch: teach --range-diff to respect -v/--reroll-count

1079 + - format-patch: extend --range-diff to accept revision range

1080 + - format-patch: add --range-diff option to embed diff in cover letter

1081 + - range-diff: relieve callers of low-level configuration burden

1082 + - range-diff: publish default creation factor

1083 + - range-diff: respect diff_option.file rather than assuming 'stdout'

1084 + - Merge branch 'es/format-patch-interdiff' into es/format-patch-rangediff

1085 + - Merge branch 'js/range-diff' into es/format-patch-rangediff

1086 + (this branch uses es/format-patch-interdiff and js/range-diff; is tangled with sb/range-diff-colors.)

1087 +

1088 + "git format-patch" learned a new "--range-diff" option to explain

1089 + the difference between this version and the previous atttempt in

1090 + the cover letter (or after the tree-dashes as a comment).

1091 +

1092 + Need to wait for the prereq topics to solidify a bit more.

1093 +

1094 +

1095 +* jk/banned-function (2018-07-26) 5 commits

1096 + - banned.h: mark strncpy() as banned

1097 + - banned.h: mark sprintf() as banned

1098 + - banned.h: mark strcat() as banned

1099 + - automatically ban strcpy()

1100 + - Merge branch 'sb/blame-color' into jk/banned-function

1101 +

1102 + It is too easy to misuse system API functions such as strcat();

1103 + these selected functions are now forbidden in this codebase and

1104 + will cause a compilation failure.

1105

1106 Will merge to 'next'.

1107

1108

965 -* tg/rerere (2018-07-16) 11 commits

966 - - rerere: recalculate conflict ID when unresolved conflict is committed

967 - - rerere: teach rerere to handle nested conflicts

968 - - rerere: return strbuf from handle path

969 - - rerere: factor out handle_conflict function

970 - - rerere: only return whether a path has conflicts or not

971 - - rerere: fix crash when conflict goes unresolved

972 - - rerere: add documentation for conflict normalization

973 - - rerere: mark strings for translation

974 - - rerere: wrap paths in output in sq

975 - - rerere: lowercase error messages

976 - - rerere: unify error messages when read_cache fails

1109 +* jk/size-t (2018-07-24) 6 commits

1110 + (merged to 'next' on 2018-08-02 at 6f861e05f0)

1111 + + strbuf_humanise: use unsigned variables

1112 + + pass st.st_size as hint for strbuf_readlink()

1113 + + strbuf_readlink: use ssize_t

1114 + + strbuf: use size_t for length in intermediate variables

1115 + + reencode_string: use size_t for string lengths

1116 + + reencode_string: use st_add/st_mult helpers

1117

1118 + Code clean-up to use size_t/ssize_t when they are the right type.

1119

979 -* jk/ui-color-always-to-auto (2018-07-18) 1 commit

980 - - Documentation: fix --color option formatting

1120 + Will merge to 'master'.

1121

982 - Doc formatting fix.

1122

984 - Will merge to 'next' and then to 'master'.

1123 +* js/t7406-recursive-submodule-update-order-fix (2018-07-23) 1 commit

1124 + (merged to 'next' on 2018-08-02 at 217ea36a37)

1125 + + t7406: avoid failures solely due to timing issues

1126

1127 + Test fix.

1128

987 -* jh/json-writer (2018-07-16) 1 commit

988 - - json_writer: new routines to create JSON data

989 - (this branch is used by jh/structured-logging.)

1129 + Will merge to 'master'.

1130

991 - Preparatory code to later add json output for telemetry data.

1131 +

1132 +* js/vscode (2018-07-30) 9 commits

1133 + - vscode: let cSpell work on commit messages, too

1134 + - vscode: add a dictionary for cSpell

1135 + - vscode: use 8-space tabs, no trailing ws, etc for Git's source code

1136 + - vscode: wrap commit messages at column 72 by default

1137 + - vscode: only overwrite C/C++ settings

1138 + - mingw: define WIN32 explicitly

1139 + - cache.h: extract enum declaration from inside a struct declaration

1140 + - vscode: hard-code a couple defines

1141 + - contrib: add a script to initialize VS Code configuration

1142 +

1143 + Add a script (in contrib/) to help users of VSCode work better with

1144 + our codebase.

1145

1146 Will merge to 'next'.

1147

1148

996 -* ag/rebase-i-in-c (2018-07-10) 13 commits

997 - - rebase -i: rewrite the rest of init_revisions_and_shortrevisions in C

998 - - rebase -i: implement the logic to initialize the variable $revision in C

999 - - rebase--interactive: remove unused modes and functions

1000 - - rebase--interactive: rewrite complete_action() in C

1001 - - sequencer: change the way skip_unnecessary_picks() returns its result

1002 - - sequencer: refactor append_todo_help() to write its message to a buffer

1003 - - rebase -i: rewrite checkout_onto() in C

1004 - - rebase -i: rewrite setup_reflog_action() in C

1005 - - sequencer: add a new function to silence a command, except if it fails

1006 - - rebase-interactive: rewrite the edit-todo functionality in C

1007 - - editor: add a function to launch the sequence editor

1008 - - rebase--interactive: rewrite append_todo_help() in C

1009 - - sequencer: make two functions and an enum from sequencer.c public

1149 +* jt/tag-following-with-proto-v2-fix (2018-07-24) 2 commits

1150 + (merged to 'next' on 2018-08-02 at d9eabdea95)

1151 + + fetch: send "refs/tags/" prefix upon CLI refspecs

1152 + + t5702: test fetch with multiple refspecs at a time

1153

1011 - Piecemeal rewrite of the remaining "rebase -i" machinery in C.

