git@vger.kernel.org mailing list mirror (one of many)
 help / Atom feed
* What's cooking in git.git (Apr 2017, #04; Wed, 19)
@ 2017-04-20  5:37 Junio C Hamano
  2017-04-20  9:59 ` Duy Nguyen
                   ` (2 more replies)
  0 siblings, 3 replies; 29+ messages in thread
From: Junio C Hamano @ 2017-04-20  5:37 UTC (permalink / raw)
  To: git

Here are the topics that have been cooking.  Commits prefixed with
'-' are only in 'pu' (proposed updates) while commits prefixed with
'+' are in 'next'.  The ones marked with '.' do not appear in any of
the integration branches, but I am still holding onto them.

An early preview for the upcoming 2.13 release has been tagged,
almost a full day earlier than I planned.

You can find the changes described here in the integration branches
of the repositories listed at

    http://git-blame.blogspot.com/p/git-public-repositories.html

--------------------------------------------------
[Graduated to "master"]

* ab/grep-plug-pathspec-leak (2017-04-16) 1 commit
  (merged to 'next' on 2017-04-18 at d8f00b5e44)
 + grep: plug a trivial memory leak

 Call clear_pathspec() to release resources immediately before the
 cmd_grep() function returns.


* ah/diff-files-ours-theirs-doc (2017-04-13) 1 commit
  (merged to 'next' on 2017-04-18 at a4f80f0c3f)
 + diff-files: document --ours etc.

 The diff options "--ours", "--theirs" exist for quite some time.
 But so far they were not documented. Now they are.


* bc/object-id (2017-03-31) 20 commits
  (merged to 'next' on 2017-04-16 at b16f1899dd)
 + Documentation: update and rename api-sha1-array.txt
 + Rename sha1_array to oid_array
 + Convert sha1_array_for_each_unique and for_each_abbrev to object_id
 + Convert sha1_array_lookup to take struct object_id
 + Convert remaining callers of sha1_array_lookup to object_id
 + Make sha1_array_append take a struct object_id *
 + sha1-array: convert internal storage for struct sha1_array to object_id
 + builtin/pull: convert to struct object_id
 + submodule: convert check_for_new_submodule_commits to object_id
 + sha1_name: convert disambiguate_hint_fn to take object_id
 + sha1_name: convert struct disambiguate_state to object_id
 + test-sha1-array: convert most code to struct object_id
 + parse-options-cb: convert sha1_array_append caller to struct object_id
 + fsck: convert init_skiplist to struct object_id
 + builtin/receive-pack: convert portions to struct object_id
 + builtin/pull: convert portions to struct object_id
 + builtin/diff: convert to struct object_id
 + Convert GIT_SHA1_RAWSZ used for allocation to GIT_MAX_RAWSZ
 + Convert GIT_SHA1_HEXSZ used for allocation to GIT_MAX_HEXSZ
 + Define new hash-size constants for allocating memory

 Conversion from unsigned char [40] to struct object_id continues.


* bw/attr-pathspec (2017-04-16) 1 commit
  (merged to 'next' on 2017-04-18 at 2644ca1269)
 + pathspec: fix segfault in clear_pathspec

 Hotfix for a topic already in 'master'.


* bw/push-options-recursively-to-submodules (2017-04-11) 5 commits
  (merged to 'next' on 2017-04-16 at d4d657724b)
 + push: propagate remote and refspec with --recurse-submodules
 + submodule--helper: add push-check subcommand
 + remote: expose parse_push_refspec function
 + push: propagate push-options with --recurse-submodules
 + push: unmark a local variable as static

 "git push --recurse-submodules --push-option=<string>" learned to
 propagate the push option recursively down to pushes in submodules.


* bw/submodule-is-active (2017-04-13) 1 commit
  (merged to 'next' on 2017-04-18 at 25b05ec29e)
 + submodule--helper: fix typo in is_active error message

 Error message fix.


* dt/gc-ignore-old-gc-logs (2017-04-16) 1 commit
  (merged to 'next' on 2017-04-18 at caadf4ff66)
 + t6500: wait for detached auto gc at the end of the test script

 Hotfix for a topic already in 'master'.


* jh/memihash-opt (2017-04-18) 4 commits
  (merged to 'next' on 2017-04-18 at 6555be6bab)
 + p0004: make perf test executable
  (merged to 'next' on 2017-04-15 at 6bfc58e19a)
 + t3008: skip lazy-init test on a single-core box
 + test-online-cpus: helper to return cpu count
  (merged to 'next' on 2017-04-11 at ec5a6f2818)
 + name-hash: fix buffer overrun

 Hotfix for a topic that is already in 'master'.


* jk/no-looking-at-dotgit-outside-repo (2017-04-16) 2 commits
  (merged to 'next' on 2017-04-18 at 68260787ad)
 + test-read-cache: setup git dir
 + has_sha1_file: don't bother if we are not in a repository

 Clean up fallouts from recent tightening of the set-up sequence,
 where Git barfs when repository information is accessed without
 first ensuring that it was started in a repository.


* ld/p4-current-branch-fix (2017-04-16) 3 commits
  (merged to 'next' on 2017-04-18 at 9e74a2609e)
 + git-p4: don't use name-rev to get current branch
 + git-p4: add read_pipe_text() internal function
 + git-p4: add failing test for name-rev rather than symbolic-ref

 "git p4" used "name-rev HEAD" when it wants to learn what branch is
 checked out; it should use "symbolic-ref HEAD".


* lt/mailinfo-in-body-header-continuation (2017-04-11) 1 commit
  (merged to 'next' on 2017-04-16 at b2f51f0780)
 + mailinfo: fix in-body header continuations

 If a patch e-mail had its first paragraph after an in-body header
 indented (even after a blank line after the in-body header line),
 the indented line was mistook as a continuation of the in-body
 header.  This has been fixed.


* nd/files-backend-git-dir (2017-04-14) 28 commits
  (merged to 'next' on 2017-04-18 at 79292aac69)
 + refs.h: add a note about sorting order of for_each_ref_*
 + t1406: new tests for submodule ref store
 + t1405: some basic tests on main ref store
 + t/helper: add test-ref-store to test ref-store functions
 + refs: delete pack_refs() in favor of refs_pack_refs()
 + files-backend: avoid ref api targeting main ref store
 + refs: new transaction related ref-store api
 + refs: add new ref-store api
 + refs: rename get_ref_store() to get_submodule_ref_store() and make it public
 + files-backend: replace submodule_allowed check in files_downcast()
 + refs: move submodule code out of files-backend.c
 + path.c: move some code out of strbuf_git_path_submodule()
 + refs.c: make get_main_ref_store() public and use it
 + refs.c: kill register_ref_store(), add register_submodule_ref_store()
 + refs.c: flatten get_ref_store() a bit
 + refs: rename lookup_ref_store() to lookup_submodule_ref_store()
 + refs.c: introduce get_main_ref_store()
 + files-backend: remove the use of git_path()
 + files-backend: add and use files_ref_path()
 + files-backend: add and use files_reflog_path()
 + files-backend: move "logs/" out of TMP_RENAMED_LOG
 + files-backend: convert git_path() to strbuf_git_path()
 + files-backend: make sure files_rename_ref() always reach the end
 + files-backend: add and use files_packed_refs_path()
 + files-backend: delete dead code in files_init_db()
 + files-backend.c: delete dead code in files_ref_iterator_begin()
 + files-backend: make files_log_ref_write() static
 + refs.h: add forward declaration for structs used in this file
 (this branch is used by mh/separate-ref-cache, nd/prune-in-worktree and nd/worktree-kill-parse-ref.)

 The "submodule" specific field in the ref_store structure is
 replaced with a more generic "gitdir" that can later be used also
 when dealing with ref_store that represents the set of refs visible
 from the other worktrees.


* sb/submodule-rm-absorb (2017-04-13) 1 commit
  (merged to 'next' on 2017-04-18 at 179081f848)
 + submodule.c: add missing ' in error messages

 Error message fix.


* sb/submodule-short-status (2017-03-29) 7 commits
  (merged to 'next' on 2017-04-16 at 24e18f4a4a)
 + submodule.c: correctly handle nested submodules in is_submodule_modified
 + short status: improve reporting for submodule changes
 + submodule.c: stricter checking for submodules in is_submodule_modified
 + submodule.c: port is_submodule_modified to use porcelain 2
 + submodule.c: convert is_submodule_modified to use strbuf_getwholeline
 + submodule.c: factor out early loop termination in is_submodule_modified
 + submodule.c: use argv_array in is_submodule_modified

 The output from "git status --short" has been extended to show
 various kinds of dirtyness in submodules differently; instead of to
 "M" for modified, 'm' and '?' can be shown to signal changes only
 to the working tree of the submodule but not the commit that is
 checked out.


* sf/putty-w-args (2017-04-16) 1 commit
  (merged to 'next' on 2017-04-18 at cbf042088c)
 + connect.c: handle errors from split_cmdline

 Hotfix for a topic already in 'master'.


* va/i18n-perl-scripts (2017-04-13) 1 commit
  (merged to 'next' on 2017-04-18 at 89f1c77091)
 + git-add--interactive.perl: add missing dot in a message

 Message fix.


* vn/revision-shorthand-for-side-branch-log (2017-04-16) 1 commit
  (merged to 'next' on 2017-04-18 at 9fbe55bfef)
 + doc/revisions: remove brackets from rev^-n shorthand

 Doc cleanup.

--------------------------------------------------
[New Topics]

* ab/push-cas-doc-n-test (2017-04-19) 1 commit
 - push: document & test --force-with-lease with multiple remotes

 Doc update.

 Will merge to 'next'.


* jk/parse-options-no-no-no (2017-04-19) 2 commits
 - SQUASH???
 - parse-options: disallow double-negations of options starting with no-

 Command line options that begin with "--no-" (e.g. "--no-checkout"
 option of "git clone") can be negated by removing "--no-"; we
 historically also allowed prefixing an extra "no" to the option
 (e.g. "--no-no-checkout"), which made the command line look ugly
 and unusual.  This proposes to forbid it.

 While I agree there is no need to support "--no-no-checkout", this
 looks more like "if it looks ugly and unusual, you do not have to
 use it".  Perhaps we can drop it?


* ss/gitmodules-ignore-doc (2017-04-19) 1 commit
 - gitmodules: clarify the ignore option values

 Doc update.

 Will merge to 'next'.


* ss/submodule-shallow-doc (2017-04-19) 1 commit
 - gitmodules: clarify what history depth a shallow clone has

 Doc update.

 Will merge to 'next'.

--------------------------------------------------
[Stalled]

* sg/clone-refspec-from-command-line-config (2017-02-27) 1 commit
 - clone: respect configured fetch respecs during initial fetch

 Expecting a reroll.
 cf. <20170227211217.73gydlxb2qu2sp3m@sigill.intra.peff.net>
 cf. <CAM0VKj=rsAfKvVccOMOoo5==Q1yW1U0zJBbUV=faKppWFm-u+g@mail.gmail.com>
 Some nits but looks ok.


* sk/dash-is-previous (2017-03-01) 5 commits
 - revert.c: delegate handling of "-" shorthand to setup_revisions
 - sha1_name.c: teach get_sha1_1 "-" shorthand for "@{-1}"
 - revision.c: args starting with "-" might be a revision
 - revision.c: swap if/else blocks
 - revision.c: do not update argv with unknown option

 A dash "-" can be written to mean "the branch that was previously
 checked out" in more places.

