{"thread":{"id":"48046","subject":"What's cooking in git.git (Mar 2018, #03; Wed, 14)","startedAt":"2018-03-15T01:35:03Z","lastAt":"2018-03-19T21:16:24Z","messageCount":13,"participants":["Junio C Hamano","Duy Nguyen","Ævar Arnfjörð Bjarmason","Lars Schneider","Jonathan Tan","Jonathan Nieder","Derrick Stolee"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"341732","messageId":"xmqqefkm6s06.fsf@gitster-ct.c.googlers.com","threadId":"48046","inReplyTo":null,"subject":"What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2018-03-15T01:34:49Z","receivedAt":"2018-03-15T01:35:03Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Here are the topics that have been cooking.  Commits prefixed with\n'-' are only in 'pu' (proposed updates) while commits prefixed with\n'+' are in 'next'.  The ones marked with '.' do not appear in any of\nthe integration branches, but I am still holding onto them.\n\nYou can find the changes described here in the integration branches\nof the repositories listed at\n\n    http://git-blame.blogspot.com/p/git-public-repositories.html\n\n--------------------------------------------------\n[Graduated to \"master\"]\n\n* ab/gc-auto-in-commit (2018-03-01) 1 commit\n  (merged to 'next' on 2018-03-02 at 96a5a4d629)\n + commit: run git gc --auto just before the post-commit hook\n\n \"git commit\" used to run \"gc --auto\" near the end, which was lost\n when the command was reimplemented in C by mistake.\n\n\n* ab/pre-auto-gc-battery (2018-02-28) 1 commit\n  (merged to 'next' on 2018-03-06 at ca9cb273cb)\n + hooks/pre-auto-gc-battery: allow gc to run on non-laptops\n\n A sample auto-gc hook (in contrib/) to skip auto-gc while on\n battery has been updated to almost always allow running auto-gc\n unless on_ac_power command is absolutely sure that we are on\n battery power (earlier, it skipped unless the command is sure that\n we are on ac power).\n\n\n* ag/userdiff-go-funcname (2018-03-01) 1 commit\n  (merged to 'next' on 2018-03-02 at ea404d1be9)\n + userdiff: add built-in pattern for golang\n\n \"git diff\" and friends learned funcname patterns for Go language\n source files.\n\n\n* bp/untracked-cache-noflush (2018-02-28) 2 commits\n  (merged to 'next' on 2018-03-02 at 709887971b)\n + untracked cache: use git_env_bool() not getenv() for customization\n + dir.c: don't flag the index as dirty for changes to the untracked cache\n\n Writing out the index file when the only thing that changed in it\n is the untracked cache information is often wasteful, and this has\n been optimized out.\n\n\n* ds/find-unique-abbrev-optim (2018-02-27) 1 commit\n  (merged to 'next' on 2018-03-02 at 0b6d4f9335)\n + sha1_name: fix uninitialized memory errors\n\n While finding unique object name abbreviation, the code may\n accidentally have read beyond the end of the array of object names\n in a pack.\n\n\n* ds/mark-parents-uninteresting-optim (2018-02-27) 1 commit\n  (merged to 'next' on 2018-03-02 at 5a42c79806)\n + revision.c: reduce object database queries\n\n Micro optimization in revision traversal code.\n\n\n* jc/test-must-be-empty (2018-02-27) 1 commit\n  (merged to 'next' on 2018-03-02 at ec129f1b97)\n + test_must_be_empty: make sure the file exists, not just empty\n\n Test framework tweak to catch developer thinko.\n\n\n* jh/status-no-ahead-behind (2018-01-24) 4 commits\n  (merged to 'next' on 2018-03-02 at 68bde8d571)\n + status: support --no-ahead-behind in long format\n + status: update short status to respect --no-ahead-behind\n + status: add --[no-]ahead-behind to status and commit for V2 format.\n + stat_tracking_info: return +1 when branches not equal\n\n \"git status\" can spend a lot of cycles to compute the relation\n between the current branch and its upstream, which can now be\n disabled with \"--no-ahead-behind\" option.\n\n\n* jk/add-i-diff-filter (2018-03-05) 2 commits\n  (merged to 'next' on 2018-03-08 at 6ef737add3)\n + add--interactive: detect bogus diffFilter output\n + t3701: add a test for interactive.diffFilter\n\n The \"interactive.diffFilter\" used by \"git add -i\" must retain\n one-to-one correspondence between its input and output, but it was\n not enforced and caused end-user confusion.  We now at least make\n sure the filtered result has the same number of lines as its input\n to detect a broken filter.\n\n\n* jk/smart-http-protocol-doc-fix (2018-03-05) 1 commit\n  (merged to 'next' on 2018-03-08 at 599b1a7c42)\n + smart-http: document flush after \"# service\" line\n\n A doc update.\n\n\n* ma/roll-back-lockfiles (2018-02-28) 5 commits\n  (merged to 'next' on 2018-03-06 at be29bf891c)\n + sequencer: do not roll back lockfile unnecessarily\n + merge: always roll back lock in `checkout_fast_forward()`\n + merge-recursive: always roll back lock in `merge_recursive_generic()`\n + sequencer: always roll back lock in `do_recursive_merge()`\n + sequencer: make lockfiles non-static\n (this branch is used by ma/skip-writing-unchanged-index.)\n\n Some codepaths used to take a lockfile and did not roll it back;\n they are automatically rolled back at program exit, so there is no\n real \"breakage\", but it still is a good practice to roll back when\n you are done with a lockfile.\n\n\n* mk/doc-pretty-fill (2018-02-27) 1 commit\n  (merged to 'next' on 2018-03-02 at 623461b127)\n + docs/pretty-formats: fix typo '% <(<N>)' -> '%<|(<N>)'\n\n Docfix.\n\n\n* nd/diff-stat-with-summary (2018-02-27) 2 commits\n  (merged to 'next' on 2018-03-06 at d543f92f5e)\n + diff: add --compact-summary\n + diff.c: refactor pprint_rename() to use strbuf\n\n \"git diff\" and friends learned \"--compact-summary\" that shows the\n information usually given with the \"--summary\" option on the same\n line as the diffstat output of the \"--stat\" option (which saves\n vertical space and keeps info on a single path at the same place).\n\n\n* nd/object-allocation-comments (2018-03-06) 2 commits\n  (merged to 'next' on 2018-03-08 at 91d553a339)\n + object.h: realign object flag allocation comment\n + object.h: update flag allocation comment\n\n Code doc update.\n\n\n* nd/parseopt-completion (2018-03-07) 45 commits\n  (merged to 'next' on 2018-03-08 at 2461b7035d)\n + completion: more subcommands in _git_notes()\n + completion: complete --{reuse,reedit}-message= for all notes subcmds\n + completion: simplify _git_notes\n + completion: don't set PARSE_OPT_NOCOMPLETE on --rerere-autoupdate\n  (merged to 'next' on 2018-03-02 at d72a6525fd)\n + completion: use __gitcomp_builtin in _git_worktree\n + completion: use __gitcomp_builtin in _git_tag\n + completion: use __gitcomp_builtin in _git_status\n + completion: use __gitcomp_builtin in _git_show_branch\n + completion: use __gitcomp_builtin in _git_rm\n + completion: use __gitcomp_builtin in _git_revert\n + completion: use __gitcomp_builtin in _git_reset\n + completion: use __gitcomp_builtin in _git_replace\n + remote: force completing --mirror= instead of --mirror\n + completion: use __gitcomp_builtin in _git_remote\n + completion: use __gitcomp_builtin in _git_push\n + completion: use __gitcomp_builtin in _git_pull\n + completion: use __gitcomp_builtin in _git_notes\n + completion: use __gitcomp_builtin in _git_name_rev\n + completion: use __gitcomp_builtin in _git_mv\n + completion: use __gitcomp_builtin in _git_merge_base\n + completion: use __gitcomp_builtin in _git_merge\n + completion: use __gitcomp_builtin in _git_ls_remote\n + completion: use __gitcomp_builtin