1154 + The wire-protocol v2 relies on the client to send "ref prefixes" to

1155 + limit the bandwidth spent on the initial ref advertisement. "git

1156 + fetch $remote branch:branch" that asks tags that point into the

1157 + history leading to the "branch" automatically followed sent to

1158 + narrow prefix and broke the tag following, which has been fixed.

1159

1013 - A reroll (which is rumored to be quite good) exists, but hasn't

1014 - been picked up yet.

1160 + Will merge to 'master'.

1161

1162

1017 -* sb/object-store-lookup (2018-06-29) 33 commits

1018 - (merged to 'next' on 2018-07-24 at dd96e29376)

1019 - + commit.c: allow lookup_commit_reference to handle arbitrary repositories

1020 - + commit.c: allow lookup_commit_reference_gently to handle arbitrary repositories

1021 - + tag.c: allow deref_tag to handle arbitrary repositories

1022 - + object.c: allow parse_object to handle arbitrary repositories

1023 - + object.c: allow parse_object_buffer to handle arbitrary repositories

1024 - + commit.c: allow get_cached_commit_buffer to handle arbitrary repositories

1025 - + commit.c: allow set_commit_buffer to handle arbitrary repositories

1026 - + commit.c: migrate the commit buffer to the parsed object store

1027 - + commit-slabs: remove realloc counter outside of slab struct

1028 - + commit.c: allow parse_commit_buffer to handle arbitrary repositories

1029 - + tag: allow parse_tag_buffer to handle arbitrary repositories

1030 - + tag: allow lookup_tag to handle arbitrary repositories

1031 - + commit: allow lookup_commit to handle arbitrary repositories

1032 - + tree: allow lookup_tree to handle arbitrary repositories

1033 - + blob: allow lookup_blob to handle arbitrary repositories

1034 - + object: allow lookup_object to handle arbitrary repositories

1035 - + object: allow object_as_type to handle arbitrary repositories

1036 - + tag: add repository argument to deref_tag

1037 - + tag: add repository argument to parse_tag_buffer

1038 - + tag: add repository argument to lookup_tag

1039 - + commit: add repository argument to get_cached_commit_buffer

1040 - + commit: add repository argument to set_commit_buffer

1041 - + commit: add repository argument to parse_commit_buffer

1042 - + commit: add repository argument to lookup_commit

1043 - + commit: add repository argument to lookup_commit_reference

1044 - + commit: add repository argument to lookup_commit_reference_gently

1045 - + tree: add repository argument to lookup_tree

1046 - + blob: add repository argument to lookup_blob

1047 - + object: add repository argument to object_as_type

1048 - + object: add repository argument to parse_object_buffer

1049 - + object: add repository argument to lookup_object

1050 - + object: add repository argument to parse_object

1051 - + Merge branch 'sb/object-store-grafts' into sb/object-store-lookup

1052 - (this branch is used by ds/commit-graph-with-grafts, ds/reachable and jt/commit-graph-per-object-store.)

1163 +* nd/pack-deltify-regression-fix (2018-07-23) 1 commit

1164 + (merged to 'next' on 2018-08-02 at f3b2bf0fef)

1165 + + pack-objects: fix performance issues on packing large deltas

1166

1054 - lookup_commit_reference() and friends have been updated to find

1055 - in-core object for a specific in-core repository instance.

1167 + In a recent update in 2.18 era, "git pack-objects" started

1168 + producing a larger than necessary packfiles by missing

1169 + opportunities to use large deltas.

1170 +

1171 + Will cook in 'next'.

1172 +

1173 +

1174 +* sb/trailers-docfix (2018-07-20) 1 commit

1175 + (merged to 'next' on 2018-08-02 at ba348fafcd)

1176 + + Documentation/git-interpret-trailers: explain possible values

1177 +

1178 + Doc update.

1179

1180 Will merge to 'master'.

1181

1182

1060 -* bc/object-id (2018-07-16) 16 commits

1061 - (merged to 'next' on 2018-07-24 at 23680778a9)