 Needs review.
 cf. <1488007487-12965-1-git-send-email-kannan.siddharth12@gmail.com>


* ls/filter-process-delayed (2017-03-06) 1 commit
 - convert: add "status=delayed" to filter process protocol

 What's the status of this one???
 cf. <xmqq60jmnmef.fsf@junio-linux.mtv.corp.google.com>


* nd/worktree-move (2017-01-27) 7 commits
 . fixup! worktree move: new command
 . worktree remove: new command
 . worktree move: refuse to move worktrees with submodules
 . worktree move: accept destination as directory
 . worktree move: new command
 . worktree.c: add update_worktree_location()
 . worktree.c: add validate_worktree()

 "git worktree" learned move and remove subcommands.

 Tentatively ejected as it seems to break 'pu' when merged.


* pb/bisect (2017-02-18) 28 commits
 - fixup! bisect--helper: `bisect_next_check` & bisect_voc shell function in C
 - bisect--helper: remove the dequote in bisect_start()
 - bisect--helper: retire `--bisect-auto-next` subcommand
 - bisect--helper: retire `--bisect-autostart` subcommand
 - bisect--helper: retire `--bisect-write` subcommand
 - bisect--helper: `bisect_replay` shell function in C
 - bisect--helper: `bisect_log` shell function in C
 - bisect--helper: retire `--write-terms` subcommand
 - bisect--helper: retire `--check-expected-revs` subcommand
 - bisect--helper: `bisect_state` & `bisect_head` shell function in C
 - bisect--helper: `bisect_autostart` shell function in C
 - bisect--helper: retire `--next-all` subcommand
 - bisect--helper: retire `--bisect-clean-state` subcommand
 - bisect--helper: `bisect_next` and `bisect_auto_next` shell function in C
 - t6030: no cleanup with bad merge base
 - bisect--helper: `bisect_start` shell function partially in C
 - bisect--helper: `get_terms` & `bisect_terms` shell function in C
 - bisect--helper: `bisect_next_check` & bisect_voc shell function in C
 - bisect--helper: `check_and_set_terms` shell function in C
 - bisect--helper: `bisect_write` shell function in C
 - bisect--helper: `is_expected_rev` & `check_expected_revs` shell function in C
 - bisect--helper: `bisect_reset` shell function in C
 - wrapper: move is_empty_file() and rename it as is_empty_or_missing_file()
 - t6030: explicitly test for bisection cleanup
 - bisect--helper: `bisect_clean_state` shell function in C
 - bisect--helper: `write_terms` shell function in C
 - bisect: rewrite `check_term_format` shell function in C
 - bisect--helper: use OPT_CMDMODE instead of OPT_BOOL

 Move more parts of "git bisect" to C.

 Expecting a reroll.
 cf. <CAFZEwPPXPPHi8KiEGS9ggzNHDCGhuqMgH9Z8-Pf9GLshg8+LPA@mail.gmail.com>
 cf. <CAFZEwPM9RSTGN54dzaw9gO9iZmsYjJ_d1SjUD4EzSDDbmh-XuA@mail.gmail.com>
 cf. <CAFZEwPNUXcNY9Qdz=_B7q2kQuaecPzJtTMGdv8YMUPEz2vnp8A@mail.gmail.com>


* sh/grep-tree-obj-tweak-output (2017-01-20) 2 commits
 - grep: use '/' delimiter for paths
 - grep: only add delimiter if there isn't one already

 "git grep", when fed a tree-ish as an input, shows each hit
 prefixed with "<tree-ish>:<path>:<lineno>:".  As <tree-ish> is
 almost always either a commit or a tag that points at a commit, the
 early part of the output "<tree-ish>:<path>" can be used as the
 name of the blob and given to "git show".  When <tree-ish> is a
 tree given in the extended SHA-1 syntax (e.g. "<commit>:", or
 "<commit>:<dir>"), however, this results in a string that does not
 name a blob (e.g. "<commit>::<path>" or "<commit>:<dir>:<path>").
 "git grep" has been taught to be a bit more intelligent about these
 cases and omit a colon (in the former case) or use slash (in the
 latter case) to produce "<commit>:<path>" and
 "<commit>:<dir>/<path>" that can be used as the name of a blob.

 Expecting a reroll?  Is this good enough with known limitations?


* jc/diff-b-m (2015-02-23) 5 commits
 . WIPWIP
 . WIP: diff-b-m
 - diffcore-rename: allow easier debugging
 - diffcore-rename.c: add locate_rename_src()
 - diffcore-break: allow debugging

 "git diff -B -M" produced incorrect patch when the postimage of a
 completely rewritten file is similar to the preimage of a removed
 file; such a resulting file must not be expressed as a rename from
 other place.

 The fix in this patch is broken, unfortunately.

 Will discard.

--------------------------------------------------
[Cooking]

* df/dir-iter-remove-subtree (2017-04-19) 5 commits
 - remove_subtree(): reimplement using iterators
 - dir_iterator: rewrite state machine model
 - dir_iterator: refactor dir_iterator_advance
 - remove_subtree(): test removing nested directories
 - dir_iterator: add tests for dir_iterator API

 Update the dir-iterator API and use it to reimplement
 remove_subtree().

 Almost there except for a minor nit.
 GSoC microproject.


* jk/ls-files-recurse-submodules-fix (2017-04-18) 2 commits
  (merged to 'next' on 2017-04-19 at d4682ceaae)
 + ls-files: fix path used when recursing into submodules
 + ls-files: fix recurse-submodules with nested submodules

 "ls-files --recurse-submodules" did not quite work well in a
 project with nested submodules.

 Will merge to 'master'.


* jt/fetch-pack-error-reporting (2017-04-17) 1 commit
  (merged to 'next' on 2017-04-19 at 6da61f7236)
 + fetch-pack: show clearer error message upon ERR

 "git fetch-pack" was not prepared to accept ERR packet that the
 upload-pack can send with a human-readable error message.  It
 showed the packet contents with ERR prefix, so there was no data
 loss, but it was redundant to say "ERR" in an error message.

 Will merge to 'master'.


* nd/conditional-config-in-early-config (2017-04-19) 3 commits
 - config: correct file reading order in read_early_config()
 - config: handle conditional include when $GIT_DIR is not set up
 - config: prepare to pass more info in git_config_with_options()

 The recently introduced conditional inclusion of configuration did
 not work well when early-config mechanism was involved.

 Will merge to 'next'.


* sb/checkout-recurse-submodules (2017-04-17) 1 commit
  (merged to 'next' on 2017-04-19 at 414e18fd9b)
 + submodule: remove a superfluous second check for the "new" variable

 Code cleanup.

 Will merge to 'master'.


* xy/format-patch-base (2017-04-17) 1 commit
  (merged to 'next' on 2017-04-19 at c10723d486)
 + doc: trivial typo in git-format-patch.txt

 Doc cleanup.

 Will merge to 'master'.


* jk/snprintf-cleanups (2017-04-17) 1 commit
  (merged to 'next' on 2017-04-19 at 4a18ea9971)
 + replace: plug a memory leak

 Hotfix for a topic that is already in 'master'.

 Will merge to 'master'.


* ab/clone-no-tags (2017-04-19) 1 commit
 - clone: add a --no-tags option to clone without tags

 "git clone" learned the "--no-tags" option not to fetch all tags
 initially, and also set up the tagopt not to follow any tags in
 subsequent fetches.

 Will merge to 'next'.


* ab/completion-push-delete-ref (2017-04-18) 1 commit
 - completion: expand "push --delete <remote> <ref>" for refs on that <remote>

 The completion script (in contrib/) learned to complete "git push
 --delete b<TAB>" to complete branch name to be deleted.

 Will merge to 'next'.


* bw/forking-and-threading (2017-04-19) 11 commits
 - run-command: block signals between fork and execve
 - run-command: add note about forking and threading
 - run-command: handle dup2 and close errors in child
 - run-command: eliminate calls to error handling functions in child
 - run-command: don't die in child when duping /dev/null
 - run-command: prepare child environment before forking
 - string-list: add string_list_remove function
 - run-command: use the async-signal-safe execv instead of execvp
 - run-command: prepare command before forking
 - t0061: run_command executes scripts without a #! line
 - t5550: use write_script to generate post-update hook

 The "run-command" APIimplementation has been made more robust
 against dead-locking in a threaded environment.

 Will merge to 'next'.


* sb/reset-recurse-submodules (2017-04-18) 4 commits
 - builtin/reset: add --recurse-submodules switch
 - submodule.c: submodule_move_head works with broken submodules
 - submodule.c: uninitialized submodules are ignored in recursive commands
 - entry.c: submodule recursing: respect force flag correctly

 "git reset" learned "--recurse-submodules" option.


* dt/xgethostname-nul-termination (2017-04-18) 2 commits
  (merged to 'next' on 2017-04-19 at 4035f442cc)
 + xgethostname: handle long hostnames
 + use HOST_NAME_MAX to size buffers for gethostname(2)

 gethostname(2) may not NUL terminate the buffer if hostname does
 not fit; unfortunately there is no easy way to see if our buffer
 was too small, but at least this will make sure we will not end up
 using garbage past the end of the buffer.

 Will merge to 'master'.


* jh/string-list-micro-optim (2017-04-15) 1 commit
  (merged to 'next' on 2017-04-19 at 2a5e339df9)
 + string-list: use ALLOC_GROW macro when reallocing string_list

 The string-list API used a custom reallocation strategy that was
 very inefficient, instead of using the usual ALLOC_GROW() macro,
 which has been fixed.

 Will merge to 'master'.


* jh/unpack-trees-micro-optim (2017-04-15) 1 commit
  (merged to 'next' on 2017-04-19 at 970c47a2f9)
 + unpack-trees: avoid duplicate ODB lookups during checkout

 In a 2- and 3-way merge of trees, more than one source trees often
 end up sharing an identical subtree; optimize by not reading the
 same tree multiple times in such a case.

 Will merge to 'master'.


* jh/verify-index-checksum-only-in-fsck (2017-04-15) 1 commit
  (merged to 'next' on 2017-04-19 at a089b97d34)
 + read-cache: force_verify_index_checksum

 The index file has a trailing SHA-1 checksum to detect file
 corruption, and historically we checked it every time the index
 file is used.  Omit the validation during normal use, and instead
 verify only in "git fsck".

 Will merge to 'master'.


* ls/travis-doc-asciidoctor (2017-04-16) 3 commits
  (merged to 'next' on 2017-04-19 at 359c32953b)
 + travis-ci: unset compiler for jobs that do not need one
 + travis-ci: parallelize documentation build
 + travis-ci: build documentation with AsciiDoc and Asciidoctor

 Have Travis CI format the documentation with both AsciiDoc and
 AsciiDoctor.

 Will merge to 'master'.


* mg/status-in-progress-info (2017-04-14) 1 commit
 - status: show in-progress info for short status

 "git status" learns an option to report various operations
 (e.g. "merging") that the user is in the middle of.

 Waiting for final confirmation.
 cf. <xmqqwpajikd2.fsf@gitster.mtv.corp.google.com>


* nd/conditional-config-include (2017-04-14) 2 commits
  (merged to 'next' on 2017-04-19 at 2e94c40b46)
 + config: resolve symlinks in conditional include's patterns
 + path.c: and an option to call real_path() in expand_user_path()

 $GIT_DIR may in some cases be normalized with all symlinks resolved
 while "gitdir" path expansion in the pattern does not receive the
 same treatment, leading to incorrect mismatch.  This has been fixed.