in _git_ls_files\n + completion: use __gitcomp_builtin in _git_init\n + completion: use __gitcomp_builtin in _git_help\n + completion: use __gitcomp_builtin in _git_grep\n + completion: use __gitcomp_builtin in _git_gc\n + completion: use __gitcomp_builtin in _git_fsck\n + completion: use __gitcomp_builtin in _git_fetch\n + completion: use __gitcomp_builtin in _git_difftool\n + completion: use __gitcomp_builtin in _git_describe\n + completion: use __gitcomp_builtin in _git_config\n + completion: use __gitcomp_builtin in _git_commit\n + completion: use __gitcomp_builtin in _git_clone\n + completion: use __gitcomp_builtin in _git_clean\n + completion: use __gitcomp_builtin in _git_cherry_pick\n + completion: use __gitcomp_builtin in _git_checkout\n + completion: use __gitcomp_builtin in _git_branch\n + completion: use __gitcomp_builtin in _git_apply\n + completion: use __gitcomp_builtin in _git_am\n + completion: use __gitcomp_builtin in _git_add\n + git-completion.bash: introduce __gitcomp_builtin\n + parse-options: let OPT__FORCE take optional flags argument\n + parse-options: add OPT_xxx_F() variants\n + parse-options: support --git-completion-helper\n\n The parse-options API has been taught an option to help the\n completion script; the command line completion has been updated to\n take advantage of this mechanism.\n\n\n* nd/worktree-move (2018-03-06) 8 commits\n  (merged to 'next' on 2018-03-06 at a26271e7de)\n + t2028: fix minor error and issues in newly-added \"worktree move\" tests\n  (merged to 'next' on 2018-03-02 at 5c514dfc92)\n + worktree remove: allow it when $GIT_WORK_TREE is already gone\n + worktree remove: new command\n + worktree move: refuse to move worktrees with submodules\n + worktree move: accept destination as directory\n + worktree move: new command\n + worktree.c: add update_worktree_location()\n + worktree.c: add validate_worktree()\n\n \"git worktree\" learned move and remove subcommands.\n\n\n* ot/ref-filter-cleanup (2018-02-21) 2 commits\n  (merged to 'next' on 2018-03-02 at 3b4c39a4b5)\n + ref-filter: get rid of goto\n + ref-filter: get rid of duplicate code\n\n Code cleanup.\n\n\n* pw/add-p-recount (2018-03-05) 9 commits\n  (merged to 'next' on 2018-03-06 at 68952f9bb0)\n + add -p: don't rely on apply's '--recount' option\n + add -p: fix counting when splitting and coalescing\n + add -p: calculate offset delta for edited patches\n + add -p: adjust offsets of subsequent hunks when one is skipped\n + t3701: add failing test for pathological context lines\n + t3701: don't hard code sha1 hash values\n + t3701: use test_write_lines and write_script\n + t3701: indent here documents\n + add -i: add function to format hunk header\n (this branch is used by pw/add-p-select.)\n\n \"git add -p\" has been lazy in coalescing split patches before\n passing the result to underlying \"git apply\", leading to corner\n case bugs; the logic to prepare the patch to be applied after hunk\n selections has been tightened.\n\n\n* rj/test-i18ngrep (2018-02-28) 2 commits\n  (merged to 'next' on 2018-03-06 at 7ea1a2352c)\n + t5536: simplify checking of messages output to stderr\n + t4151: consolidate multiple calls to test_i18ngrep\n\n Test updates.\n\n\n* rs/perf-repeat-thrice-by-default (2018-02-27) 1 commit\n  (merged to 'next' on 2018-03-02 at 4898b3c450)\n + perf: use GIT_PERF_REPEAT_COUNT=3 by default even without config file\n\n Perf test regression fix.\n\n\n* rv/grep-cleanup (2018-02-23) 2 commits\n  (merged to 'next' on 2018-03-02 at 4aafca15f9)\n + grep: simplify grep_oid and grep_file\n + grep: move grep_source_init outside critical section\n\n Threaded \"git grep\" has been optimized to avoid allocation in code\n section that is covered under a mutex.\n\n\n* sg/subtree-signed-commits (2018-02-23) 1 commit\n  (merged to 'next' on 2018-03-02 at c5f6fd33e6)\n + subtree: fix add and pull for GPG-signed commits\n\n \"git subtree\" script (in contrib/) scripted around \"git log\", whose\n output got affected by end-user configuration like log.showsignature\n\n\n* sg/test-x (2018-02-28) 11 commits\n  (merged to 'next' on 2018-03-06 at ab0684b27c)\n + travis-ci: run tests with '-x' tracing\n + t/README: add a note about don't saving stderr of compound commands\n + t1510-repo-setup: mark as untraceable with '-x'\n + t9903-bash-prompt: don't check the stderr of __git_ps1()\n + t5570-git-daemon: don't check the stderr of a subshell\n + t5526: use $TRASH_DIRECTORY to specify the path of GIT_TRACE log file\n + t5500-fetch-pack: don't check the stderr of a subshell\n + t3030-merge-recursive: don't check the stderr of a subshell\n + t1507-rev-parse-upstream: don't check the stderr of a shell function\n + t: add means to disable '-x' tracing for individual test scripts\n + t: prevent '-x' tracing from interfering with test helpers' stderr\n\n Running test scripts under -x option of the shell is often not a\n useful way to debug them, because the error messages from the\n commands tests try to capture and inspect are contaminated by the\n tracing output by the shell.  An earlier work done to make it more\n pleasant to run tests under -x with recent versions of bash is\n extended to cover posix shells that do not support BASH_XTRACEFD.\n\n\n* sg/travis-build-during-script-phase (2018-01-08) 1 commit\n  (merged to 'next' on 2018-03-02 at 29e1585ae7)\n + travis-ci: build Git during the 'script' phase\n\n Build the executable in 'script' phase in Travis CI integration, to\n follow the established practice, rather than during 'before_script'\n phase.  This allows the CI categorize the failures better ('failed'\n is project's fault, 'errored' is build environment's).\n\n--------------------------------------------------\n[New Topics]\n\n* ab/man-sec-list (2018-03-08) 1 commit\n  (merged to 'next' on 2018-03-09 at 9626b691e2)\n + git manpage: note git-security@googlegroups.com\n\n Doc update.\n\n Will merge to 'master'.\n\n\n* sg/cvs-tests-with-x (2018-03-08) 2 commits\n  (merged to 'next' on 2018-03-09 at 6ec749c7b7)\n + t9402-git-cvsserver-refs: don't check the stderr of a subshell\n + t9400-git-cvsserver-server: don't rely on the output of 'test_cmp'\n\n Allow running a couple of tests with \"sh -x\".\n\n Will merge to 'master'.\n\n\n* tl/userdiff-csharp-async (2018-03-08) 1 commit\n  (merged to 'next' on 2018-03-09 at 6dcf76e118)\n + userdiff.c: add C# async keyword in diff pattern\n\n Update funcname pattern used for C# to recognize \"async\" keyword.\n\n Will merge to 'master'.\n\n\n* ti/fetch-everything-local-optim (2018-03-14) 1 commit\n - fetch-pack.c: use oidset to check existence of loose object\n\n A \"git fetch\" from a repository with insane number of refs into a\n repository that is already up-to-date still wasted too many cycles\n making many lstat(2) calls to see if these objects at the tips\n exist as loose objects locally.  These lstat(2) calls are optimized\n away by enumerating all loose objects beforehand.\n\n It is unknown if the new strategy negatively affects existing use\n cases, fetching into a repository with many loose objects from a\n repository with small number of refs.\n\n Will merge to 'next'.\n\n\n* ab/nuke-emacs-contrib (2018-03-13) 1 commit\n - git{,-blame}.el: remove old bitrotting Emacs code\n\n The scripts in contrib/emacs/ have outlived their usefulness and\n have been removed.\n\n Will merge to 'next'.