1062 - + pretty: switch hard-coded constants to the_hash_algo

1063 - + sha1-file: convert constants to uses of the_hash_algo

1064 - + log-tree: switch GIT_SHA1_HEXSZ to the_hash_algo->hexsz

1065 - + diff: switch GIT_SHA1_HEXSZ to use the_hash_algo

1066 - + builtin/merge-recursive: make hash independent

1067 - + builtin/merge: switch to use the_hash_algo

1068 - + builtin/fmt-merge-msg: make hash independent

1069 - + builtin/update-index: simplify parsing of cacheinfo

1070 - + builtin/update-index: convert to using the_hash_algo

1071 - + refs/files-backend: use the_hash_algo for writing refs

1072 - + sha1-name: use the_hash_algo when parsing object names

1073 - + strbuf: allocate space with GIT_MAX_HEXSZ

1074 - + commit: express tree entry constants in terms of the_hash_algo

1075 - + hex: switch to using the_hash_algo

1076 - + tree-walk: replace hard-coded constants with the_hash_algo

1077 - + cache: update object ID functions for the_hash_algo

1183 +* sg/coccicheck-updates (2018-07-23) 5 commits

1184 + (merged to 'next' on 2018-08-02 at b5548ff3a9)

1185 + + coccinelle: extract dedicated make target to clean Coccinelle's results

1186 + + coccinelle: put sane filenames into output patches

1187 + + coccinelle: exclude sha1dc source files from static analysis

1188 + + coccinelle: use $(addsuffix) in 'coccicheck' make target

1189 + + coccinelle: mark the 'coccicheck' make target as .PHONY

1190

1079 - Conversion from uchar[40] to struct object_id continues.

1191 + Update the way we use Coccinelle to find out-of-style code that

1192 + need to be modernised.

1193

1194 Will merge to 'master'.

1195

1196

1084 -* en/dirty-merge-fixes (2018-07-11) 9 commits

1085 - (merged to 'next' on 2018-07-24 at 7b6ca3507c)

1086 - + merge: fix misleading pre-merge check documentation

1087 - + merge-recursive: enforce rule that index matches head before merging

1088 - + t6044: add more testcases with staged changes before a merge is invoked

1089 - + merge-recursive: fix assumption that head tree being merged is HEAD

1090 - + merge-recursive: make sure when we say we abort that we actually abort

1091 - + t6044: add a testcase for index matching head, when head doesn't match HEAD

1092 - + t6044: verify that merges expected to abort actually abort

1093 - + index_has_changes(): avoid assuming operating on the_index

1094 - + read-cache.c: move index_has_changes() from merge.c

1197 +* sg/fast-import-dump-refs-on-checkpoint-fix (2018-07-20) 1 commit

1198 + (merged to 'next' on 2018-08-02 at f5c05b5a2c)

1199 + + t9300: wait for background fast-import process to die after killing it

1200

1096 - The recursive merge strategy did not properly ensure there was no

1097 - change between HEAD and the index before performing its operation,

1098 - which has been corrected.

1201 + Test update.

1202

1203 Will merge to 'master'.

1204

1205

1103 -* en/t6036-merge-recursive-tests (2018-07-11) 6 commits

1104 - (merged to 'next' on 2018-07-24 at 75055cb6e1)

1105 - + t6036: add a failed conflict detection case: regular files, different modes

1106 - + t6036: add a failed conflict detection case with conflicting types

1107 - + t6036: add a failed conflict detection case with submodule add/add

1108 - + t6036: add a failed conflict detection case with submodule modify/modify

1109 - + t6036: add a failed conflict detection case with symlink add/add

1110 - + t6036: add a failed conflict detection case with symlink modify/modify

1206 +* sg/travis-cocci-diagnose-failure (2018-07-23) 2 commits

1207 + (merged to 'next' on 2018-08-02 at 54808a8778)

1208 + + travis-ci: fail if Coccinelle static analysis found something to transform

1209 + + travis-ci: run Coccinelle static analysis with two parallel jobs

1210

1112 - Tests to cover various conflicting cases have been added for

1113 - merge-recursive.

1211 + Update the way we run static analysis tool at TravisCI to make it

1212 + easier to use its findings.

1213

1214 Will merge to 'master'.

1215

1216

1118 -* en/t6036-recursive-corner-cases (2018-07-12) 2 commits

1119 - (merged to 'next' on 2018-07-24 at b7b3514ef4)

1120 - + t6036: fix broken && chain in sub-shell

1121 - + t6036: add lots of detail for directory/file conflicts in recursive case

1217 +* ab/newhash-is-sha256 (2018-07-26) 2 commits

1218 + - doc hash-function-transition: pick SHA-256 as NewHash

1219 + - doc hash-function-transition: note the lack of a changelog

1220

1123 - Tests to cover more D/F conflict cases have been added for

1124 - merge-recursive.

1221 + Documentation update.

1222 +

1223 + Will merge to 'next'.

1224 +

1225 +

1226 +* bb/redecl-enum-fix (2018-07-26) 1 commit

1227 + - packfile: ensure that enum object_type is defined

1228 +

1229 + Compilation fix.

1230 +

1231 + Will merge to 'next'.