 Will merge to 'master'.


* nd/worktree-add-lock (2017-04-16) 2 commits
 - SQUASH???
 - worktree add: add --lock option

 Allow to lock a worktree immediately after it's created. This helps
 prevent a race between "git worktree add; git worktree lock" and
 "git worktree prune".

 Waiting for a response to SQUASH???


* ps/pathspec-empty-prefix-origin (2017-04-14) 1 commit
 - pathspec: honor `PATHSPEC_PREFIX_ORIGIN` with empty prefix

 A recent update broke "git add -p ../foo" from a subdirectory.

 Will merge to 'next'.


* sr/http-proxy-configuration-fix (2017-04-13) 2 commits
  (merged to 'next' on 2017-04-19 at d8f507e951)
 + http: fix the silent ignoring of proxy misconfiguraion
 + http: honor empty http.proxy option to bypass proxy

 "http.proxy" set to an empty string is used to disable the usage of
 proxy.  We broke this early last year.

 Will merge to 'master'.


* tb/doc-eol-normalization (2017-04-13) 1 commit
  (merged to 'next' on 2017-04-19 at fe15d04d0a)
 + gitattributes.txt: document how to normalize the line endings

 Doc update.

 Will merge to 'master'.


* gb/rebase-signoff (2017-04-18) 3 commits
  (merged to 'next' on 2017-04-19 at c3c04e6a71)
 + rebase: pass --[no-]signoff option to git am
 + builtin/am: fold am_signoff() into am_append_signoff()
 + builtin/am: honor --signoff also when --rebasing

 "git rebase" learns "--signoff" option.

 Will merge to 'master'.


* jh/add-index-entry-optim (2017-04-19) 5 commits
 - read-cache: speed up has_dir_name (part 2)
 - read-cache: speed up has_dir_name (part 1)
 - read-cache: speed up add_index_entry during checkout
 - p0006-read-tree-checkout: perf test to time read-tree
 - read-cache: add strcmp_offset function

 "git checkout" that handles a lot of paths has been optimized by
 reducing the number of unnecessary checks of paths in the
 has_dir_name() function.

 Will merge to 'next'.


* ab/grep-threading-cleanup (2017-04-16) 8 commits
 - grep: given --threads with NO_PTHREADS=YesPlease, warn
 - pack-objects: fix buggy warning about threads under NO_PTHREADS=YesPlease
 - pack-object & index-pack: add test for --threads warning under NO_PTHREADS
 - tests: add a PTHREADS prerequisite
 - grep: skip pthreads overhead when using one thread
 - grep: don't redundantly compile throwaway patterns under threading
 - grep: add tests for --threads=N and grep.threads
 - grep: assert that threading is enabled when calling grep_{lock,unlock}


* bw/submodule-with-bs-path (2017-04-16) 1 commit
  (merged to 'next' on 2017-04-19 at 692775477b)
 + submodule: prevent backslash expantion in submodule names

 "git submodule" script does not work well with strange pathnames.
 Protect it from a path with slashes in them, at least.

 Will merge to 'master'.


* jk/loose-object-fsck (2017-04-16) 1 commit
  (merged to 'next' on 2017-04-19 at 0cc7810018)
 + sha1_file: remove an used fd variable

 Code cleanup.

 Will merge to 'master'.


* jk/quarantine-received-objects (2017-04-16) 3 commits
  (merged to 'next' on 2017-04-19 at 2c6fcf1d6d)
 + refs: reject ref updates while GIT_QUARANTINE_PATH is set
 + receive-pack: document user-visible quarantine effects
 + receive-pack: drop tmp_objdir_env from run_update_hook

 Add finishing touches to a recent topic.

 Will merge to 'master'.


* ls/travis-coccicheck (2017-04-16) 1 commit
 - travis-ci: add static analysis build job to run coccicheck

 Travis CI learns to run coccicheck.

 Will merge to 'next'.


* rs/misc-cppcheck-fixes (2017-04-17) 3 commits
  (merged to 'next' on 2017-04-19 at e8fca7f593)
 + server-info: avoid calling fclose(3) twice in update_info_file()
 + files_for_each_reflog_ent_reverse(): close stream and free strbuf on error
 + am: close stream on error, but not stdin

 Various small fixes.

 Will merge to 'master'.


* km/t1400-modernization (2017-04-16) 1 commit
  (merged to 'next' on 2017-04-19 at 7490be39bf)
 + t1400: use consistent style for test_expect_success calls

 Code cleanup.

 Will merge to 'master'.


* dt/http-postbuffer-can-be-large (2017-04-13) 1 commit
  (merged to 'next' on 2017-04-19 at dc9de26fc9)
 + http.postbuffer: allow full range of ssize_t values

 Allow the http.postbuffer configuration variable to be set to a
 size that can be expressed in size_t, which can be larger than
 ulong on some platforms.

 Will merge to 'master'.


* pc/t2027-git-to-pipe-cleanup (2017-04-14) 1 commit
 - t2027: avoid using pipes

 Having a git command on the upstream side of a pipe in a test
 script will hide the exit status from the command, which may cause
 us to fail to notice a breakage; rewrite tests in a script to avoid
 this issue.

 Will merge to 'next'.


* bp/sub-process-convert-filter (2017-03-30) 8 commits
 - convert: update subprocess_read_status to not die on EOF
 - sub-process: move sub-process functions into separate files
 - convert: rename reusable sub-process functions
 - convert: update generic functions to only use generic data structures
 - convert: separate generic structures and variables from the filter specific ones
 - convert: split start_multi_file_filter() into two separate functions
 - convert: move packet_write_list() into pkt-line.c as packet_writel()
 - pkt-line: add packet_read_line_gently()

 Code from "conversion using external process" codepath has been
 extracted to a separate sub-process.[ch] module.

 Reroll exists but needs more work.
 cf. <20170407120354.17736-1-benpeart@microsoft.com>


* cc/split-index-config (2017-03-30) 1 commit
 - read-cache: avoid using git_path() in freshen_shared_index()

 The split-index code configuration code used an unsafe git_path()
 function without copying its result out.

 Needs to be explained better.
 The code looked OK, though.


* mg/name-rev-debug (2017-03-31) 4 commits
 - describe: pass --debug down to name-rev
 - name-rev: provide debug output
 - name-rev: favor describing with tags and use committer date to tiebreak
 - name-rev: refactor logic to see if a new candidate is a better name

 "git describe --debug --contains" did not add any meaningful
 information, even though without "--contains" it did.

 Expecting a reroll of the tip two.
 cf. <xmqqshltxnwt.fsf@gitster.mtv.corp.google.com>


* mh/separate-ref-cache (2017-04-16) 20 commits
 - do_for_each_entry_in_dir(): delete function
 - files_pack_refs(): use reference iteration
 - commit_packed_refs(): use reference iteration
 - cache_ref_iterator_begin(): make function smarter
 - get_loose_ref_cache(): new function
 - get_loose_ref_dir(): function renamed from get_loose_refs()
 - do_for_each_entry_in_dir(): eliminate `offset` argument
 - refs: handle "refs/bisect/" in `loose_fill_ref_dir()`
 - ref-cache: use a callback function to fill the cache
 - refs: record the ref_store in ref_cache, not ref_dir
 - ref-cache: introduce a new type, ref_cache
 - refs: split `ref_cache` code into separate files
 - ref-cache: rename `remove_entry()` to `remove_entry_from_dir()`
 - ref-cache: rename `find_ref()` to `find_ref_entry()`
 - ref-cache: rename `add_ref()` to `add_ref_entry()`
 - refs_verify_refname_available(): use function in more places
 - refs_verify_refname_available(): implement once for all backends
 - refs_ref_iterator_begin(): new function
 - refs_read_raw_ref(): new function
 - get_ref_dir(): don't call read_loose_refs() for "refs/bisect"

 The internals of the refs API around the cached refs has been
 streamlined.

 Will merge to 'next'.


* ja/doc-l10n (2017-03-20) 3 commits
 - SQUASH???
 - l10n: add git-add.txt to localized man pages
 - l10n: introduce framework for localizing man pages

 A proposal to use po4a to localize our manual pages.


* nd/prune-in-worktree (2017-04-19) 12 commits
 - rev-list: expose and document --single-worktree
 - revision.c: --reflog add HEAD reflog from all worktrees
 - files-backend: make reflog iterator go through per-worktree reflog
 - revision.c: --all adds HEAD from all worktrees
 - refs: remove dead for_each_*_submodule()
 - revision.c: use refs_for_each*() instead of for_each_*_submodule()
 - refs: add refs_head_ref()
 - refs: move submodule slash stripping code to get_submodule_ref_store
 - refs.c: refactor get_submodule_ref_store(), share common free block
 - revision.c: --indexed-objects add objects from all worktrees
 - revision.c: refactor add_index_objects_to_pending()
 - revision.h: new flag in struct rev_info wrt. worktree-related refs
 (this branch uses nd/worktree-kill-parse-ref.)

 "git gc" and friends when multiple worktrees are used off of a
 single repository did not consider the index and per-worktree refs
 of other worktrees as the root for reachability traversal, making
 objects that are in use only in other worktrees to be subject to
 garbage collection.

 Waiting for nd/worktree-kill-parse-ref to settle.


* nd/worktree-kill-parse-ref (2017-04-16) 5 commits
 - refs: kill set_worktree_head_symref()
 - worktree.c: kill parse_ref() in favor of refs_resolve_ref_unsafe()
 - refs: introduce get_worktree_ref_store()
 - refs.c: make submodule ref store hashmap generic
 - environment.c: fix potential segfault by get_git_common_dir()
 (this branch is used by nd/prune-in-worktree.)

 "git gc" did not interact well with "git worktree"-managed
 per-worktree refs.

 Will merge to 'next'.


* jc/bundle (2016-03-03) 6 commits
 - index-pack: --clone-bundle option
 - Merge branch 'jc/index-pack' into jc/bundle
 - bundle v3: the beginning
 - bundle: keep a copy of bundle file name in the in-core bundle header
 - bundle: plug resource leak
 - bundle doc: 'verify' is not about verifying the bundle

 The beginning of "split bundle", which could be one of the
 ingredients to allow "git clone" traffic off of the core server
 network to CDN.

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20  5:37 What's cooking in git.git (Apr 2017, #04; Wed, 19) Junio C Hamano
@ 2017-04-20  9:59 ` Duy Nguyen
  2017-04-20 15:35   ` Jeff King
  2017-04-20 22:46   ` Junio C Hamano
       [not found] ` <D61D47BD-9750-4FB6-892E-013504E03738@gmail.com>
  2017-04-20 15:32 ` Lars Schneider
  2 siblings, 2 replies; 29+ messages in thread
From: Duy Nguyen @ 2017-04-20  9:59 UTC (permalink / raw)
  To: Junio C Hamano; +Cc: git

On Wed, Apr 19, 2017 at 10:37:21PM -0700, Junio C Hamano wrote:
> * nd/worktree-add-lock (2017-04-16) 2 commits
>  - SQUASH???
>  - worktree add: add --lock option
> 
>  Allow to lock a worktree immediately after it's created. This helps
>  prevent a race between "git worktree add; git worktree lock" and
>  "git worktree prune".
> 
>  Waiting for a response to SQUASH???

Looking good. I would add some comment, lest ';' feel lonely. But it's
really personal taste.