\n\n\n* bc/object-id (2018-03-14) 36 commits\n - convert: convert to struct object_id\n - sha1_file: introduce a constant for max header length\n - Convert lookup_replace_object to struct object_id\n - sha1_file: convert read_sha1_file to struct object_id\n - sha1_file: convert read_object_with_reference to object_id\n - tree-walk: convert tree entry functions to object_id\n - streaming: convert istream internals to struct object_id\n - tree-walk: convert get_tree_entry_follow_symlinks internals to object_id\n - builtin/notes: convert static functions to object_id\n - builtin/fmt-merge-msg: convert remaining code to object_id\n - sha1_file: convert sha1_object_info* to object_id\n - Convert remaining callers of sha1_object_info_extended to object_id\n - packfile: convert unpack_entry to struct object_id\n - sha1_file: convert retry_bad_packed_offset to struct object_id\n - sha1_file: convert assert_sha1_type to object_id\n - builtin/mktree: convert to struct object_id\n - streaming: convert open_istream to use struct object_id\n - sha1_file: convert check_sha1_signature to struct object_id\n - sha1_file: convert read_loose_object to use struct object_id\n - builtin/index-pack: convert struct ref_delta_entry to object_id\n - archive: convert sha1_file_to_archive to struct object_id\n - archive: convert write_archive_entry_fn_t to object_id\n - builtin/mktag: convert to struct object_id\n - replace_object: convert struct replace_object to object_id\n - send-pack: convert remaining functions to struct object_id\n - http-walker: convert struct object_request to use struct object_id\n - Convert find_unique_abbrev* to struct object_id\n - wt-status: convert struct wt_status_state to object_id\n - strbuf: convert strbuf_add_unique_abbrev to use struct object_id\n - ref-filter: convert grab_objectname to struct object_id\n - tree: convert read_tree_recursive to struct object_id\n - resolve-undo: convert struct resolve_undo_info to object_id\n - cache-tree: convert remnants to struct object_id\n - cache-tree: convert write_*_as_tree to object_id\n - builtin/write-tree: convert to struct object_id\n - bulk-checkin: convert index_bulk_checkin to struct object_id\n\n Conversion from uchar[20] to struct object_id continues.\n\n Will merge to 'next'.\n\n\n* jh/fsck-promisors (2018-03-13) 1 commit\n - sha1_file: restore OBJECT_INFO_QUICK functionality\n\n A hotfix to a topic that graduated recently.\n\n Will merge to 'next'.\n\n\n* ma/shortlog-revparse (2018-03-13) 3 commits\n - shortlog: do not accept revisions when run outside repo\n - shortlog: add usage-string for stdin-reading\n - git-shortlog.txt: reorder usages\n\n \"git shortlog cruft\" aborted with a BUG message when run outside a\n Git repository.  The command has been taught to complain about\n extra and unwanted arguments on its command line instead in such a\n case.\n\n Expecting a reroll.\n cf. <CAN0heSoDSf+3VmRowHmiTPDDzfdvUjQi+tjpPF6RW172ZE5BWw@mail.gmail.com>\n\n\n* ab/pcre-v2 (2018-03-14) 3 commits\n - Makefile: make USE_LIBPCRE=YesPlease mean v2, not v1\n - configure: detect redundant --with-libpcre & --with-libpcre1\n - configure: fix a regression in PCRE v1 detection\n\n Git can be built to use either v1 or v2 of the PCRE library, and so\n far, the build-time configuration USE_LIBPCRE=YesPlease instructed\n the build procedure to use v1, but now it means v2.  USE_LIBPCRE1\n and USE_LIBPCRE2 can be used to explicitly choose which version to\n use, as before.\n\n Will merge to 'next'.\n\n--------------------------------------------------\n[Stalled]\n\n* sb/blame-color (2018-02-13) 3 commits\n - builtin/blame: highlight recently changed lines\n - builtin/blame: add option to color metadata fields separately\n - builtin/blame: dim uninteresting metadata\n\n Expecting a reroll.\n cf. https://public-inbox.org/git/20171110011002.10179-1-sbeller@google.com/#t\n error messages are funny, can segfault, ...\n\n\n* av/fsmonitor-updates (2018-01-04) 6 commits\n - fsmonitor: use fsmonitor data in `git diff`\n - fsmonitor: remove debugging lines from t/t7519-status-fsmonitor.sh\n - fsmonitor: make output of test-dump-fsmonitor more concise\n - fsmonitor: update helper tool, now that flags are filled later\n - fsmonitor: stop inline'ing mark_fsmonitor_valid / _invalid\n - dir.c: update comments to match argument name\n\n Code clean-up on fsmonitor integration, plus optional utilization\n of the fsmonitor data in diff-files.\n\n Waiting for an update.\n cf. <alpine.DEB.2.21.1.1801042335130.32@MININT-6BKU6QN.europe.corp.microsoft.com>\n\n\n* pb/bisect-helper-2 (2017-10-28) 8 commits\n - t6030: make various test to pass GETTEXT_POISON tests\n - bisect--helper: `bisect_start` shell function partially in C\n - bisect--helper: `get_terms` & `bisect_terms` shell function in C\n - bisect--helper: `bisect_next_check` shell function in C\n - bisect--helper: `check_and_set_terms` shell function in C\n - wrapper: move is_empty_file() and rename it as is_empty_or_missing_file()\n - bisect--helper: `bisect_write` shell function in C\n - bisect--helper: `bisect_reset` shell function in C\n\n Expecting a reroll.\n cf. <0102015f5e5ee171-f30f4868-886f-47a1-a4e4-b4936afc545d-000000@eu-west-1.amazonses.com>\n\n\n* dj/runtime-prefix (2017-12-05) 4 commits\n . exec_cmd: RUNTIME_PREFIX on some POSIX systems\n . Makefile: add Perl runtime prefix support\n . Makefile: add support for \"perllibdir\"\n . Makefile: generate Perl header from template file\n\n A build-time option has been added to allow Git to be told to refer\n to its associated files relative to the main binary, in the same\n way that has been possible on Windows for quite some time, for\n Linux, BSDs and Darwin.\n\n Perhaps it is about time to reboot the effort?\n\n\n* mk/http-backend-content-length (2017-11-27) 4 commits\n - SQUASH???\n - t5560-http-backend-noserver.sh: add CONTENT_LENGTH cases\n - SQUASH???\n - http-backend: respect CONTENT_LENGTH as specified by rfc3875\n\n The http-backend (used for smart-http transport) used to slurp the\n whole input until EOF, without paying attention to CONTENT_LENGTH\n that is supplied in the environment and instead expecting the Web\n server to close the input stream.  This has been fixed.\n\n Expecting a reroll.\n Suggested fixes to be used when rerolling is queued, but I'd\n prefer _not_ squashing them myself.\n\n Also, it may be too complex solution for the problem.\n cf. <20171204171308.GA13332@sigill.intra.peff.net>\n\n\n* jk/drop-ancient-curl (2017-08-09) 5 commits\n - http: #error on too-old curl\n - curl: remove ifdef'd code never used with curl >=7.19.4\n - http: drop support for curl < 7.19.4\n - http: drop support for curl < 7.16.0\n - http: drop support for curl < 7.11.1\n\n Some code in http.c that has bitrot is being removed.\n\n Expecting a reroll.\n\n\n* mk/use-size-t-in-zlib (2017-08-10) 1 commit\n . zlib.c: use size_t for size\n\n The wrapper to call into zlib followed our long tradition to use\n \"unsigned long\" for sizes of regions in memory, which have been\n updated to use \"size_t\".\n\n Needs resurrecting by making sure the fix is good and still applies\n (or adjusted to today's codebase).\n\n--------------------------------------------------\n[Cooking]\n\n* bb/git-gui-ssh-key-files (2018-03-02) 2 commits\n - Merge branch 'bb/ssh-key-files' of git-gui into bb/git-gui-ssh-key-files\n - git-gui: search for all current SSH key types\n\n \"git gui\" learned that \"~/.ssh/id_ecdsa.pub\" and\n \"~/.ssh/id_ed25519.pub\" are also possible SSH key files.