1232 +

1233 +

1234 +* jh/structured-logging (2018-07-25) 25 commits

1235 + - structured-logging: add config data facility

1236 + - structured-logging: t0420 tests for interacitve child_summary

1237 + - structured-logging: t0420 tests for child process detail events

1238 + - structured-logging: add child process classification

1239 + - structured-logging: add detail-events for child processes

1240 + - structured-logging: add structured logging to remote-curl

1241 + - structured-logging: t0420 tests for aux-data

1242 + - structured-logging: add aux-data for size of sparse-checkout file

1243 + - structured-logging: add aux-data for index size

1244 + - structured-logging: add aux-data facility

1245 + - structured-logging: t0420 tests for timers

1246 + - structured-logging: add timer around preload_index

1247 + - structured-logging: add timer around wt-status functions

1248 + - structured-logging: add timer around do_write_index

1249 + - structured-logging: add timer around do_read_index

1250 + - structured-logging: add timer facility

1251 + - structured-logging: add detail-event for lazy_init_name_hash

1252 + - structured-logging: add detail-event facility

1253 + - structured-logging: t0420 basic tests

1254 + - structured-logging: set sub_command field for checkout command

1255 + - structured-logging: set sub_command field for branch command

1256 + - structured-logging: add session-id to log events

1257 + - structured-logging: add structured logging framework

1258 + - structured-logging: add STRUCTURED_LOGGING=1 to Makefile

1259 + - structured-logging: design document

1260 + (this branch uses jh/json-writer.)

1261 +

1262 + Will merge to 'next'.

1263 +

1264 +

1265 +* en/abort-df-conflict-fixes (2018-07-31) 2 commits

1266 + - read-cache: fix directory/file conflict handling in read_index_unmerged()

1267 + - t1015: demonstrate directory/file conflict recovery failures

1268 +

1269 + "git merge --abort" etc. did not clean things up properly when

1270 + there were conflicted entries in certain order that are involved

1271 + in D/F conflicts. This has been corrected.

1272 +

1273 + Will merge to 'next'.

1274 +

1275 +

1276 +* hs/gpgsm (2018-07-20) 7 commits

1277 + (merged to 'next' on 2018-08-02 at db28bffe4f)

1278 + + gpg-interface t: extend the existing GPG tests with GPGSM

1279 + + gpg-interface: introduce new signature format "x509" using gpgsm

1280 + + gpg-interface: introduce new config to select per gpg format program

1281 + + gpg-interface: do not hardcode the key string len anymore

1282 + + gpg-interface: introduce an abstraction for multiple gpg formats

1283 + + t/t7510: check the validation of the new config gpg.format

1284 + + gpg-interface: add new config to select how to sign a commit

1285 +

1286 + Teach "git tag -s" etc. a few configuration varaibles (gpg.format

1287 + that can be set to "openpgp" or "x509", and gpg.<format>.program

1288 + that is used to specify what program to use to deal with the format)

1289 + to allow x.509 certs with CMS via "gpgsm" to be used instead of

1290 + openpgp via "gnupg".

1291

1292 Will merge to 'master'.

1293

1294

1129 -* en/t6042-insane-merge-rename-testcases (2018-07-03) 3 commits

1130 - (merged to 'next' on 2018-07-24 at 65c80f72da)

1131 - + t6042: add testcase covering long chains of rename conflicts

1132 - + t6042: add testcase covering rename/rename(2to1)/delete/delete conflict

1133 - + t6042: add testcase covering rename/add/delete conflict type

1295 +* jn/gc-auto (2018-07-17) 3 commits

1296 + - gc: do not return error for prior errors in daemonized mode

1297 + - gc: exit with status 128 on failure

1298 + - gc: improve handling of errors reading gc.log

1299

1135 - Various glitches in the heuristics of merge-recursive strategy have

1136 - been documented in new tests.

1300 + "gc --auto" ended up calling exit(-1) upon error, which has been

1301 + corrected to use exit(1). Also the error reporting behaviour when

1302 + daemonized has been updated to exit with zero status when stopping

1303 + due to a previously discovered error (which implies there is no

1304 + point running gc to improve the situation); we used to exit with

1305 + failure in such a case.

1306 +

1307 + Stuck in review?

1308 + cf. <20180717201348.GD26218@sigill.intra.peff.net>

1309 +

1310 +

1311 +* sb/submodule-update-in-c (2018-07-18) 6 commits

1312 + - submodule--helper: introduce new update-module-mode helper

1313 + - builtin/submodule--helper: factor out method to update a single submodule

1314 + - builtin/submodule--helper: store update_clone information in a struct

1315 + - builtin/submodule--helper: factor out submodule updating

1316 + - git-submodule.sh: rename unused variables

1317 + - git-submodule.sh: align error reporting for update mode to use path

1318 +

1319 + "git submodule update" is getting rewritten piece-by-piece into C.

1320 +

1321 + Will merge to 'next'.

1322 +

1323 +

1324 +* sl/commit-dry-run-with-short-output-fix (2018-07-30) 4 commits

1325 + . commit: fix exit code when doing a dry run

1326 + . wt-status: teach wt_status_collect about merges in progress

1327 + . wt-status: rename commitable to committable

1328 + . t7501: add coverage for flags which imply dry runs

1329 +

1330 + "git commit --dry-run" gave a correct exit status even during a

1331 + conflict resolution toward a merge, but it did not with the

1332 + "--short" option, which has been corrected.

1333 +

1334 + Seems to break 7512, 3404 and 7060 in 'pu'.