-- 8< --
diff --git a/builtin/worktree.c b/builtin/worktree.c
index 5ebdcce793..bc75676bf3 100644
--- a/builtin/worktree.c
+++ b/builtin/worktree.c
@@ -310,7 +310,7 @@ static int add_worktree(const char *path, const char *refname,
 	strbuf_reset(&sb);
 	strbuf_addf(&sb, "%s/locked", sb_repo.buf);
 	if (!ret && opts->keep_locked)
-		;
+		;	      /* --lock wants to keep "locked" file */
 	else
 		unlink_or_warn(sb.buf);
 	argv_array_clear(&child_env);
-- 8< --

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
       [not found] ` <D61D47BD-9750-4FB6-892E-013504E03738@gmail.com>
@ 2017-04-20 13:24   ` Johannes Schindelin
  2017-04-20 16:56     ` Brandon Williams
  2017-04-20 22:58   ` Junio C Hamano
  1 sibling, 1 reply; 29+ messages in thread
From: Johannes Schindelin @ 2017-04-20 13:24 UTC (permalink / raw)
  To: Lars Schneider; +Cc: Junio C Hamano, git, peff, bmwill

Hi Lars & Junio,

On Thu, 20 Apr 2017, Lars Schneider wrote:

> > * bw/forking-and-threading (2017-04-19) 11 commits
> > - run-command: block signals between fork and execve
> > - run-command: add note about forking and threading
> > - run-command: handle dup2 and close errors in child
> > - run-command: eliminate calls to error handling functions in child
> > - run-command: don't die in child when duping /dev/null
> > - run-command: prepare child environment before forking
> > - string-list: add string_list_remove function
> > - run-command: use the async-signal-safe execv instead of execvp
> > - run-command: prepare command before forking
> > - t0061: run_command executes scripts without a #! line
> > - t5550: use write_script to generate post-update hook
> > 
> > The "run-command" APIimplementation has been made more robust
> > against dead-locking in a threaded environment.
> > 
> > Will merge to 'next'.
> 
> There might be a problem on Windows with this (that's just a hunch, i can't test this right now):
> https://travis-ci.org/git/git/jobs/223830474

Thanks for keeping track of Travis' failure reports. From what I see, the
latest iteration (which does not seem to have made it to `pu` yet) has the
!MINGW prerequisite which should fix the issue. Hopefully my suggested
addition to the commit message will make it into the commit history, too.

Ciao,
Dscho

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20  5:37 What's cooking in git.git (Apr 2017, #04; Wed, 19) Junio C Hamano
  2017-04-20  9:59 ` Duy Nguyen
       [not found] ` <D61D47BD-9750-4FB6-892E-013504E03738@gmail.com>
@ 2017-04-20 15:32 ` Lars Schneider
  2017-04-20 22:52   ` Junio C Hamano
  2 siblings, 1 reply; 29+ messages in thread
From: Lars Schneider @ 2017-04-20 15:32 UTC (permalink / raw)
  To: git

Sorry for sending this email multiple times. My mobile email client created html... Should be fixed now!

> 
> * ls/filter-process-delayed (2017-03-06) 1 commit
> - convert: add "status=delayed" to filter process protocol
> 
> What's the status of this one???
> cf. <xmqq60jmnmef.fsf@junio-linux.mtv.corp.google.com>

posted v3 here:
https://public-inbox.org/git/20170409191107.20547-1-larsxschneider@gmail.com/ 

From my point of view the only open question is if we use an index ("delay-id" favored by Peff) or the path (favored by Taylor) to identify delayed items: 
https://public-inbox.org/git/20170227095825.jhdspwy6oa6mvvvm@sigill.intra.peff.net/ 
https://public-inbox.org/git/20170412173404.GA49694@Ida/#r 

I slightly favor the path approach as it simplifies the protocol. But I also trust Peff's opinion.


> * ls/travis-doc-asciidoctor (2017-04-16) 3 commits
> (merged to 'next' on 2017-04-19 at 359c32953b)
> + travis-ci: unset compiler for jobs that do not need one
> + travis-ci: parallelize documentation build
> + travis-ci: build documentation with AsciiDoc and Asciidoctor
> 
> Have Travis CI format the documentation with both AsciiDoc and
> AsciiDoctor.
> 
> Will merge to 'master'.

Can you hold this until next week?
The I would post a v2 and check asciidoc stderr: 
https://public-inbox.org/git/D9F45212-91F7-4BB1-A0EC-74A84DA81860@gmail.com/ 

I can also post an additional patch if you prefer to merge now.

Thanks, Lars



^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20  9:59 ` Duy Nguyen
@ 2017-04-20 15:35   ` Jeff King
  2017-04-20 22:51     ` Junio C Hamano
  2017-04-20 22:46   ` Junio C Hamano
  1 sibling, 1 reply; 29+ messages in thread
From: Jeff King @ 2017-04-20 15:35 UTC (permalink / raw)
  To: Duy Nguyen; +Cc: Junio C Hamano, git

On Thu, Apr 20, 2017 at 04:59:21PM +0700, Duy Nguyen wrote:

> On Wed, Apr 19, 2017 at 10:37:21PM -0700, Junio C Hamano wrote:
> > * nd/worktree-add-lock (2017-04-16) 2 commits
> >  - SQUASH???
> >  - worktree add: add --lock option
> > 
> >  Allow to lock a worktree immediately after it's created. This helps
> >  prevent a race between "git worktree add; git worktree lock" and
> >  "git worktree prune".
> > 
> >  Waiting for a response to SQUASH???
> 
> Looking good. I would add some comment, lest ';' feel lonely. But it's
> really personal taste.
> 
> -- 8< --
> diff --git a/builtin/worktree.c b/builtin/worktree.c
> index 5ebdcce793..bc75676bf3 100644
> --- a/builtin/worktree.c
> +++ b/builtin/worktree.c
> @@ -310,7 +310,7 @@ static int add_worktree(const char *path, const char *refname,
>  	strbuf_reset(&sb);
>  	strbuf_addf(&sb, "%s/locked", sb_repo.buf);
>  	if (!ret && opts->keep_locked)
> -		;
> +		;	      /* --lock wants to keep "locked" file */
>  	else
>  		unlink_or_warn(sb.buf);

I know this is just a drive-by comment, but given that the "else" is the
only thing that does anything, would it be simpler to flip the
conditional? The strbuf manipulation above also could go inside it.
E.g.:

diff --git a/builtin/worktree.c b/builtin/worktree.c
index 5ebdcce79..05ade547c 100644
--- a/builtin/worktree.c
+++ b/builtin/worktree.c
@@ -307,12 +307,11 @@ static int add_worktree(const char *path, const char *refname,
 	junk_git_dir = NULL;
 
 done:
-	strbuf_reset(&sb);
-	strbuf_addf(&sb, "%s/locked", sb_repo.buf);
-	if (!ret && opts->keep_locked)
-		;
-	else
+	if (ret || !opts->keep_locked) {
+		strbuf_reset(&sb);
+		strbuf_addf(&sb, "%s/locked", sb_repo.buf);
 		unlink_or_warn(sb.buf);
+	}
 	argv_array_clear(&child_env);
 	strbuf_release(&sb);
 	strbuf_release(&symref);

(Since it's in its own block I'd also consider just introducing a new
variable for the path, but I guess reusing "sb" for each path is already
a pattern in the function).

-Peff

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20 13:24   ` Johannes Schindelin
@ 2017-04-20 16:56     ` Brandon Williams
  2017-04-20 23:18       ` Brandon Williams
  0 siblings, 1 reply; 29+ messages in thread
From: Brandon Williams @ 2017-04-20 16:56 UTC (permalink / raw)
  To: Johannes Schindelin; +Cc: Lars Schneider, Junio C Hamano, git, peff

On 04/20, Johannes Schindelin wrote:
> Hi Lars & Junio,
> 
> On Thu, 20 Apr 2017, Lars Schneider wrote:
> 
> > > * bw/forking-and-threading (2017-04-19) 11 commits
> > > - run-command: block signals between fork and execve
> > > - run-command: add note about forking and threading
> > > - run-command: handle dup2 and close errors in child
> > > - run-command: eliminate calls to error handling functions in child
> > > - run-command: don't die in child when duping /dev/null
> > > - run-command: prepare child environment before forking
> > > - string-list: add string_list_remove function
> > > - run-command: use the async-signal-safe execv instead of execvp
> > > - run-command: prepare command before forking
> > > - t0061: run_command executes scripts without a #! line
> > > - t5550: use write_script to generate post-update hook
> > > 
> > > The "run-command" APIimplementation has been made more robust
> > > against dead-locking in a threaded environment.
> > > 
> > > Will merge to 'next'.
> > 
> > There might be a problem on Windows with this (that's just a hunch, i can't test this right now):
> > https://travis-ci.org/git/git/jobs/223830474
> 
> Thanks for keeping track of Travis' failure reports. From what I see, the
> latest iteration (which does not seem to have made it to `pu` yet) has the
> !MINGW prerequisite which should fix the issue. Hopefully my suggested
> addition to the commit message will make it into the commit history, too.

Thanks for catching this.  And as you pointed out the latest reroll
should fix the issue.

> 
> Ciao,
> Dscho

-- 
Brandon Williams

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20  9:59 ` Duy Nguyen
  2017-04-20 15:35   ` Jeff King
@ 2017-04-20 22:46   ` Junio C Hamano
  1 sibling, 0 replies; 29+ messages in thread
From: Junio C Hamano @ 2017-04-20 22:46 UTC (permalink / raw)
  To: Duy Nguyen; +Cc: git

Duy Nguyen <pclouds@gmail.com> writes:

> Looking good. I would add some comment, lest ';' feel lonely. But it's
> really personal taste.

... which matches mine.   Thanks for the update (which I'll squash in).

>
> -- 8< --
> diff --git a/builtin/worktree.c b/builtin/worktree.c
> index 5ebdcce793..bc75676bf3 100644
> --- a/builtin/worktree.c
> +++ b/builtin/worktree.c
> @@ -310,7 +310,7 @@ static int add_worktree(const char *path, const char *refname,
>  	strbuf_reset(&sb);
>  	strbuf_addf(&sb, "%s/locked", sb_repo.buf);
>  	if (!ret && opts->keep_locked)
> -		;
> +		;	      /* --lock wants to keep "locked" file */
>  	else
>  		unlink_or_warn(sb.buf);
>  	argv_array_clear(&child_env);
> -- 8< --

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20 15:35   ` Jeff King
@ 2017-04-20 22:51     ` Junio C Hamano
  0 siblings, 0 replies; 29+ messages in thread
From: Junio C Hamano @ 2017-04-20 22:51 UTC (permalink / raw)
  To: Jeff King; +Cc: Duy Nguyen, git

Jeff King <peff@peff.net> writes:

>>  	if (!ret && opts->keep_locked)
>> -		;
>> +		;	      /* --lock wants to keep "locked" file */
>>  	else
>>  		unlink_or_warn(sb.buf);
>
> I know this is just a drive-by comment, but given that the "else" is the
> only thing that does anything, would it be simpler to flip the
> conditional? The strbuf manipulation above also could go inside it.
> E.g.:
>
> diff --git a/builtin/worktree.c b/builtin/worktree.c
> index 5ebdcce79..05ade547c 100644
> --- a/builtin/worktree.c
> +++ b/builtin/worktree.c
> @@ -307,12 +307,11 @@ static int add_worktree(const char *path, const char *refname,
>  	junk_git_dir = NULL;
>  
>  done:
> -	strbuf_reset(&sb);
> -	strbuf_addf(&sb, "%s/locked", sb_repo.buf);
> -	if (!ret && opts->keep_locked)
> -		;
> -	else
> +	if (ret || !opts->keep_locked) {
> +		strbuf_reset(&sb);
> +		strbuf_addf(&sb, "%s/locked", sb_repo.buf);
>  		unlink_or_warn(sb.buf);
> +	}
>  	argv_array_clear(&child_env);
>  	strbuf_release(&sb);
>  	strbuf_release(&symref);
>
> (Since it's in its own block I'd also consider just introducing a new
> variable for the path, but I guess reusing "sb" for each path is already
> a pattern in the function).