\n\n\n* bp/git-gui-bind-kp-enter (2018-03-02) 2 commits\n - Merge branch 'bp/bind-kp-enter' of git-gui into bp/git-gui-bind-kp-enter\n - git-gui: bind CTRL/CMD+numpad ENTER to do_commit\n\n \"git gui\" performs commit upon CTRL/CMD+ENTER but the\n CTRL/CMD+KP_ENTER (i.e. enter key on the numpad) did not have the\n same key binding.  It now does.\n\n\n* cb/git-gui-ttk-style (2018-03-05) 2 commits\n - Merge branch 'cb/ttk-style' of git-gui into cb/git-gui-ttk-style\n - git-gui: workaround ttk:style theme use\n\n \"git gui\" has been taught to work with old versions of tk (like\n 8.5.7) that do not support \"ttk::style theme use\" as a way to query\n the current theme.\n\n\n* nd/pack-objects-pack-struct (2018-03-05) 9 commits\n - pack-objects: reorder 'hash' to pack struct object_entry\n - pack-objects: refer to delta objects by index instead of pointer\n - pack-objects: move in_pack out of struct object_entry\n - pack-objects: move in_pack_pos out of struct object_entry\n - pack-objects: note about in_pack_header_size\n - pack-objects: use bitfield for object_entry::depth\n - pack-objects: use bitfield for object_entry::dfs_state\n - pack-objects: turn type and in_pack_type to bitfields\n - pack-objects: document holes in struct object_entry.h\n\n \"git pack-objects\" needs to allocate tons of \"struct object_entry\"\n while doing its work, and shrinking its size helps the performance\n quite a bit.\n\n Will merge to 'next'.\n\n\n* nd/repack-keep-pack (2018-03-07) 6 commits\n - SQUASH???\n - pack-objects: display progress in get_object_details()\n - pack-objects: show some progress when counting kept objects\n - gc --auto: exclude base pack if not enough mem to \"repack -ad\"\n - repack: add --keep-pack option\n - t7700: have closing quote of a test at the beginning of line\n\n \"git gc\" in a large repository takes a lot of time as it considers\n to repack all objects into one pack by default.  The command has\n been taught to pretend as if the largest existing packfile is\n marked with \".keep\" so that it is left untouched while objects in\n other packs and loose ones are repacked.\n\n Expecting a reroll.\n cf. <CACsJy8BW_EtxQvgL=YrCXCQY7cEWCQxgfkeH=Gd=X=uVYhPJcw@mail.gmail.com>\n Except for final finishing touches, this looked more-or-less ready\n for 'next'.\n\n\n* nd/worktree-prune (2018-03-06) 3 commits\n - worktree prune: improve prune logic when worktree is moved\n - worktree: delete dead code\n - gc.txt: more details about what gc does\n\n The way \"git worktree prune\" worked internally has been simplified,\n by assuming how \"git worktree move\" moves an existing worktree to a\n different place.\n\n Will merge to 'next'.\n\n\n* pw/add-p-select (2018-03-06) 3 commits\n - add -p: optimize line selection for short hunks\n - add -p: allow line selection to be inverted\n - add -p: select individual hunk lines\n\n \"git add -p\" interactive interface learned to let users choose\n individual added/removed lines to be used in the operation, instead\n of accepting or rejecting a whole hunk.\n\n Expecting a reroll.\n cf. <6476d776-dbf7-09cf-1c65-e413798b9987@talktalk.net>\n\n\n* ld/p4-unshelve (2018-02-22) 1 commit\n - git-p4: add unshelve command\n\n \"git p4\" learned to \"unshelve\" shelved commit from P4.\n\n Will hold, perhaps drop and use format-change that uses a proper \"diff\".\n cf. <CAE5ih7_ooDMqVtTMoQ70s5XCkncr04HY0JkqSp1UmKQeG81oaA@mail.gmail.com>\n\n\n* ps/contains-id-error-message (2018-03-06) 1 commit\n - parse-options: squelch usage help on certain errors\n\n \"git tag --contains no-such-commit\" gave a full list of options\n after giving an error message.\n\n Expecting a reroll.\n cf. <CAN0heSrnERYggyJ8bL1nAV=X2JQYS8aOkc6nWZhnZu3oqhzAfA@mail.gmail.com>\n Rebooted and fixed the root cause of the issue at a lower level.\n\n\n* nd/remove-ignore-env-field (2018-03-05) 5 commits\n - repository: delete ignore_env member\n - sha1_file.c: move delayed getenv(altdb) back to setup_git_env()\n - repository.c: delete dead functions\n - repository.c: move env-related setup code back to environment.c\n - repository: initialize the_repository in main()\n (this branch is used by sb/object-store and sb/packfiles-in-repository.)\n\n Code clean-up for the \"repository\" abstraction.\n\n Will merge to 'next'.\n\n\n* sb/object-store (2018-03-05) 27 commits\n - sha1_file: allow sha1_loose_object_info to handle arbitrary repositories\n - sha1_file: allow map_sha1_file to handle arbitrary repositories\n - sha1_file: allow map_sha1_file_1 to handle arbitrary repositories\n - sha1_file: allow open_sha1_file to handle arbitrary repositories\n - sha1_file: allow stat_sha1_file to handle arbitrary repositories\n - sha1_file: allow sha1_file_name to handle arbitrary repositories\n - sha1_file: add repository argument to sha1_loose_object_info\n - sha1_file: add repository argument to map_sha1_file\n - sha1_file: add repository argument to map_sha1_file_1\n - sha1_file: add repository argument to open_sha1_file\n - sha1_file: add repository argument to stat_sha1_file\n - sha1_file: add repository argument to sha1_file_name\n - sha1_file: allow prepare_alt_odb to handle arbitrary repositories\n - sha1_file: allow link_alt_odb_entries to handle arbitrary repositories\n - sha1_file: add repository argument to prepare_alt_odb\n - sha1_file: add repository argument to link_alt_odb_entries\n - sha1_file: add repository argument to read_info_alternates\n - sha1_file: add repository argument to link_alt_odb_entry\n - sha1_file: add raw_object_store argument to alt_odb_usable\n - pack: move approximate object count to object store\n - pack: move prepare_packed_git_run_once to object store\n - object-store: close all packs upon clearing the object store\n - object-store: move packed_git and packed_git_mru to object store\n - object-store: free alt_odb_list\n - object-store: move alt_odb_list and alt_odb_tail to object store\n - object-store: migrate alternates struct and functions from cache.h\n - repository: introduce raw object store field\n (this branch is used by sb/packfiles-in-repository; uses nd/remove-ignore-env-field.)\n\n Refactoring the internal global data structure to make it possible\n to open multiple repositories, work with and then close them.\n\n Rerolled by Duy on top of a separate preliminary clean-up topic.\n The resulting structure of the topics looked very sensible.\n\n Waiting for a follow-up discussion.\n\n\n* sb/packfiles-in-repository (2018-03-05) 12 commits\n - packfile: keep prepare_packed_git() private\n - packfile: allow find_pack_entry to handle arbitrary repositories\n - packfile: add repository argument to find_pack_entry\n - packfile: allow reprepare_packed_git to handle arbitrary repositories\n - packfile: allow prepare_packed_git to handle arbitrary repositories\n - packfile: allow prepare_packed_git_one to handle arbitrary repositories\n - packfile: add repository argument to reprepare_packed_git\n - packfile: add repository argument to prepare_packed_git\n - packfile: add repository argument to prepare_packed_git_one\n - packfile: allow install_packed_git to handle arbitrary repositories\n - packfile: allow rearrange_packed_git to handle arbitrary repositories\n - packfile: allow prepare_packed_git_mru to handle arbitrary repositories\n (this branch uses nd/remove-ignore-env-field and sb/object-store.)\n\n Refactoring of the internal global data structure continues.