1335 +

1336 +

1337 +* tg/rerere (2018-07-16) 11 commits

1338 + - rerere: recalculate conflict ID when unresolved conflict is committed

1339 + - rerere: teach rerere to handle nested conflicts

1340 + - rerere: return strbuf from handle path

1341 + - rerere: factor out handle_conflict function

1342 + - rerere: only return whether a path has conflicts or not

1343 + - rerere: fix crash when conflict goes unresolved

1344 + - rerere: add documentation for conflict normalization

1345 + - rerere: mark strings for translation

1346 + - rerere: wrap paths in output in sq

1347 + - rerere: lowercase error messages

1348 + - rerere: unify error messages when read_cache fails

1349

1138 - Will merge to 'master'.

1350 + Fixes to "git rerere" corner cases, especially when conflict

1351 + markers cannot be parsed in the file.

1352

1140 - I am not sure if there is a single "correct" answer everybody can

1141 - agree on for each of these "insane" cases, though.

1353 + I am not sure about the "nested" stuff, though.

1354

1355

1144 -* en/t7405-recursive-submodule-conflicts (2018-07-11) 3 commits

1145 - (merged to 'next' on 2018-07-24 at 6cb7d02298)

1146 - + t7405: verify 'merge --abort' works after submodule/path conflicts

1147 - + t7405: add a directory/submodule conflict

1148 - + t7405: add a file/submodule conflict

1356 +* jk/ui-color-always-to-auto (2018-07-18) 1 commit

1357 + (merged to 'next' on 2018-08-02 at 1a054baf0e)

1358 + + Documentation: fix --color option formatting

1359

1150 - Tests to cover conflict cases that involve submodules have been

1151 - added for merge-recursive.

1360 + Doc formatting fix.

1361

1362 Will merge to 'master'.

1363

1364

1156 -* es/test-fixes (2018-07-17) 26 commits

1157 - (merged to 'next' on 2018-07-24 at fd6796a3ef)

1158 - + t5608: fix broken &&-chain

1159 - + t9119: fix broken &&-chains

1160 - + t9000-t9999: fix broken &&-chains

1161 - + t7000-t7999: fix broken &&-chains

1162 - + t6000-t6999: fix broken &&-chains

1163 - + t5000-t5999: fix broken &&-chains

1164 - + t4000-t4999: fix broken &&-chains

1165 - + t3030: fix broken &&-chains

1166 - + t3000-t3999: fix broken &&-chains

1167 - + t2000-t2999: fix broken &&-chains

1168 - + t1000-t1999: fix broken &&-chains

1169 - + t0000-t0999: fix broken &&-chains

1170 - + t9814: simplify convoluted check that command correctly errors out

1171 - + t9001: fix broken "invoke hook" test

1172 - + t7810: use test_expect_code() instead of hand-rolled comparison

1173 - + t7400: fix broken "submodule add/reconfigure --force" test

1174 - + t7201: drop pointless "exit 0" at end of subshell

1175 - + t6036: fix broken "merge fails but has appropriate contents" tests

1176 - + t5505: modernize and simplify hard-to-digest test

1177 - + t5406: use write_script() instead of birthing shell script manually

1178 - + t5405: use test_must_fail() instead of checking exit code manually

1179 - + t/lib-submodule-update: fix "absorbing" test

1180 - + t: drop unnecessary terminating semicolon in subshell

1181 - + t: use sane_unset() rather than 'unset' with broken &&-chain

1182 - + t: use test_write_lines() instead of series of 'echo' commands

1183 - + t: use test_might_fail() instead of manipulating exit code manually

1184 - (this branch is used by es/chain-lint-in-subshell.)

1365 +* jh/json-writer (2018-07-16) 1 commit

1366 + (merged to 'next' on 2018-08-02 at d841450c7d)

1367 + + json_writer: new routines to create JSON data

1368 + (this branch is used by jh/structured-logging.)

1369

1186 - Test clean-up and corrections.

1370 + Preparatory code to later add json output for telemetry data.

1371

1372 Will merge to 'master'.

1373

1374

1191 -* jk/fsck-gitmodules-gently (2018-07-16) 6 commits

1192 - (merged to 'next' on 2018-07-24 at 5b15c800db)

1193 - + fsck: downgrade gitmodulesParse default to "info"