Yeah, when I looked at Duy's version and thought about changing to
lose the empty statement myself, I was afraid that the "if"
condition might become less transparent to grasp, but now I see what
you actually written down, I think "if we got an error, or the
caller didn't ask us to, remove that file" is easy enough to
understand.

Thanks.

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20 15:32 ` Lars Schneider
@ 2017-04-20 22:52   ` Junio C Hamano
  0 siblings, 0 replies; 29+ messages in thread
From: Junio C Hamano @ 2017-04-20 22:52 UTC (permalink / raw)
  To: Lars Schneider; +Cc: git

Lars Schneider <larsxschneider@gmail.com> writes:

> Sorry for sending this email multiple times. My mobile email client created html... Should be fixed now!
>
>> 
>> * ls/filter-process-delayed (2017-03-06) 1 commit
>> - convert: add "status=delayed" to filter process protocol
>> 
>> What's the status of this one???
>> cf. <xmqq60jmnmef.fsf@junio-linux.mtv.corp.google.com>
>
> posted v3 here:
> https://public-inbox.org/git/20170409191107.20547-1-larsxschneider@gmail.com/ 
>
> From my point of view the only open question is if we use an index ("delay-id" favored by Peff) or the path (favored by Taylor) to identify delayed items: 
> https://public-inbox.org/git/20170227095825.jhdspwy6oa6mvvvm@sigill.intra.peff.net/ 
> https://public-inbox.org/git/20170412173404.GA49694@Ida/#r 
>
> I slightly favor the path approach as it simplifies the protocol. But I also trust Peff's opinion.
>

Thanks for a nice summary.

>> * ls/travis-doc-asciidoctor (2017-04-16) 3 commits
>> (merged to 'next' on 2017-04-19 at 359c32953b)
>> + travis-ci: unset compiler for jobs that do not need one
>> + travis-ci: parallelize documentation build
>> + travis-ci: build documentation with AsciiDoc and Asciidoctor
>> 
>> Have Travis CI format the documentation with both AsciiDoc and
>> AsciiDoctor.
>> 
>> Will merge to 'master'.
>
> Can you hold this until next week?
> The I would post a v2 and check asciidoc stderr: 
> https://public-inbox.org/git/D9F45212-91F7-4BB1-A0EC-74A84DA81860@gmail.com/ 
>
> I can also post an additional patch if you prefer to merge now.

Thanks; will hold.

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
       [not found] ` <D61D47BD-9750-4FB6-892E-013504E03738@gmail.com>
  2017-04-20 13:24   ` Johannes Schindelin
@ 2017-04-20 22:58   ` Junio C Hamano
  2017-04-21  9:50     ` Johannes Schindelin
  1 sibling, 1 reply; 29+ messages in thread
From: Junio C Hamano @ 2017-04-20 22:58 UTC (permalink / raw)
  To: Lars Schneider; +Cc: git, peff, bmwill, Johannes Schindelin

Lars Schneider <larsxschneider@gmail.com> writes:

>> * bw/forking-and-threading (2017-04-19) 11 commits
>> - run-command: block signals between fork and execve
>> - run-command: add note about forking and threading
>> - run-command: handle dup2 and close errors in child
>> - run-command: eliminate calls to error handling functions in child
>> - run-command: don't die in child when duping /dev/null
>> - run-command: prepare child environment before forking
>> - string-list: add string_list_remove function
>> - run-command: use the async-signal-safe execv instead of execvp
>> - run-command: prepare command before forking
>> - t0061: run_command executes scripts without a #! line
>> - t5550: use write_script to generate post-update hook
>> 
>> The "run-command" APIimplementation has been made more robust
>> against dead-locking in a threaded environment.
>> 
>> Will merge to 'next'.
>
> There might be a problem on Windows with this (that's just a hunch, i can't test this right now):
> https://travis-ci.org/git/git/jobs/223830474

Thanks for keeping an eye on Travis output. My eyes learned to
ignore the Windows section as its failures often seem to be due to
timing out.

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20 16:56     ` Brandon Williams
@ 2017-04-20 23:18       ` Brandon Williams
  2017-04-21  0:56         ` Junio C Hamano
  0 siblings, 1 reply; 29+ messages in thread
From: Brandon Williams @ 2017-04-20 23:18 UTC (permalink / raw)
  To: Johannes Schindelin; +Cc: Lars Schneider, Junio C Hamano, git, peff

On 04/20, Brandon Williams wrote:
> On 04/20, Johannes Schindelin wrote:
> > Hi Lars & Junio,
> > 
> > On Thu, 20 Apr 2017, Lars Schneider wrote:
> > 
> > > > * bw/forking-and-threading (2017-04-19) 11 commits
> > > > - run-command: block signals between fork and execve
> > > > - run-command: add note about forking and threading
> > > > - run-command: handle dup2 and close errors in child
> > > > - run-command: eliminate calls to error handling functions in child
> > > > - run-command: don't die in child when duping /dev/null
> > > > - run-command: prepare child environment before forking
> > > > - string-list: add string_list_remove function
> > > > - run-command: use the async-signal-safe execv instead of execvp
> > > > - run-command: prepare command before forking
> > > > - t0061: run_command executes scripts without a #! line

I just double checked what differences existed between what I have
locally and what is queued at bw/forking-and-threading and it looks like
the changes (adding the !MINGW) to just this one patch were missed,
while the rest of them were picked up.  Just and fyi.

> > > > - t5550: use write_script to generate post-update hook
> > > > 
> > > > The "run-command" APIimplementation has been made more robust
> > > > against dead-locking in a threaded environment.
> > > > 
> > > > Will merge to 'next'.
> > > 
> > > There might be a problem on Windows with this (that's just a hunch, i can't test this right now):
> > > https://travis-ci.org/git/git/jobs/223830474
> > 
> > Thanks for keeping track of Travis' failure reports. From what I see, the
> > latest iteration (which does not seem to have made it to `pu` yet) has the
> > !MINGW prerequisite which should fix the issue. Hopefully my suggested
> > addition to the commit message will make it into the commit history, too.
> 
> Thanks for catching this.  And as you pointed out the latest reroll
> should fix the issue.
> 
> > 
> > Ciao,
> > Dscho
> 
> -- 
> Brandon Williams

-- 
Brandon Williams

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20 23:18       ` Brandon Williams
@ 2017-04-21  0:56         ` Junio C Hamano
  0 siblings, 0 replies; 29+ messages in thread
From: Junio C Hamano @ 2017-04-21  0:56 UTC (permalink / raw)
  To: Brandon Williams; +Cc: Johannes Schindelin, Lars Schneider, git, peff

Brandon Williams <bmwill@google.com> writes:

> On 04/20, Brandon Williams wrote:
>> On 04/20, Johannes Schindelin wrote:
>> > Hi Lars & Junio,
>> > 
>> > On Thu, 20 Apr 2017, Lars Schneider wrote:
>> > 
>> > > > * bw/forking-and-threading (2017-04-19) 11 commits
>> > > > - run-command: block signals between fork and execve
>> > > > - run-command: add note about forking and threading
>> > > > - run-command: handle dup2 and close errors in child
>> > > > - run-command: eliminate calls to error handling functions in child
>> > > > - run-command: don't die in child when duping /dev/null
>> > > > - run-command: prepare child environment before forking
>> > > > - string-list: add string_list_remove function
>> > > > - run-command: use the async-signal-safe execv instead of execvp
>> > > > - run-command: prepare command before forking
>> > > > - t0061: run_command executes scripts without a #! line
>
> I just double checked what differences existed between what I have
> locally and what is queued at bw/forking-and-threading and it looks like
> the changes (adding the !MINGW) to just this one patch were missed,
> while the rest of them were picked up.  Just and fyi.

Thanks.  Queued with a warning in the log message.


^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-20 22:58   ` Junio C Hamano
@ 2017-04-21  9:50     ` Johannes Schindelin
  2017-04-21 12:29       ` Christian Couder
  2017-04-24  0:25       ` Junio C Hamano
  0 siblings, 2 replies; 29+ messages in thread
From: Johannes Schindelin @ 2017-04-21  9:50 UTC (permalink / raw)
  To: Junio C Hamano; +Cc: Lars Schneider, git, peff, bmwill

Hi Junio,

On Thu, 20 Apr 2017, Junio C Hamano wrote:

> Lars Schneider <larsxschneider@gmail.com> writes:
> 
> >> * bw/forking-and-threading (2017-04-19) 11 commits
> >> - run-command: block signals between fork and execve
> >> - run-command: add note about forking and threading
> >> - run-command: handle dup2 and close errors in child
> >> - run-command: eliminate calls to error handling functions in child
> >> - run-command: don't die in child when duping /dev/null
> >> - run-command: prepare child environment before forking
> >> - string-list: add string_list_remove function
> >> - run-command: use the async-signal-safe execv instead of execvp
> >> - run-command: prepare command before forking
> >> - t0061: run_command executes scripts without a #! line
> >> - t5550: use write_script to generate post-update hook
> >> 
> >> The "run-command" APIimplementation has been made more robust
> >> against dead-locking in a threaded environment.
> >> 
> >> Will merge to 'next'.
> >
> > There might be a problem on Windows with this (that's just a hunch, i can't test this right now):
> > https://travis-ci.org/git/git/jobs/223830474
> 
> Thanks for keeping an eye on Travis output. My eyes learned to
> ignore the Windows section as its failures often seem to be due to
> timing out.

Part of the reason is that you push out all of the branches in one go,
typically at the very end of your work day. The idea of Continuous
Integration is a little orthogonal to that style, suggesting to build &
test whenever new changes come into the integration branch.

As a consequence, my original setup was a little overloaded: the VM sat
idle most of the time, and when you pushed, it was overloaded.

To accommodate even that use case, I managed to pry away some resources
that should be mostly idle at the time you push, and that should be able
to run up to 4 builds in parallel (the number "4" is not really magic, it
is the number of integration branches of git.git).

Since that "let's aggregate everything and only push out the final result
at the end of the day" approach does not really allow the Continuous
Testing to identify problems associated with individual topic branches, I
have another job that bisects the breakages (with all associated problems
I reported earlier, as you apply some patches on top of really ancient
commits and bisect wants to test all merge bases first) because the
required time *definitely* would let Travis time out all the time. Those
bisect results are even less visible than the Travis results, see e.g.
https://github.com/git/git/commit/2e3a8b9035a#commitcomment-21836854.

Having said that, I do not think that it makes sense for you to change
your habits, as proper Continuous Integration (as opposed to a variation
of Continuous Testing that we use, really) would take a lot more of a
change than you are comfortable with: it would look a lot more Pull
Request centric than the current mailing list centered process.