\n\n Waiting for a follow-up discussion.\n\n\n* ma/skip-writing-unchanged-index (2018-03-01) 1 commit\n - write_locked_index(): add flag to avoid writing unchanged index\n\n Internal API clean-up to allow write_locked_index() optionally skip\n writing the in-core index when it is not modified.\n\n Will merge to 'next'.\n\n\n* ab/perl-fixes (2018-03-05) 13 commits\n  (merged to 'next' on 2018-03-09 at 262d84c1ba)\n + perl Git::LoadCPAN: emit better errors under NO_PERL_CPAN_FALLBACKS\n + Makefile: add NO_PERL_CPAN_FALLBACKS knob\n + perl: move the perl/Git/FromCPAN tree to perl/FromCPAN\n + perl: generalize the Git::LoadCPAN facility\n + perl: move CPAN loader wrappers to another namespace\n + perl: update our copy of Mail::Address\n + perl: update our ancient copy of Error.pm\n + git-send-email: unconditionally use Net::{SMTP,Domain}\n + Git.pm: hard-depend on the File::{Temp,Spec} modules\n + gitweb: hard-depend on the Digest::MD5 5.8 module\n + Git.pm: add the \"use warnings\" pragma\n + Git.pm: remove redundant \"use strict\" from sub-package\n + perl: *.pm files should not have the executable bit\n\n Clean-up to various pieces of Perl code we have.\n\n Will merge to 'master'.\n\n\n* ds/commit-graph (2018-03-14) 17 commits\n - SQUASH??? sparse fixes\n - commit-graph: implement \"--additive\" option\n - commit-graph: build graph from starting commits\n - commit-graph: read only from specific pack-indexes\n - commit: integrate commit graph with commit parsing\n - commit-graph: close under reachability\n - commit-graph: add core.commitGraph setting\n - commit-graph: implement git commit-graph read\n - commit-graph: implement 'git-commit-graph write'\n - commit-graph: implement write_commit_graph()\n - commit-graph: create git-commit-graph builtin\n - graph: add commit graph design document\n - commit-graph: add format document\n - csum-file: refactor finalize_hashfile() method\n - csum-file: rename hashclose() to finalize_hashfile()\n - Merge branch 'jk/cached-commit-buffer' into HEAD\n - Merge branch 'jt/binsearch-with-fanout' into HEAD\n\n Precompute and store information necessary for ancestry traversal\n in a separate file to optimize graph walking.\n\n It seems that this topic is getting there.\n\n\n* ma/config-page-only-in-list-mode (2018-02-21) 3 commits\n - config: change default of `pager.config` to \"on\"\n - config: respect `pager.config` in list/get-mode only\n - t7006: add tests for how git config paginates\n\n In a way similar to how \"git tag\" learned to honor the pager\n setting only in the list mode, \"git config\" learned to ignore the\n pager setting when it is used for setting values (i.e. when the\n purpose of the operation is not to \"show\").\n\n Will merge to 'next'.\n\n\n* pc/submodule-helper-foreach (2018-02-02) 5 commits\n - submodule: port submodule subcommand 'foreach' from shell to C\n - submodule foreach: document variable '$displaypath'\n - submodule foreach: clarify the '$toplevel' variable documentation\n - submodule foreach: document '$sm_path' instead of '$path'\n - submodule foreach: correct '$path' in nested submodules from a subdirectory\n\n Expecting a response to review comments\n e.g. cf. <20180206150044.1bffbb573c088d38c8e44bf5@google.com>\n\n\n* tg/worktree-add-existing-branch (2018-02-05) 3 commits\n - worktree: teach \"add\" to check out existing branches\n - worktree: be clearer when \"add\" dwim-ery kicks in\n - worktree: improve message when creating a new worktree\n\n \"git worktree add\" learned to check out an existing branch.\n\n Expecting a reroll.\n cf. <CAPig+cRLohiqR_Drh7P0q3XbvC22WLjNwH0YLZo3dqFzZZuAPw@mail.gmail.com>\n cf. <CACsJy8BEKYqW+Ne_WY2RBaSbb9OKcjREtrawStj=eJsVsia_Jw@mail.gmail.com>\n The general idea is good, just end-user facing messages are found\n suboptimal.\n\n\n* np/send-email-header-parsing (2017-12-15) 1 commit\n  (merged to 'next' on 2018-03-09 at 91ef7216f7)\n + send-email: extract email-parsing code into a subroutine\n (this branch is used by cl/send-email-reply-to.)\n\n Code refactoring.\n\n Will merge to 'master'.\n\n\n* cl/send-email-reply-to (2018-03-06) 2 commits\n  (merged to 'next' on 2018-03-09 at 3d3c3ab441)\n + send-email: support separate Reply-To address\n + send-email: rename variable for clarity\n (this branch uses np/send-email-header-parsing.)\n\n \"git send-email\" learned \"--reply-to=<address>\" option.\n\n Will merge to 'master'.\n\n\n* js/rebase-recreate-merge (2018-02-23) 12 commits\n - rebase -i: introduce --recreate-merges=[no-]rebase-cousins\n - pull: accept --rebase=recreate to recreate the branch topology\n - sequencer: handle post-rewrite for merge commands\n - sequencer: make refs generated by the `label` command worktree-local\n - rebase: introduce the --recreate-merges option\n - rebase-helper --make-script: introduce a flag to recreate merges\n - sequencer: fast-forward merge commits, if possible\n - sequencer: introduce the `merge` command\n - sequencer: introduce new commands to reset the revision\n - git-rebase--interactive: clarify arguments\n - sequencer: make rearrange_squash() a bit more obvious\n - sequencer: avoid using errno clobbered by rollback_lock_file()\n\n \"git rebase\" learned \"--recreate-merges\" to transplant the whole\n topology of commit graph elsewhere.\n\n Will merge to 'next'.\n\n\n* bw/protocol-v2 (2018-03-14) 35 commits\n - remote-curl: don't request v2 when pushing\n - remote-curl: implement stateless-connect command\n - http: eliminate \"# service\" line when using protocol v2\n - http: don't always add Git-Protocol header\n - http: allow providing extra headers for http requests\n - remote-curl: store the protocol version the server responded with\n - remote-curl: create copy of the service name\n - pkt-line: add packet_buf_write_len function\n - transport-helper: introduce stateless-connect\n - transport-helper: refactor process_connect_service\n - transport-helper: remove name parameter\n - connect: don't request v2 when pushing\n - connect: refactor git_connect to only get the protocol version once\n - fetch-pack: support shallow requests\n - fetch-pack: perform a fetch using v2\n - upload-pack: introduce fetch server command\n - push: pass ref prefixes when pushing\n - fetch: pass ref prefixes when fetching\n - ls-remote: pass ref prefixes when requesting a remote's refs\n - transport: convert transport_get_remote_refs to take a list of ref prefixes\n - transport: convert get_refs_list to take a list of ref prefixes\n - connect: request remote refs using v2\n - ls-refs: introduce ls-refs server command\n - serve: introduce git-serve\n - test-pkt-line: introduce a packet-line test helper\n - protocol: introduce enum protocol_version value protocol_v2\n - transport: store protocol version\n - connect: discover protocol version outside of get_remote_heads\n - connect: convert get_remote_heads to use struct packet_reader\n - transport: use get_refs_via_connect to get refs\n - upload-pack: factor out processing lines\n - upload-pack: convert to a builtin\n - pkt-line: add delim packet support\n - pkt-line: allow peeking a packet line without consuming it\n - pkt-line: introduce packet_read_with_status\n\n The beginning of the next-gen transfer protocol.