1194 - + fsck: split ".gitmodules too large" error from parse failure

1195 - + fsck: silence stderr when parsing .gitmodules

1196 - + config: add options parameter to git_config_from_mem

1197 - + config: add CONFIG_ERROR_SILENT handler

1198 - + config: turn die_on_error into caller-facing enum

1375 +* ag/rebase-i-in-c (2018-07-31) 20 commits

1376 + - rebase -i: move rebase--helper modes to rebase--interactive

1377 + - rebase -i: remove git-rebase--interactive.sh

1378 + - rebase--interactive2: rewrite the submodes of interactive rebase in C

1379 + - rebase -i: implement the main part of interactive rebase as a builtin

1380 + - rebase -i: rewrite init_basic_state() in C

1381 + - rebase -i: rewrite write_basic_state() in C

1382 + - rebase -i: rewrite the rest of init_revisions_and_shortrevisions() in C

1383 + - rebase -i: implement the logic to initialize $revisions in C

1384 + - rebase -i: remove unused modes and functions

1385 + - rebase -i: rewrite complete_action() in C

1386 + - t3404: todo list with commented-out commands only aborts

1387 + - sequencer: change the way skip_unnecessary_picks() returns its result

1388 + - sequencer: refactor append_todo_help() to write its message to a buffer

1389 + - rebase -i: rewrite checkout_onto() in C

1390 + - rebase -i: rewrite setup_reflog_action() in C

1391 + - sequencer: add a new function to silence a command, except if it fails

1392 + - rebase -i: rewrite the edit-todo functionality in C

1393 + - editor: add a function to launch the sequence editor

1394 + - rebase -i: rewrite append_todo_help() in C

1395 + - sequencer: make two functions and an enum from sequencer.c public

1396

1200 - Recent "security fix" to pay attention to contents of ".gitmodules"

1201 - while accepting "git push" was a bit overly strict than necessary,

1202 - which has been adjusted.

1397 + Rewrite of the remaining "rebase -i" machinery in C.

1398

1204 - Will merge to 'master'.

1399 + Will merge to 'next'.

1400

1401

1207 -* js/range-diff (2018-07-25) 21 commits

1402 +* js/range-diff (2018-07-30) 21 commits

1403 - range-diff: use dim/bold cues to improve dual color mode

1404 - range-diff: make --dual-color the default mode

1405 - range-diff: left-pad patch numbers

@@ -1226,55 +1421,14 @@ of the repositories listed at

1421 - range-diff: first rudimentary implementation

1422 - Introduce `range-diff` to compare iterations of a topic branch

1423 - linear-assignment: a function to solve least-cost assignment problems

1229 - (this branch is used by es/format-patch-rangediff.)

1424 + (this branch is used by es/format-patch-rangediff and sb/range-diff-colors.)

1425

1426 "git tbdiff" that lets us compare individual patches in two

1427 iterations of a topic has been rewritten and made into a built-in

1428 command.

1429

1235 - Undecided.

1236 -

1237 - Many "The feature is useful" comments without much real review; we

1238 - know the feature is great as this mimicks tbdiff already so that is

1239 - not news.

1240 -

1241 - I've squashed an obvious documentation fix in before rebasing the

1242 - other topic that depends on it. I _think_ we would probably be

1243 - better off to _disable_ whitespace-error coloring altogether when

1244 - showing diff-of-diff, and get rid of the workaround that only is

1245 - applicable to the context lines of the outer diff (unless we first

1246 - define how whitespace errors in diff-of-diff should be colored and

1247 - implement it correctly, that is, but after seeing these two

1248 - attempts, it seems that is harder than it is worth).

1249 -

1250 -

1251 -* jt/commit-graph-per-object-store (2018-07-17) 7 commits

1252 - (merged to 'next' on 2018-07-24 at 090d1a4d59)

1253 - + commit-graph: add repo arg to graph readers

1254 - + commit-graph: store graph in struct object_store

1255 - + commit-graph: add free_commit_graph

1256 - + commit-graph: add missing forward declaration

1257 - + object-store: add missing include

1258 - + commit-graph: refactor preparing commit graph

1259 - + Merge branch 'ds/commit-graph-fsck' into jt/commit-graph-per-object-store

1260 - (this branch is used by ds/commit-graph-with-grafts and ds/reachable; uses ds/commit-graph-fsck and sb/object-store-lookup.)

1261 -

1262 - The singleton commit-graph in-core instance is made per in-core

1263 - repository instance.

1264 -

1265 - Will merge to 'master'.

1266 -

1267 -

1268 -* kg/gc-auto-windows-workaround (2018-07-09) 1 commit

1269 - (merged to 'next' on 2018-07-24 at 71c05d27b6)

1270 - + gc --auto: release pack files before auto packing

1271 -

1272 - "git gc --auto" opens file descriptors for the packfiles before

1273 - spawning "git repack/prune", which would upset Windows that does

1274 - not want a process to work on a file that is open by another

1275 - process. The issue has been worked around.

1276 -

1277 - Will merge to 'master'.

1430 + It seems there will another hopefully the final reroll coming.

1431 + cf. <nycvar.QRO.7.76.6.1808011800570.71@tvgsbejvaqbjf.bet>

1432

1433

1434 * lt/date-human (2018-07-09) 1 commit

@@ -1300,42 +1454,17 @@ of the repositories listed at

1454 header. These cases have been optimzied by calling

1455 oid_object_info() API.

1456

1303 - What's the doneness of this one?

1457 + Will merge to 'next'.

1458

1459

1306 -* pk/rebase-in-c (2018-07-23) 4 commits

1460 +* pk/rebase-in-c (2018-07-30) 3 commits

1461 - builtin/rebase: support running "git rebase <upstream>"

1308 - - sequencer: refactor the code to detach HEAD to checkout.c

1462 - rebase: refactor common shell functions into their own file

1463 - rebase: start implementing it as a builtin

1464

1312 - Piecemeal rewrite of the "rebase" machinery in C.

1313 - Expecting a reroll, but it seems that this is getting quite close.