Ciao,
Dscho

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-21  9:50     ` Johannes Schindelin
@ 2017-04-21 12:29       ` Christian Couder
  2017-04-22 11:48         ` Johannes Schindelin
  2017-04-22 13:37         ` Johannes Sixt
  2017-04-24  0:25       ` Junio C Hamano
  1 sibling, 2 replies; 29+ messages in thread
From: Christian Couder @ 2017-04-21 12:29 UTC (permalink / raw)
  To: Johannes Schindelin; +Cc: Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

On Fri, Apr 21, 2017 at 11:50 AM, Johannes Schindelin
<Johannes.Schindelin@gmx.de> wrote:
>
> Since that "let's aggregate everything and only push out the final result
> at the end of the day" approach does not really allow the Continuous
> Testing to identify problems associated with individual topic branches, I
> have another job that bisects the breakages

Great!

> (with all associated problems
> I reported earlier, as you apply some patches on top of really ancient
> commits and bisect wants to test all merge bases first)

First bisect should ask you to test merge bases only if there are
"good" commits that are not ancestors of the "bad" commit.
So I don't know how you want to bisect, but it seems to me that you
could just go back to a previous know "good" commit when the one you
have is not an ancestor of the "bad" commit.
This would get rid of the whole merge base problem.

Second yeah there is probably an old bug in bisect there. In theory in
most cases bisect should ask you to test only one merge base, as:

- if the merge base is "bad", it means that the bug has been fixed
between the merge base and your "good" commit, and bisecting will
stop,
- if the merge base is "good", it means that all the merge bases that
are ancestor of this merge base are also good, so there is no need to
test them

But that theory can work only if bisect is asking you to test a merge
base that is a descendant of many (if not all the other merge bases),
but unfortunately I don't think bisect is smart enough for that yet.

> because the
> required time *definitely* would let Travis time out all the time. Those
> bisect results are even less visible than the Travis results, see e.g.
> https://github.com/git/git/commit/2e3a8b9035a#commitcomment-21836854.

Nice that they exists though!

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-21 12:29       ` Christian Couder
@ 2017-04-22 11:48         ` Johannes Schindelin
  2017-04-22 17:32           ` Christian Couder
  2017-04-22 13:37         ` Johannes Sixt
  1 sibling, 1 reply; 29+ messages in thread
From: Johannes Schindelin @ 2017-04-22 11:48 UTC (permalink / raw)
  To: Christian Couder; +Cc: Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

Hi Christian,

On Fri, 21 Apr 2017, Christian Couder wrote:

> On Fri, Apr 21, 2017 at 11:50 AM, Johannes Schindelin
> <Johannes.Schindelin@gmx.de> wrote:
> >
> > (with all associated problems I reported earlier, as you apply some
> > patches on top of really ancient commits and bisect wants to test all
> > merge bases first)
> 
> First bisect should ask you to test merge bases only if there are
> "good" commits that are not ancestors of the "bad" commit.

Please note that this is a stateless job. The only "state" I have is the
branch name.

So when something goes wrong, I have *no* indicator what is a known good
state.

The strategy I implemented was to use knowledge about the branches and
their relations. So when there is a bug in `pu`, the script first tests
whether the same test passes in `next`. And if it does, that is my
known-good state.

In the meantime, I cheat and mark all merge-bases as known-good, too. But
that is by no means a correct assumption: sometimes Junio decides to base
a patch on top of a really ancient commit, one that may be broken on
Windows. So there you are, I cannot win, I just tried to implement
something that works reasonably well, most of the time. It still takes way
too long.

> Second yeah there is probably an old bug in bisect there. In theory in
> most cases bisect should ask you to test only one merge base, as:
> 
> - if the merge base is "bad", it means that the bug has been fixed
> between the merge base and your "good" commit, and bisecting will
> stop,
> - if the merge base is "good", it means that all the merge bases that
> are ancestor of this merge base are also good, so there is no need to
> test them

That is not necessarily correct. If there are two merge bases, one may be
broken, and then that one is the first bad commit.

> > because the required time *definitely* would let Travis time out all
> > the time. Those bisect results are even less visible than the Travis
> > results, see e.g.
> > https://github.com/git/git/commit/2e3a8b9035a#commitcomment-21836854.
> 
> Nice that they exists though!

Yeah, well, it took enough of my time to implement, too ;-)

Ciao,
Dscho

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-21 12:29       ` Christian Couder
  2017-04-22 11:48         ` Johannes Schindelin
@ 2017-04-22 13:37         ` Johannes Sixt
  2017-04-24 14:24           ` Johannes Schindelin
  2017-04-25  2:00           ` Christian Couder
  1 sibling, 2 replies; 29+ messages in thread
From: Johannes Sixt @ 2017-04-22 13:37 UTC (permalink / raw)
  To: Christian Couder; +Cc: Johannes Schindelin, Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

Am 21.04.2017 um 14:29 schrieb Christian Couder:
> First bisect should ask you to test merge bases only if there are
> "good" commits that are not ancestors of the "bad" commit.

That's a tangent, but I have never understood why this needs to be so. 
Consider this:

     o--o--o--o--o--o--o--o--B
    /           /
  -o--o--o--o--g--o--o--o--o--G

When I mark B as bad and G as good, why would g have to be tested first? 
This is exactly what I do when I bisect in Git history: I mark the 
latest commits on git-gui and gitk sub-histories as good, because I know 
they can't possibly be bad. (In my setup, these two histories are ahead 
of pu and next.)

-- Hannes


^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-22 11:48         ` Johannes Schindelin
@ 2017-04-22 17:32           ` Christian Couder
  2017-04-24 14:08             ` Johannes Schindelin
  0 siblings, 1 reply; 29+ messages in thread
From: Christian Couder @ 2017-04-22 17:32 UTC (permalink / raw)
  To: Johannes Schindelin; +Cc: Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

Hi Dscho,

On Sat, Apr 22, 2017 at 1:48 PM, Johannes Schindelin
<Johannes.Schindelin@gmx.de> wrote:
>>
>> First bisect should ask you to test merge bases only if there are
>> "good" commits that are not ancestors of the "bad" commit.
>
> Please note that this is a stateless job. The only "state" I have is the
> branch name.
>
> So when something goes wrong, I have *no* indicator what is a known good
> state.

Maybe we could consider the last release a known good state?

> The strategy I implemented was to use knowledge about the branches and
> their relations. So when there is a bug in `pu`, the script first tests
> whether the same test passes in `next`. And if it does, that is my
> known-good state.
>
> In the meantime, I cheat and mark all merge-bases as known-good, too. But
> that is by no means a correct assumption: sometimes Junio decides to base
> a patch on top of a really ancient commit, one that may be broken on
> Windows. So there you are, I cannot win, I just tried to implement
> something that works reasonably well, most of the time. It still takes way
> too long.
>
>> Second yeah there is probably an old bug in bisect there. In theory in
>> most cases bisect should ask you to test only one merge base, as:
>>
>> - if the merge base is "bad", it means that the bug has been fixed
>> between the merge base and your "good" commit, and bisecting will
>> stop,
>> - if the merge base is "good", it means that all the merge bases that
>> are ancestor of this merge base are also good, so there is no need to
>> test them
>
> That is not necessarily correct. If there are two merge bases, one may be
> broken, and then that one is the first bad commit.

I wrote "in most cases" above because I think that it is not usual for
git.git to have branches that start from a commit that is not among
the set given by `git rev-list --first-parent master`.

So "in most cases" all the merge bases are among the first parents of
master, which means that if bisect was smart enough to first test the
last of these merge bases and to discard the merge bases that are
among its ancestors, then bisect would not need to test the other
merge bases.

Even if there were a few merge bases that were not among the first
parents of master, the number of merge bases to test could be reduced
a lot.

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-21  9:50     ` Johannes Schindelin
  2017-04-21 12:29       ` Christian Couder
@ 2017-04-24  0:25       ` Junio C Hamano
  2017-04-24 14:19         ` Johannes Schindelin
  1 sibling, 1 reply; 29+ messages in thread
From: Junio C Hamano @ 2017-04-24  0:25 UTC (permalink / raw)
  To: Johannes Schindelin; +Cc: Lars Schneider, git, peff, bmwill

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

> Part of the reason is that you push out all of the branches in one go,
> typically at the very end of your work day. The idea of Continuous
> Integration is a little orthogonal to that style, suggesting to build &
> test whenever new changes come into the integration branch.
>
> As a consequence, my original setup was a little overloaded: the VM sat
> idle most of the time, and when you pushed, it was overloaded.

I do not see pushing out all them in one go is making the problem
worse for you, though.

As of this writing, master..pu counts 60+ first-parent merges.
Instead of pushing out the final one at the end of the day, I could
push out after every merge.  Behind the scenes, because some topics
are extended or tweaked while I read the list discussion, the number
of merges I am doing during a day is about twice or more than that
before I reach the final version for the day.  

Many issues can be noticed locally even before the patches hit a
topic, before the topic gets merged to 'pu', or before the tentative
'pu' is pushed out, and breakage at each of these points can be
locally corrected without bothering external test setups.  I've been
assuming that pushing out all in one go at the end will help
reducing the load at external test setups.

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-22 17:32           ` Christian Couder
@ 2017-04-24 14:08             ` Johannes Schindelin
  0 siblings, 0 replies; 29+ messages in thread
From: Johannes Schindelin @ 2017-04-24 14:08 UTC (permalink / raw)
  To: Christian Couder; +Cc: Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

Hi Christian,

On Sat, 22 Apr 2017, Christian Couder wrote:

> On Sat, Apr 22, 2017 at 1:48 PM, Johannes Schindelin
> <Johannes.Schindelin@gmx.de> wrote:
> >>
> >> First bisect should ask you to test merge bases only if there are
> >> "good" commits that are not ancestors of the "bad" commit.
> >
> > Please note that this is a stateless job. The only "state" I have is
> > the branch name.
> >
> > So when something goes wrong, I have *no* indicator what is a known
> > good state.
> 
> Maybe we could consider the last release a known good state?

You mean the latest release. I would expect that we won't have a last
release in a long time...

Using the latest release as a 'known good' would not improve on using the
strategy I chose, as the latest release is at most up-to-date with
maint/master. In the case of `pu`, for example, it is much better to test
`next` and use it as a known good state if it passes.

If we had a Pull Request centric workflow with one integration branch (as
opposed to four), it would be relatively easy, as `master` would always be
expected to serve as a "known good" state, and it would be the policy that
the build has to pass before Pull Requests would be accepted.

But we don't, and that's that.

Ciao,
Johannes

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-24  0:25       ` Junio C Hamano
@ 2017-04-24 14:19         ` Johannes Schindelin
  2017-04-24 15:18           ` Ævar Arnfjörð Bjarmason
  0 siblings, 1 reply; 29+ messages in thread
From: Johannes Schindelin @ 2017-04-24 14:19 UTC (permalink / raw)
  To: Junio C Hamano; +Cc: Lars Schneider, git, peff, bmwill

Hi Junio,

On Sun, 23 Apr 2017, Junio C Hamano wrote:

> Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:
> 
> > Part of the reason is that you push out all of the branches in one go,
> > typically at the very end of your work day. The idea of Continuous
> > Integration is a little orthogonal to that style, suggesting to build
> > & test whenever new changes come into the integration branch.
> >
> > As a consequence, my original setup was a little overloaded: the VM
> > sat idle most of the time, and when you pushed, it was overloaded.
> 
> I do not see pushing out all them in one go is making the problem worse
> for you, though.