\n\n\n* ls/checkout-encoding (2018-03-09) 10 commits\n - convert: add round trip check based on 'core.checkRoundtripEncoding'\n - convert: add tracing for 'working-tree-encoding' attribute\n - convert: advise canonical UTF encoding names\n - convert: check for detectable errors in UTF encodings\n - convert: add 'working-tree-encoding' attribute\n - utf8: add function to detect a missing UTF-16/32 BOM\n - utf8: add function to detect prohibited UTF-16/32 BOM\n - strbuf: add a case insensitive starts_with()\n - strbuf: add xstrdup_toupper()\n - strbuf: remove unnecessary NUL assignment in xstrdup_tolower()\n\n The new \"checkout-encoding\" attribute can ask Git to convert the\n contents to the specified encoding when checking out to the working\n tree (and the other way around when checking in).\n\n Expecting a reroll.\n cf. <66370A41-A048-44E7-9BF8-4631C50AAE63@gmail.com>\n Modulo minor design decision corrections, the series is almost there.\n\n\n* en/rename-directory-detection (2018-02-27) 29 commits\n  (merged to 'next' on 2018-03-06 at d42470f86e)\n + merge-recursive: ensure we write updates for directory-renamed file\n + merge-recursive: avoid spurious rename/rename conflict from dir renames\n + directory rename detection: new testcases showcasing a pair of bugs\n + merge-recursive: fix remaining directory rename + dirty overwrite cases\n + merge-recursive: fix overwriting dirty files involved in renames\n + merge-recursive: avoid clobbering untracked files with directory renames\n + merge-recursive: apply necessary modifications for directory renames\n + merge-recursive: when comparing files, don't include trees\n + merge-recursive: check for file level conflicts then get new name\n + merge-recursive: add computation of collisions due to dir rename & merging\n + merge-recursive: check for directory level conflicts\n + merge-recursive: add get_directory_renames()\n + merge-recursive: make a helper function for cleanup for handle_renames\n + merge-recursive: split out code for determining diff_filepairs\n + merge-recursive: make !o->detect_rename codepath more obvious\n + merge-recursive: fix leaks of allocated renames and diff_filepairs\n + merge-recursive: introduce new functions to handle rename logic\n + merge-recursive: move the get_renames() function\n + directory rename detection: tests for handling overwriting dirty files\n + directory rename detection: tests for handling overwriting untracked files\n + directory rename detection: miscellaneous testcases to complete coverage\n + directory rename detection: testcases exploring possibly suboptimal merges\n + directory rename detection: more involved edge/corner testcases\n + directory rename detection: testcases checking which side did the rename\n + directory rename detection: files/directories in the way of some renames\n + directory rename detection: partially renamed directory testcase/discussion\n + directory rename detection: testcases to avoid taking detection too far\n + directory rename detection: directory splitting testcases\n + directory rename detection: basic testcases\n\n Rename detection logic in \"diff\" family that is used in \"merge\" has\n learned to guess when all of x/a, x/b and x/c have moved to z/a,\n z/b and z/c, it is likely that x/d added in the meantime would also\n want to move to z/d by taking the hint that the entire directory\n 'x' moved to 'z'.  A bug causing dirty files involved in a rename\n to be overwritten during merge has also been fixed as part of this\n work.\n\n Will merge to 'master'.\n\n--------------------------------------------------\n[Discarded]\n\n* cc/require-tcl-tk-for-build (2017-11-29) 2 commits\n - travis-ci: avoid new tcl/tk build requirement\n - Makefile: check that tcl/tk is installed\n\n A first-time builder of Git may have installed neither tclsh nor\n msgfmt, in which case git-gui and gitk part will fail and break the\n build.  As a workaround, refuse to run a build when tclsh is not\n installed and NO_TCLTK is not set.\n\n Stalled for too long without any response; will discard.\n I still feel that requring tclsh to be installed, with or without\n \"escape hatch\" for experts, may be too heavy-handed.\n\n\n* mg/merge-base-fork-point (2017-09-17) 3 commits\n - merge-base: find fork-point outside partial reflog\n - merge-base: return fork-point outside reflog\n - t6010: test actual test output\n\n \"merge-base --fork-point $branch $commit\" is used to guess on which\n commit among the commits that were once at the tip of the $branch the\n $commit was built on top of, and it learns these historical tips from\n the reflog of the $branch.  When the true fork-point is lost due to\n pruning of old reflog entries, the command does not give any output,\n because it has no way to guess correctly and does not want to mislead\n the user with a wrong guess.\n\n The command has been updated to give the best but not known to be\n correct guess, based on a hope that a merge-base between $commit and a\n virtual merge across all the reflog entries that still are available\n for $branch may still be a closer to the true fork-point than the\n merge-base between $commit and the current tip of the $branch.\n\n This may have to be offered by an additional option, to allow the\n users that are prepared to see a potentially incorrect guess to opt\n into the feature, without affecting the current callers that may not\n be prepared to accept a guess that is not known to be correct.\n\n Stalled for too long without any response; will discard.\n\n\n* mg/status-in-progress-info (2017-05-10) 2 commits\n - status --short --inprogress: spell it as --in-progress\n - status: show in-progress info for short status\n\n \"git status\" learns an option to report various operations\n (e.g. \"merging\") that the user is in the middle of.\n\n Stalled for too long without any response; will discard.\n cf. <xmqqmvakcdqw.fsf@gitster.mtv.corp.google.com>\n\n\n* nd/tilde-expand-opt-file-value (2018-02-14) 2 commits\n - init-db: change --template type to OPTION_FILENAME\n - parse-options: expand $HOME on filename options\n\n \"git cmd --opt=~u/path/to/file\" did not tilde-expand \"~u\" part to\n the path to the home directory of user 'u'\n\n This may make the resulting whole more confusing, though.\n cf. <87wozffavp.fsf@evledraar.gmail.com>\n\n\n* dp/git-el-ls-files-excludes (2018-03-05) 1 commit\n . git.el: handle default excludesfile properly\n\n The \"git.el\" script (in contrib/) has been taught to use\n \"--exclude-standard\" option when driving \"ls-files\", instead of\n locating the exclude files on its own.  With this change, it now\n honors $XDG_CONFIG_HOME when it is set.\n\n The author of \"git.el\" recommends that users consider switching to\n \"magit\", though.\n\n Discarded, as contrib/emacs/ hierarchy is being emptied by the\n ab/nuke-emacs-contrib topic.\n"},{"id":"341734","messageId":"CACsJy8D9-OaU5F+bugpgaPBPV_cdtqeYJ_5iYYMcrDj5d-9z5Q@mail.gmail.com","threadId":"48046","inReplyTo":"xmqqefkm6s06.fsf@gitster-ct.c.googlers.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Duy Nguyen","fromEmail":"pclouds@gmail.com","sentAt":"2018-03-15T06:30:53Z","receivedAt":"2018-03-15T06:31:30Z","isPatch":false,"sender":{"key":"pclouds@gmail.com","avatar":"https://avatars.githubusercontent.com/u/720?v=4"},"body":"On Thu, Mar 15, 2018 at 2:34 AM, Junio C Hamano <gitster@pobox.com> wrote:\n> * nd/pack-objects-pack-struct (2018-03-05) 9 commits\n>  - pack-objects: reorder 'hash' to pack struct object_entry\n>  - pack-objects: refer to delta objects by index instead of pointer\n>  - pack-objects: move in_pack out of struct object_entry\n>  - pack-objects: move in_pack_pos out of struct object_entry\n>  - pack-objects: note about in_pack_header_size\n>  - pack-objects: use bitfield for object_entry::depth\n>  - pack-objects: use bitfield for object_entry::dfs_state\n>  - pack-objects: turn type and in_pack_type to bitfields\n>  - pack-objects: document holes in struct object_entry.h\n>\n>  \"git pack-objects\" needs to allocate tons of \"struct object_entry\"\n>  while doing its work, and shrinking its size helps the performance\n>  quite a bit.