1314 - cf. <CAOZc8M8YmLwJOzG-1jyz8ft4W_tJMwNs6kSV8inX1q_zmDW8Sg@mail.gmail.com>

1315 -

1316 -

1317 -* jt/fetch-nego-tip (2018-07-03) 1 commit

1318 - (merged to 'next' on 2018-07-24 at a9e299006d)

1319 - + fetch-pack: support negotiation tip whitelist

1320 - (this branch uses jt/fetch-pack-negotiator; is tangled with jt/fetch-negotiator-skipping.)

1321 -

1322 - "git fetch" learned a new option "--negotiation-tip" to limit the

1323 - set of commits it tells the other end as "have", to reduce wasted

1324 - bandwidth and cycles, which would be helpful when the receiving

1325 - repository has a lot of refs that have little to do with the

1326 - history at the remote it is fetching from.

1327 -

1328 - Will merge to 'master'.

1329 -

1330 -

1331 -* tb/grep-only-matching (2018-07-09) 2 commits

1332 - (merged to 'next' on 2018-07-24 at 7e878b9d95)

1333 - + grep.c: teach 'git grep --only-matching'

1334 - + grep.c: extract show_line_header()

1335 -

1336 - "git grep" learned the "--only-matching" option.

1465 + Rewrite of the "rebase" machinery in C.

1466

1338 - Will merge to 'master'.

1467 + Will merge to 'next'.

1468

1469

1470 * jk/branch-l-1-repurpose (2018-06-22) 1 commit

@@ -1346,7 +1475,7 @@ of the repositories listed at

1475 Will hold in 'pu' until jk/branch-l-0-deprecation progresses sufficiently.

1476

1477

1349 -* cc/remote-odb (2018-07-16) 9 commits

1478 +* cc/remote-odb (2018-08-02) 9 commits

1479 - Documentation/config: add odb.<name>.promisorRemote

1480 - t0410: test fetching from many promisor remotes

1481 - Use odb.origin.partialclonefilter instead of core.partialclonefilter

@@ -1357,6 +1486,17 @@ of the repositories listed at

1486 - Add initial remote odb support

1487 - fetch-object: make functions return an error code

1488

1489 + Implement lazy fetches of missing objects to complement the

1490 + experimental partial clone feature.

1491 +

1492 + I haven't seen much interest in this topic on list. What's the

1493 + doneness of this thing?

1494 +

1495 + I do not particularly mind adding code to support a niche feature

1496 + as long as it is cleanly made and it is clear that the feature

1497 + won't negatively affect those who do not use it, so a review from

1498 + that point of view may also be appropriate.

1499 +

1500

1501 * ds/multi-pack-index (2018-07-20) 23 commits

1502 - midx: clear midx on repack

@@ -1388,136 +1528,25 @@ of the repositories listed at

1528 consulting many pack .idx files; a new mechanism to have a single

1529 file that consolidates all of these .idx files is introduced.

1530

1391 - Ready to move to 'next', with some known issues to be followed up?

1392 - cf. <xmqqefg8uplg.fsf@gitster-ct.c.googlers.com>

1393 - cf. <CAPig+cTU--KrGcv4C_CwBZEuec4dgm_tJqL=CFWKT6vxxR016w@mail.gmail.com>

1394 -

1395 -

1396 -* jt/fetch-pack-negotiator (2018-06-15) 7 commits

1397 - (merged to 'next' on 2018-07-24 at 438efcd6b1)

1398 - + fetch-pack: introduce negotiator API

1399 - + fetch-pack: move common check and marking together

1400 - + fetch-pack: make negotiation-related vars local

1401 - + fetch-pack: use ref adv. to prune "have" sent

1402 - + fetch-pack: directly end negotiation if ACK ready

1403 - + fetch-pack: clear marks before re-marking

1404 - + fetch-pack: split up everything_local()

1405 - (this branch is used by jt/fetch-nego-tip and jt/fetch-negotiator-skipping.)

1406 -

1407 - Code restructuring and a small fix to transport protocol v2 during

1408 - fetching.

1409 -

1410 - Will merge to 'master'.

1411 -

1412 -

1413 -* is/parsing-line-range (2018-06-15) 2 commits

1414 - (merged to 'next' on 2018-07-24 at a06b453f32)

1415 - + log: prevent error if line range ends past end of file

1416 - + blame: prevent error if range ends past end of file

1417 -

1418 - Parsing of -L[<N>][,[<M>]] parameters "git blame" and "git log"

1419 - take has been tweaked.

1420 -

1421 - Will merge to 'master'.

1422 -

1423 -

1424 -* ab/checkout-default-remote (2018-06-11) 8 commits

1425 - (merged to 'next' on 2018-07-24 at 6ef645f485)

1426 - + checkout & worktree: introduce checkout.defaultRemote

1427 - + checkout: add advice for ambiguous "checkout <branch>"

1428 - + builtin/checkout.c: use "ret" variable for return

1429 - + checkout: pass the "num_matches" up to callers

1430 - + checkout.c: change "unique" member to "num_matches"

1431 - + checkout.c: introduce an *_INIT macro

1432 - + checkout.h: wrap the arguments to unique_tracking_name()

1433 - + checkout tests: index should be clean after dwim checkout

1434 -

1435 - "git checkout" and "git worktree add" learned to honor

1436 - checkout.defaultRemote when auto-vivifying a local branch out of a

1437 - remote tracking branch in a repository with multiple remotes that

1438 - have tracking branches that share the same names.