Oh no, you don't see that? Then let me spell it out a little more
clearly: when you push out four branches at the same time, the same
Virtual Machine that hosts all of the build agents has to build each and
everyone of them, then run the entire test suite.

As I have pointed out at several occasions (but I was probably complaining
too much about it, so you probably ignored it), the test suite uses shell
scripting a lot, and as a consequence it is really, really slow on
Windows. Meaning that even on a high-end VM, it typically takes 1.5 hours
to run the test suite. That's without SVN tests.

So now we have up to four build agents banging at the same CPU and RAM,
competing for resources. Now it takes more like 2-3 hours to run the
entire build & test.

The situation usually gets a little worse, even: you sometimes push out
several iterations of `pu` in relatively rapid succession, "rapid" being
relative to the time taken by the builds.

That means that there are sometimes four jobs still hogging the VM when
the next request to build & test `pu` arrives, and sometimes there is
another one queued before the first job finishes.

Naturally, the last two jobs will have started barely before Travis
decides that it waited long enough (3 hours) to call it quits.

To answer your implied question: the situation would be much, much better
if the branches with more time in-between.

But as I said, I understand that it would be asking you way too much to
change your process that seems to work well for you.

> As of this writing, master..pu counts 60+ first-parent merges.
> Instead of pushing out the final one at the end of the day, I could
> push out after every merge.  Behind the scenes, because some topics
> are extended or tweaked while I read the list discussion, the number
> of merges I am doing during a day is about twice or more than that
> before I reach the final version for the day.  
> 
> Many issues can be noticed locally even before the patches hit a
> topic, before the topic gets merged to 'pu', or before the tentative
> 'pu' is pushed out, and breakage at each of these points can be
> locally corrected without bothering external test setups.  I've been
> assuming that pushing out all in one go at the end will help
> reducing the load at external test setups.

Pushing out only four updates at the end of the day is probably better
than pushing after every merge, for sure.

Ciao,
Dscho

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-22 13:37         ` Johannes Sixt
@ 2017-04-24 14:24           ` Johannes Schindelin
  2017-04-24 16:34             ` Philip Oakley
  2017-04-25  2:00           ` Christian Couder
  1 sibling, 1 reply; 29+ messages in thread
From: Johannes Schindelin @ 2017-04-24 14:24 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: Christian Couder, Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

Hi Hannes,

On Sat, 22 Apr 2017, Johannes Sixt wrote:

> Am 21.04.2017 um 14:29 schrieb Christian Couder:
> > First bisect should ask you to test merge bases only if there are
> > "good" commits that are not ancestors of the "bad" commit.
> 
> That's a tangent, but I have never understood why this needs to be so.
> Consider this:
> 
>    o--o--o--o--o--o--o--o--B
>    /           /
>  -o--o--o--o--g--o--o--o--o--G
> 
> When I mark B as bad and G as good, why would g have to be tested first? This
> is exactly what I do when I bisect in Git history: I mark the latest commits
> on git-gui and gitk sub-histories as good, because I know they can't possibly
> be bad. (In my setup, these two histories are ahead of pu and next.)

I guess the idea behind bisect's reasoning is that you could have merged
the "wrong" branch first.

Ciao,
Dscho

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-24 14:19         ` Johannes Schindelin
@ 2017-04-24 15:18           ` Ævar Arnfjörð Bjarmason
  2017-04-25  0:56             ` Junio C Hamano
  0 siblings, 1 reply; 29+ messages in thread
From: Ævar Arnfjörð Bjarmason @ 2017-04-24 15:18 UTC (permalink / raw)
  To: Johannes Schindelin; +Cc: Junio C Hamano, Lars Schneider, Git Mailing List, Jeff King, Brandon Williams

On Mon, Apr 24, 2017 at 4:19 PM, Johannes Schindelin
<Johannes.Schindelin@gmx.de> wrote:
> Hi Junio,
>
> On Sun, 23 Apr 2017, Junio C Hamano wrote:
>
>> Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:
>>
>> > Part of the reason is that you push out all of the branches in one go,
>> > typically at the very end of your work day. The idea of Continuous
>> > Integration is a little orthogonal to that style, suggesting to build
>> > & test whenever new changes come into the integration branch.
>> >
>> > As a consequence, my original setup was a little overloaded: the VM
>> > sat idle most of the time, and when you pushed, it was overloaded.
>>
>> I do not see pushing out all them in one go is making the problem worse
>> for you, though.
>
> Oh no, you don't see that? Then let me spell it out a little more
> clearly: when you push out four branches at the same time, the same
> Virtual Machine that hosts all of the build agents has to build each and
> everyone of them, then run the entire test suite.
>
> As I have pointed out at several occasions (but I was probably complaining
> too much about it, so you probably ignored it), the test suite uses shell
> scripting a lot, and as a consequence it is really, really slow on
> Windows. Meaning that even on a high-end VM, it typically takes 1.5 hours
> to run the test suite. That's without SVN tests.
>
> So now we have up to four build agents banging at the same CPU and RAM,
> competing for resources. Now it takes more like 2-3 hours to run the
> entire build & test.
>
> The situation usually gets a little worse, even: you sometimes push out
> several iterations of `pu` in relatively rapid succession, "rapid" being
> relative to the time taken by the builds.
>
> That means that there are sometimes four jobs still hogging the VM when
> the next request to build & test `pu` arrives, and sometimes there is
> another one queued before the first job finishes.
>
> Naturally, the last two jobs will have started barely before Travis
> decides that it waited long enough (3 hours) to call it quits.
>
> To answer your implied question: the situation would be much, much better
> if the branches with more time in-between.
>
> But as I said, I understand that it would be asking you way too much to
> change your process that seems to work well for you.

Is getting the results of these builds time-critical? If not perhaps
an acceptable solution would be to use a source repo that's
time-delayed, e.g. 24hrs behind on average from Junio's git.git, and
where commits are pushed in at some configurable trickle.

>> As of this writing, master..pu counts 60+ first-parent merges.
>> Instead of pushing out the final one at the end of the day, I could
>> push out after every merge.  Behind the scenes, because some topics
>> are extended or tweaked while I read the list discussion, the number
>> of merges I am doing during a day is about twice or more than that
>> before I reach the final version for the day.
>>
>> Many issues can be noticed locally even before the patches hit a
>> topic, before the topic gets merged to 'pu', or before the tentative
>> 'pu' is pushed out, and breakage at each of these points can be
>> locally corrected without bothering external test setups.  I've been
>> assuming that pushing out all in one go at the end will help
>> reducing the load at external test setups.
>
> Pushing out only four updates at the end of the day is probably better
> than pushing after every merge, for sure.
>
> Ciao,
> Dscho

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-24 14:24           ` Johannes Schindelin
@ 2017-04-24 16:34             ` Philip Oakley
  2017-04-25  2:17               ` Christian Couder
  0 siblings, 1 reply; 29+ messages in thread
From: Philip Oakley @ 2017-04-24 16:34 UTC (permalink / raw)
  To: Johannes Schindelin, Johannes Sixt; +Cc: Christian Couder, Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

From: "Johannes Schindelin" <Johannes.Schindelin@gmx.de>
> Hi Hannes,
>
> On Sat, 22 Apr 2017, Johannes Sixt wrote:
>
>> Am 21.04.2017 um 14:29 schrieb Christian Couder:
>> > First bisect should ask you to test merge bases only if there are
>> > "good" commits that are not ancestors of the "bad" commit.
>>
>> That's a tangent, but I have never understood why this needs to be so.
>> Consider this:
>>
>>    o--o--o--o--o--o--o--o--B
>>    /           /
>>  -o--o--o--o--g--o--o--o--o--G
>>
>> When I mark B as bad and G as good, why would g have to be tested first? 
>> This
>> is exactly what I do when I bisect in Git history: I mark the latest 
>> commits
>> on git-gui and gitk sub-histories as good, because I know they can't 
>> possibly
>> be bad. (In my setup, these two histories are ahead of pu and next.)
>
> I guess the idea behind bisect's reasoning is that you could have merged
> the "wrong" branch first.
>
> Ciao,
> Dscho
>

Sorry if I'm bikeshedding here, but it does look like adding an alternate 
'bisect' strategy may be useful for this style of integration testing.

To me, given the multiplicity of independent branches being brought 
together, it should be possible to do a check on each of the branches 
separately, before the tests along the line of integration . The tests would 
not be a true 'bisection' as they are separate single point tests, but they 
do establish good commits at the tips of those branches.

Thus, for each of the merges in the --first-parent traversal, the option 
could test (in the OS of choice) the *second parent* commit of the merge. 
This sets the known good points. The breakages during the integration then 
become easier to bisect, as it is only looking for the integration of a good 
series into mainline that breaks. [1]

In addition, an aside that dscho made earlier about the merge-base of some 
branches relative to Windows may have been missed. The normal bisect process 
assumes that we start from a set of good merge bases. However, dscho noticed 
that occasionally some series may choose an older point on maint (etc.) that 
itself would _not_ be classed as good when tested on Windows (or on other 
OS's). Those older mergebases can make the classic bisect across all the 
commits in the DAG between here and there a tortuous process, especially if 
the local OS implementation percieves the base points as bad! (which breaks 
expectations).

--
Philip


[1] It maybe that this can be approached via an alternate DAG, which could 
be 'faked' up as if each of the topic branches had been squashed from being 
a long series down to a single commit (the final tree of the series), and 
then likewise (same tree) for the integration merges, such that all the 
singleton commits are not tested, only the pre and post merge commits on the 
first-parent traverse. 


^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-24 15:18           ` Ævar Arnfjörð Bjarmason
@ 2017-04-25  0:56             ` Junio C Hamano
  0 siblings, 0 replies; 29+ messages in thread
From: Junio C Hamano @ 2017-04-25  0:56 UTC (permalink / raw)
  To: Ævar Arnfjörð Bjarmason; +Cc: Johannes Schindelin, Lars Schneider, Git Mailing List, Jeff King, Brandon Williams

Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:

> Is getting the results of these builds time-critical? If not perhaps
> an acceptable solution would be to use a source repo that's
> time-delayed, e.g. 24hrs behind on average from Junio's git.git, and
> where commits are pushed in at some configurable trickle.

Because the tip of 'pu' is updated every day, but the tips of
individual topics are not, the tip of the more stable side of
integration branches are less often updated, I would think that an
automated procedure (whether it is Travis or Dscho's Windows thing)
that is most efficient for us would:

 - treat updates to 'maint', 'master' and 'next' as they do right
   now;

 - use update to 'pu' as a mere "trigger" and not test 'pu' as an
   aggregate as it is not very interesting (merges near the tip may
   be merging known-broken topics, like yesterday's pushout).

And arrange 'pu' update to trigger a procedure like this:

 - Grab the list of merge commits beyond what matches 'next':

   $ git rev-list --first-parent '^pu^{/^### match next}' pu

 - For each of these merges:

   - If its second parent (i.e. the tip of a topic) was already
     tested with the current 'master', do not bother testing it
     again;

   - Otherwise, call its second parent "it" in the below:

     - Create a merge of it (and nothing else) on top of 'master',
       and test it.  Each topic is intended to be able to graduate
       without other topics in 'pu', so this will catch potential
       breakage early and help blocking a broken topic from getting
       merged to 'master'.

     - When the above fails, test it in isolation to see if it
       itself is broken.  If this succeeds, we may have discovered
       unintended interactions between the topic and other topics
       already in 'master'.  Or the topic itself may already been
       broken.