\n>\n>  Will merge to 'next'.\n\nHold it. A reroll is coming. I'm a bit busy this week and can't really do much.\n\n> * nd/worktree-prune (2018-03-06) 3 commits\n>  - worktree prune: improve prune logic when worktree is moved\n>  - worktree: delete dead code\n>  - gc.txt: more details about what gc does\n>\n>  The way \"git worktree prune\" worked internally has been simplified,\n>  by assuming how \"git worktree move\" moves an existing worktree to a\n>  different place.\n>\n>  Will merge to 'next'.\n\nSame.\n-- \nDuy\n"},{"id":"341736","messageId":"87r2olenw9.fsf@evledraar.gmail.com","threadId":"48046","inReplyTo":"xmqqefkm6s06.fsf@gitster-ct.c.googlers.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Ævar Arnfjörð Bjarmason","fromEmail":"avarab@gmail.com","sentAt":"2018-03-15T08:36:22Z","receivedAt":"2018-03-15T08:36:33Z","isPatch":false,"sender":{"key":"avarab@gmail.com","avatar":"https://avatars.githubusercontent.com/u/45301?v=4"},"body":"\nOn Thu, Mar 15 2018, Junio C. Hamano jotted:\n\n> * ti/fetch-everything-local-optim (2018-03-14) 1 commit\n>  - fetch-pack.c: use oidset to check existence of loose object\n\nLooks good to me, and great to have such an optimization in.\n\n> * ab/pcre-v2 (2018-03-14) 3 commits\n>\n>  Will merge to 'next'.\n>\n>[...]\n>\n> * ab/perl-fixes (2018-03-05) 13 commits\n>\n>  Will merge to 'master'.\n\nGiven 2.17-rc1 next Tuesday according to your calendar, what's the\nstatus of these two landing in 2.17?\n\nI'd like to annoy packagers with all this perl/ stuff in just one\nrelease instead of spreading out out over two, and without ab/perl-fixes\nthey'll need another hack to avoid installing our Error.pm.\n\nThe ab/pcre-v2 is less important, but given the minimal nature of it\nwould be very nice to have in 2.17 as well so we get off the\nmostly-unmaintained v1 sooner than later.\n\n> * nd/repack-keep-pack (2018-03-07) 6 commits\n>  - SQUASH???\n>  - pack-objects: display progress in get_object_details()\n>  - pack-objects: show some progress when counting kept objects\n>  - gc --auto: exclude base pack if not enough mem to \"repack -ad\"\n>  - repack: add --keep-pack option\n>  - t7700: have closing quote of a test at the beginning of line\n>\n>  \"git gc\" in a large repository takes a lot of time as it considers\n>  to repack all objects into one pack by default.  The command has\n>  been taught to pretend as if the largest existing packfile is\n>  marked with \".keep\" so that it is left untouched while objects in\n>  other packs and loose ones are repacked.\n>\n>  Expecting a reroll.\n>  cf. <CACsJy8BW_EtxQvgL=YrCXCQY7cEWCQxgfkeH=Gd=X=uVYhPJcw@mail.gmail.com>\n>  Except for final finishing touches, this looked more-or-less ready\n>  for 'next'.\n>\n>\n\nAs I noted in 87a7vdqegi.fsf@evledraar.gmail.com and\n877eqhq7ha.fsf@evledraar.gmail.com (both at:\nhttps://public-inbox.org/git/?q=87a7vdqegi.fsf%40evledraar.gmail.com) I\nthink we should change the too-specific behavior here to be more generic\n(and am happy to do the work pending feedback from Duy on what he thinks\nabout it).\n\nI'm also interested to know from those at Microsoft (CC'd some) if the\nmechanism I've proposed is something closer to what they could\neventually use to gc windows.git.\n\nI know that now it doesn't GC now, and they have some side-channel\nmechanism for pre-deploying large (daily?) packs to clients, if it's\nadjusted as I suggest gc could be told not to touch packs of that size,\nleaving only stray small packs from \"git pull\" and loose objects to GC.\n\nI may also have entirely misunderstood how it works, this is from brief\nin-person conversations at Git Merge.\n\nBut as far as mainlining some of that eventually I think it would be\ngood to get feedback on whether the mechanism I proposed would get in\ntheir way more or less than what Duy has, or be entirely irrelevant\nbecause they need something else.\n\nThanks!\n"},{"id":"341765","messageId":"xmqqwoyd5lf3.fsf@gitster-ct.c.googlers.com","threadId":"48046","inReplyTo":"CACsJy8D9-OaU5F+bugpgaPBPV_cdtqeYJ_5iYYMcrDj5d-9z5Q@mail.gmail.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2018-03-15T16:54:40Z","receivedAt":"2018-03-15T16:54:48Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Duy Nguyen <pclouds@gmail.com> writes:\n\n> On Thu, Mar 15, 2018 at 2:34 AM, Junio C Hamano <gitster@pobox.com> wrote:\n>> * nd/pack-objects-pack-struct (2018-03-05) 9 commits\n>> ...\n>>  Will merge to 'next'.\n>\n> Hold it. A reroll is coming. I'm a bit busy this week and can't really do much.\n>\n>> * nd/worktree-prune (2018-03-06) 3 commits\n>> ...\n> Same.\n\nThanks, quick responses like this message really helps.\n"},{"id":"341802","messageId":"xmqqk1ud5jmo.fsf@gitster-ct.c.googlers.com","threadId":"48046","inReplyTo":"87r2olenw9.fsf@evledraar.gmail.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2018-03-15T17:33:19Z","receivedAt":"2018-03-15T17:33:29Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:\n\n> Given 2.17-rc1 next Tuesday according to your calendar, what's the\n> status of these two landing in 2.17?\n\nA rule of thumb for an average topic with just a handful of commits\nis to stay in 'next' for about a week before graduating.  A larger\ntopic with wider implications to the codebase may want to be a few\nweeks in 'next' (or stay there since before -rc0 and during the\nfeature freeze to the final), while a more trivially correct ones\nmay be there only overnight.\n\nI tend to think that the perl-fixes topic is something we'd rather\nwant to push packgers to try sooner rather than later; there aren't\nroom for \"bugs\" that affects program execution, but the impact is\nmore to the installation procedure people have around our build\nprocedure.\n\nI do not have a firm opinion on pcre v1 vs v2; the patches are only\nabout flipping the default and not robbing any capabilities from the\npackagers and the users, so I do not see any need to be unusually\ncareful to merge it down (compared to other topics).\n"},{"id":"341824","messageId":"7D05D553-D914-402F-8AE1-C49C1E1D64AB@gmail.com","threadId":"48046","inReplyTo":"xmqqefkm6s06.fsf@gitster-ct.c.googlers.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Lars Schneider","fromEmail":"larsxschneider@gmail.com","sentAt":"2018-03-15T19:18:02Z","receivedAt":"2018-03-15T19:18:11Z","isPatch":false,"sender":{"key":"larsxschneider@gmail.com","avatar":"https://avatars.githubusercontent.com/u/477434?v=4"},"body":"\n> On 15 Mar 2018, at 02:34, Junio C Hamano <gitster@pobox.com> wrote:\n> \n> ...