1439 -

1440 - Will merge to 'master'.

1441 -

1442 -

1443 -* ds/commit-graph-fsck (2018-07-16) 23 commits

1444 - (merged to 'next' on 2018-07-24 at 6a802adc7a)

1445 - + coccinelle: update commit.cocci

1446 - + commit-graph: update design document

1447 - + gc: automatically write commit-graph files

1448 - + commit-graph: add '--reachable' option

1449 - + commit-graph: use string-list API for input

1450 - + fsck: verify commit-graph

1451 - + commit-graph: verify contents match checksum

1452 - + commit-graph: test for corrupted octopus edge

1453 - + commit-graph: verify commit date

1454 - + commit-graph: verify generation number

1455 - + commit-graph: verify parent list

1456 - + commit-graph: verify root tree OIDs

1457 - + commit-graph: verify objects exist

1458 - + commit-graph: verify corrupt OID fanout and lookup

1459 - + commit-graph: verify required chunks are present

1460 - + commit-graph: verify catches corrupt signature

1461 - + commit-graph: add 'verify' subcommand

1462 - + commit-graph: load a root tree from specific graph

1463 - + commit: force commit to parse from object database

1464 - + commit-graph: parse commit from chosen graph

1465 - + commit-graph: fix GRAPH_MIN_SIZE

1466 - + commit-graph: UNLEAK before die()

1467 - + t5318-commit-graph.sh: use core.commitGraph

1468 - (this branch is used by ds/commit-graph-with-grafts, ds/reachable and jt/commit-graph-per-object-store.)

1469 -

1470 - "git fsck" learns to make sure the optional commit-graph file is in

1471 - a sane state.

1472 -

1473 - Will merge to 'master'.

1474 -

1475 -

1476 -* ma/wrapped-info (2018-05-28) 2 commits

1477 - - usage: prefix all lines in `vreportf()`, not just the first

1478 - - usage: extract `prefix_suffix_lines()` from `advise()`

1479 -

1480 - An attempt to help making multi-line messages fed to warning(),

1481 - error(), and friends more easily translatable.

1482 -

1483 - Will discard and wait for a cleaned-up rewrite.

1484 - cf. <20180529213957.GF7964@sigill.intra.peff.net>

1531 + Will merge to and cook in 'next'.

1532

1533

1487 -* jm/cache-entry-from-mem-pool (2018-07-03) 8 commits

1488 - (merged to 'next' on 2018-07-24 at 9be51a88dc)

1489 - + block alloc: add validations around cache_entry lifecyle

1490 - + block alloc: allocate cache entries from mem_pool

1491 - + mem-pool: fill out functionality

1492 - + mem-pool: add life cycle management functions

1493 - + mem-pool: only search head block for available space

1494 - + block alloc: add lifecycle APIs for cache_entry structs

1495 - + read-cache: teach make_cache_entry to take object_id

1496 - + read-cache: teach refresh_cache_entry to take istate

1534 +--------------------------------------------------

1535 +[Discarded]

1536

1498 - For a large tree, the index needs to hold many cache entries

1499 - allocated on heap. These cache entries are now allocated out of a

1500 - dedicated memory pool to amortize malloc(3) overhead.

1537 +* am/sequencer-author-script-fix (2018-07-18) 1 commit

1538 + . sequencer.c: terminate the last line of author-script properly

1539

1502 - Will merge to 'master'.

1540 + The author-script that records the author information created by

1541 + the sequencer machinery lacked the closing single quote on the last

1542 + entry.

1543

1504 - This makes each cache-entry larger by either 4 or 8 bytes, which is

1505 - a bit sad, though.

1544 + Superseded by another topic.

1545

1546

1508 -* sb/diff-color-move-more (2018-07-19) 10 commits

1509 - (merged to 'next' on 2018-07-24 at 89c893cab2)

1510 - + diff.c: offer config option to control ws handling in move detection

1511 - + diff.c: add white space mode to move detection that allows indent changes

1512 - + diff.c: factor advance_or_nullify out of mark_color_as_moved

1513 - + diff.c: decouple white space treatment from move detection algorithm

1514 - + diff.c: add a blocks mode for moved code detection

1515 - + diff.c: adjust hash function signature to match hashmap expectation

1516 - + diff.c: do not pass diff options as keydata to hashmap

1517 - + t4015: avoid git as a pipe input

1518 - + xdiff/xdiffi.c: remove unneeded function declarations

1519 - + xdiff/xdiff.h: remove unused flags

1547 +* jc/push-cas-opt-comment (2018-08-01) 1 commit

1548 + . push: comment on a funny unbalanced option help

1549

1521 - "git diff --color-moved" feature has further been tweaked.

1550 + Code clarification.

1551

1523 - Will merge to 'master'.

1552 + Superseded by another topic.

What's cooking (2018/08 #01)