     - In either case, record the test result so that the same tip
       of the topic would not have to be tested again, merely
       because the tip of 'pu' got updated next day due to updates
       in other topics.

We can s/master/next/ in the above description (I am undecided on
pros-and-cons between using 'master' and 'next').

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-22 13:37         ` Johannes Sixt
  2017-04-24 14:24           ` Johannes Schindelin
@ 2017-04-25  2:00           ` Christian Couder
  2017-04-25  5:51             ` Johannes Sixt
  1 sibling, 1 reply; 29+ messages in thread
From: Christian Couder @ 2017-04-25  2:00 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: Johannes Schindelin, Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

On Sat, Apr 22, 2017 at 3:37 PM, Johannes Sixt <j6t@kdbg.org> wrote:
> Am 21.04.2017 um 14:29 schrieb Christian Couder:
>>
>> First bisect should ask you to test merge bases only if there are
>> "good" commits that are not ancestors of the "bad" commit.
>
>
> That's a tangent, but I have never understood why this needs to be so.
> Consider this:
>
>     o--o--o--o--o--o--o--o--B
>    /           /
>  -o--o--o--o--g--o--o--o--o--G
>
> When I mark B as bad and G as good, why would g have to be tested first?

It is because g could be bad if the bug has been fixed between g and G.
If this happens and we don't test g, we would give a wrong result.

> This is exactly what I do when I bisect in Git history: I mark the latest
> commits on git-gui and gitk sub-histories as good, because I know they can't
> possibly be bad. (In my setup, these two histories are ahead of pu and
> next.)

Yeah, it is safe to do that in this case as we test the merge bases.

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-24 16:34             ` Philip Oakley
@ 2017-04-25  2:17               ` Christian Couder
  0 siblings, 0 replies; 29+ messages in thread
From: Christian Couder @ 2017-04-25  2:17 UTC (permalink / raw)
  To: Philip Oakley; +Cc: Johannes Schindelin, Johannes Sixt, Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

On Mon, Apr 24, 2017 at 6:34 PM, Philip Oakley <philipoakley@iee.org> wrote:
>
> Sorry if I'm bikeshedding here, but it does look like adding an alternate
> 'bisect' strategy may be useful for this style of integration testing.
>
> To me, given the multiplicity of independent branches being brought
> together, it should be possible to do a check on each of the branches
> separately, before the tests along the line of integration . The tests would
> not be a true 'bisection' as they are separate single point tests, but they
> do establish good commits at the tips of those branches.
>
> Thus, for each of the merges in the --first-parent traversal, the option
> could test (in the OS of choice) the *second parent* commit of the merge.
> This sets the known good points. The breakages during the integration then
> become easier to bisect, as it is only looking for the integration of a good
> series into mainline that breaks. [1]

I think what you suggest is very similar in practice as the
--first-parent strategy that we have been suggesting for some years to
add to bisect as a GSoC project:

https://git.github.io/SoC-2017-Ideas/

> In addition, an aside that dscho made earlier about the merge-base of some
> branches relative to Windows may have been missed. The normal bisect process
> assumes that we start from a set of good merge bases.

When a good commit is not an ancestor of the bad commit, we test the
merge bases first and we abort if one of them is not good.

> However, dscho noticed
> that occasionally some series may choose an older point on maint (etc.) that
> itself would _not_ be classed as good when tested on Windows (or on other
> OS's). Those older mergebases can make the classic bisect across all the
> commits in the DAG between here and there a tortuous process, especially if
> the local OS implementation percieves the base points as bad! (which breaks
> expectations).

My understanding of the bisecting problem that Dscho is reporting is
that the process is sometimes buggy or tortuous because bisect is not
smart enough to optimize the testing of the merge bases when there are
a lot of them.

It is not because bisect expects or assumes that some merge bases are
good or bad when in fact they are not.

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-25  2:00           ` Christian Couder
@ 2017-04-25  5:51             ` Johannes Sixt
  2017-04-25  6:52               ` Junio C Hamano
  0 siblings, 1 reply; 29+ messages in thread
From: Johannes Sixt @ 2017-04-25  5:51 UTC (permalink / raw)
  To: Christian Couder; +Cc: Johannes Schindelin, Junio C Hamano, Lars Schneider, git, Jeff King, Brandon Williams

Am 25.04.2017 um 04:00 schrieb Christian Couder:
> On Sat, Apr 22, 2017 at 3:37 PM, Johannes Sixt <j6t@kdbg.org> wrote:
>> Am 21.04.2017 um 14:29 schrieb Christian Couder:
>>>
>>> First bisect should ask you to test merge bases only if there are
>>> "good" commits that are not ancestors of the "bad" commit.
>>
>>
>> That's a tangent, but I have never understood why this needs to be so.
>> Consider this:
>>
>>     o--o--o--o--o--o--o--o--B
>>    /           /
>>  -o--o--o--o--g--o--o--o--o--G
>>
>> When I mark B as bad and G as good, why would g have to be tested first?
>
> It is because g could be bad if the bug has been fixed between g and G.
> If this happens and we don't test g, we would give a wrong result.

Gah! So, a typical messy workflow "requires" this behavior. A clean 
branchy workflow like Git's does not because we know that a breakage is 
either on a topic branch, B, or a (the?) bad commit is an ancester of 
(the integration branch) G.

>
>> This is exactly what I do when I bisect in Git history: I mark the latest
>> commits on git-gui and gitk sub-histories as good, because I know they can't
>> possibly be bad. (In my setup, these two histories are ahead of pu and
>> next.)
>
> Yeah, it is safe to do that in this case as we test the merge bases.

The idea of marking git-gui and gitk histories that none of their 
commits is checked out: it erases all Git source code from the working 
directory, and a later bisection step places all code back and it 
requires a full build. Not a big deal with Git, but there are much 
larger code bases.

The current bisect behavior makes this idea unworkable. For me, it was a 
big step backwards when it was implemented. :-(

-- Hannes


^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-25  5:51             ` Johannes Sixt
@ 2017-04-25  6:52               ` Junio C Hamano
  2017-04-25 18:26                 ` Johannes Sixt
  0 siblings, 1 reply; 29+ messages in thread
From: Junio C Hamano @ 2017-04-25  6:52 UTC (permalink / raw)
  To: Johannes Sixt; +Cc: Christian Couder, Johannes Schindelin, Lars Schneider, git, Jeff King, Brandon Williams

Johannes Sixt <j6t@kdbg.org> writes:

> The idea of marking git-gui and gitk histories that none of their
> commits is checked out: it erases all Git source code from the working
> directory, and a later bisection step places all code back and it
> requires a full build. Not a big deal with Git, but there are much
> larger code bases.
>
> The current bisect behavior makes this idea unworkable. For me, it was
> a big step backwards when it was implemented. :-(

I think it is sort-of unfair to blame bisect for that.  It talks
more about how "coolest merge evar!" cross project merges are
unworkable in practice.  With a proper submodule support, nobody
should be doing it, but we started the cross project merge way
before we had submodules, so...

^ permalink raw reply	[flat|nested] 29+ messages in thread

* Re: What's cooking in git.git (Apr 2017, #04; Wed, 19)
  2017-04-25  6:52               ` Junio C Hamano
@ 2017-04-25 18:26                 ` Johannes Sixt
  0 siblings, 0 replies; 29+ messages in thread
From: Johannes Sixt @ 2017-04-25 18:26 UTC (permalink / raw)
  To: Junio C Hamano; +Cc: Christian Couder, Johannes Schindelin, Lars Schneider, git, Jeff King, Brandon Williams

Am 25.04.2017 um 08:52 schrieb Junio C Hamano:
> Johannes Sixt <j6t@kdbg.org> writes:
>
>> The idea of marking git-gui and gitk histories that none of their
>> commits is checked out: it erases all Git source code from the working
>> directory, and a later bisection step places all code back and it
>> requires a full build. Not a big deal with Git, but there are much
>> larger code bases.
>>
>> The current bisect behavior makes this idea unworkable. For me, it was
>> a big step backwards when it was implemented. :-(
>
> I think it is sort-of unfair to blame bisect for that.  It talks
> more about how "coolest merge evar!" cross project merges are
> unworkable in practice.

No, it has nothing to do with the "coolest merge ever". The same problem 
occurs when I'm working on a topic branch:

      A--....--B   <- topic
     /
  --g--....--G     <- integration branch ("good by default")

While fiddling with B, I find a bug and suspect it somewhere on the 
topic from A to B. When A has a change that causes a substantial or 
complete rebuild, testing g causes burden for no good reason.

But, as I said, this is a tangent. I have no mission to change git bisect.

-- Hannes


^ permalink raw reply	[flat|nested] 29+ messages in thread

end of thread, back to index

Thread overview: 29+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-04-20  5:37 What's cooking in git.git (Apr 2017, #04; Wed, 19) Junio C Hamano
2017-04-20  9:59 ` Duy Nguyen
2017-04-20 15:35   ` Jeff King
2017-04-20 22:51     ` Junio C Hamano
2017-04-20 22:46   ` Junio C Hamano
     [not found] ` <D61D47BD-9750-4FB6-892E-013504E03738@gmail.com>
2017-04-20 13:24   ` Johannes Schindelin
2017-04-20 16:56     ` Brandon Williams
2017-04-20 23:18       ` Brandon Williams
2017-04-21  0:56         ` Junio C Hamano
2017-04-20 22:58   ` Junio C Hamano
2017-04-21  9:50     ` Johannes Schindelin
2017-04-21 12:29       ` Christian Couder
2017-04-22 11:48         ` Johannes Schindelin
2017-04-22 17:32           ` Christian Couder
2017-04-24 14:08             ` Johannes Schindelin
2017-04-22 13:37         ` Johannes Sixt
2017-04-24 14:24           ` Johannes Schindelin
2017-04-24 16:34             ` Philip Oakley
2017-04-25  2:17               ` Christian Couder
2017-04-25  2:00           ` Christian Couder
2017-04-25  5:51             ` Johannes Sixt
2017-04-25  6:52               ` Junio C Hamano
2017-04-25 18:26                 ` Johannes Sixt
2017-04-24  0:25       ` Junio C Hamano
2017-04-24 14:19         ` Johannes Schindelin
2017-04-24 15:18           ` Ævar Arnfjörð Bjarmason
2017-04-25  0:56             ` Junio C Hamano
2017-04-20 15:32 ` Lars Schneider
2017-04-20 22:52   ` Junio C Hamano

git@vger.kernel.org mailing list mirror (one of many)

Archives are clonable:
	git clone --mirror https://public-inbox.org/git
	git clone --mirror http://ou63pmih66umazou.onion/git
	git clone --mirror http://czquwvybam4bgbro.onion/git
	git clone --mirror http://hjrcffqmbrq6wope.onion/git

Newsgroups are available over NNTP:
	nntp://news.public-inbox.org/inbox.comp.version-control.git
	nntp://ou63pmih66umazou.onion/inbox.comp.version-control.git
	nntp://czquwvybam4bgbro.onion/inbox.comp.version-control.git
	nntp://hjrcffqmbrq6wope.onion/inbox.comp.version-control.git
	nntp://news.gmane.org/gmane.comp.version-control.git

 note: .onion URLs require Tor: https://www.torproject.org/
       or Tor2web: https://www.tor2web.org/

AGPL code for this site: git clone https://public-inbox.org/ public-inbox