\n> \n> * ls/checkout-encoding (2018-03-09) 10 commits\n> - convert: add round trip check based on 'core.checkRoundtripEncoding'\n> - convert: add tracing for 'working-tree-encoding' attribute\n> - convert: advise canonical UTF encoding names\n> - convert: check for detectable errors in UTF encodings\n> - convert: add 'working-tree-encoding' attribute\n> - utf8: add function to detect a missing UTF-16/32 BOM\n> - utf8: add function to detect prohibited UTF-16/32 BOM\n> - strbuf: add a case insensitive starts_with()\n> - strbuf: add xstrdup_toupper()\n> - strbuf: remove unnecessary NUL assignment in xstrdup_tolower()\n> \n> The new \"checkout-encoding\" attribute can ask Git to convert the\n> contents to the specified encoding when checking out to the working\n> tree (and the other way around when checking in).\n> \n> Expecting a reroll.\n> cf. <66370A41-A048-44E7-9BF8-4631C50AAE63@gmail.com>\n> Modulo minor design decision corrections, the series is almost there.\n\nIf I hurry up (IOW: send a reroll tonight), would this topic\nhave a chance for 2.17-rc1 or is it too late?\n\nThanks,\nLars\n\n\n"},{"id":"341851","messageId":"13DCA899-54E1-49C9-A90D-E0A24C9F7D10@gmail.com","threadId":"48046","inReplyTo":"7D05D553-D914-402F-8AE1-C49C1E1D64AB@gmail.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Lars Schneider","fromEmail":"larsxschneider@gmail.com","sentAt":"2018-03-15T23:00:45Z","receivedAt":"2018-03-15T23:00:54Z","isPatch":false,"sender":{"key":"larsxschneider@gmail.com","avatar":"https://avatars.githubusercontent.com/u/477434?v=4"},"body":"\n> On 15 Mar 2018, at 20:18, Lars Schneider <larsxschneider@gmail.com> wrote:\n> \n> \n>> On 15 Mar 2018, at 02:34, Junio C Hamano <gitster@pobox.com> wrote:\n>> \n>> ...\n>> \n>> * ls/checkout-encoding (2018-03-09) 10 commits\n>> - convert: add round trip check based on 'core.checkRoundtripEncoding'\n>> - convert: add tracing for 'working-tree-encoding' attribute\n>> - convert: advise canonical UTF encoding names\n>> - convert: check for detectable errors in UTF encodings\n>> - convert: add 'working-tree-encoding' attribute\n>> - utf8: add function to detect a missing UTF-16/32 BOM\n>> - utf8: add function to detect prohibited UTF-16/32 BOM\n>> - strbuf: add a case insensitive starts_with()\n>> - strbuf: add xstrdup_toupper()\n>> - strbuf: remove unnecessary NUL assignment in xstrdup_tolower()\n>> \n>> The new \"checkout-encoding\" attribute can ask Git to convert the\n>> contents to the specified encoding when checking out to the working\n>> tree (and the other way around when checking in).\n>> \n>> Expecting a reroll.\n>> cf. <66370A41-A048-44E7-9BF8-4631C50AAE63@gmail.com>\n>> Modulo minor design decision corrections, the series is almost there.\n> \n> If I hurry up (IOW: send a reroll tonight), would this topic\n> have a chance for 2.17-rc1 or is it too late?\n\nI just send out a reroll in case the series is still relevant\nfor 2.17-rc1.\n\nhttps://public-inbox.org/git/20180315225746.18119-1-lars.schneider@autodesk.com/\n\nThanks,\nLars"},{"id":"341857","messageId":"xmqq1sgk4z7h.fsf@gitster-ct.c.googlers.com","threadId":"48046","inReplyTo":"7D05D553-D914-402F-8AE1-C49C1E1D64AB@gmail.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2018-03-16T00:54:26Z","receivedAt":"2018-03-16T00:54:57Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Lars Schneider <larsxschneider@gmail.com> writes:\n\n> If I hurry up (IOW: send a reroll tonight), would this topic\n> have a chance for 2.17-rc1 or is it too late?\n\nThat depends on how well the reroll is made, but for an\naverage-sized and average-importance topic like this one that is not\nyet in 'next' yet, even if it is very close, I'd have to say that it\nis cutting a bit too close.\n\n\n"},{"id":"341950","messageId":"20180316143124.5b22d70861344cbcbbad46dd@google.com","threadId":"48046","inReplyTo":"xmqqefkm6s06.fsf@gitster-ct.c.googlers.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Jonathan Tan","fromEmail":"jonathantanmy@google.com","sentAt":"2018-03-16T21:31:24Z","receivedAt":"2018-03-16T21:31:31Z","isPatch":false,"sender":{"key":"jonathantanmy@fastmail.com","avatar":null},"body":"On Wed, 14 Mar 2018 18:34:49 -0700\nJunio C Hamano <gitster@pobox.com> wrote:\n\n> * sb/object-store (2018-03-05) 27 commits\n\n[snip list of commits]\n\n>  (this branch is used by sb/packfiles-in-repository; uses nd/remove-ignore-env-field.)\n> \n>  Refactoring the internal global data structure to make it possible\n>  to open multiple repositories, work with and then close them.\n> \n>  Rerolled by Duy on top of a separate preliminary clean-up topic.\n>  The resulting structure of the topics looked very sensible.\n> \n>  Waiting for a follow-up discussion.\n\nWould it be possible for this set to go in independently of\nnd/remove-ignore-env-field? I understand that some patches might be\ncleaner if ignore_env is first removed, but this patch set has already\nundergone several rounds of review and (I think) is an improvement to\nthe codebase on its own.\n"},{"id":"341954","messageId":"xmqqh8pfzo1h.fsf@gitster-ct.c.googlers.com","threadId":"48046","inReplyTo":"20180316143124.5b22d70861344cbcbbad46dd@google.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2018-03-16T21:52:10Z","receivedAt":"2018-03-16T21:52:17Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Jonathan Tan <jonathantanmy@google.com> writes:\n\n> On Wed, 14 Mar 2018 18:34:49 -0700\n> Junio C Hamano <gitster@pobox.com> wrote:\n>\n>> * sb/object-store (2018-03-05) 27 commits\n>\n> [snip list of commits]\n>\n>>  (this branch is used by sb/packfiles-in-repository; uses nd/remove-ignore-env-field.)\n>> \n>>  Refactoring the internal global data structure to make it possible\n>>  to open multiple repositories, work with and then close them.\n>> \n>>  Rerolled by Duy on top of a separate preliminary clean-up topic.\n>>  The resulting structure of the topics looked very sensible.\n>> \n>>  Waiting for a follow-up discussion.\n>\n> Would it be possible for this set to go in independently of\n> nd/remove-ignore-env-field? I understand that some patches might be\n> cleaner if ignore_env is first removed, but this patch set has already\n> undergone several rounds of review and (I think) is an improvement to\n> the codebase on its own.\n\nI thought the \"remove-ignore-env-field\" thing is a quite small and\nmore-or-less straightforward improvements that would serve as a good\npreparatory change to give a solid foundation to the object-store\ntopic.\n\nI was hoping to hear quick Acks for remove-ignore-env (and also\nDuy's reroll of their topics on it) from people involved in all the\nrelated topics, so that we can advance them more-or-less at the same\ntime.\n\n\n"},{"id":"341956","messageId":"20180316220824.GC151588@aiede.svl.corp.google.com","threadId":"48046","inReplyTo":"xmqqh8pfzo1h.fsf@gitster-ct.c.googlers.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Jonathan Nieder","fromEmail":"jrnieder@gmail.com","sentAt":"2018-03-16T22:08:24Z","receivedAt":"2018-03-16T22:08:32Z","isPatch":false,"sender":{"key":"jrnieder@gmail.com","avatar":"https://avatars.githubusercontent.com/u/281595?v=4"},"body":"Hi,\n\nJunio C Hamano wrote:\n> Jonathan Tan <jonathantanmy@google.com> writes:\n>> On Wed, 14 Mar 2018 18:34:49 -0700\n>> Junio C Hamano <gitster@pobox.com> wrote:\n\n>>> * sb/object-store (2018-03-05) 27 commits\n>>\n>> [snip list of commits]\n>>\n>>>  (this branch is used by sb/packfiles-in-repository; uses nd/remove-ignore-env-field.)\n>>>\n>>>  Refactoring the internal global data structure to make it possible\n>>>  to open multiple repositories, work with and then close them.\n>>>\n>>>  Rerolled by Duy on top of a separate preliminary clean-up topic.\n>>>  The resulting structure of the topics looked very sensible.\n>>>\n>>>  Waiting for a follow-up discussion.\n>>\n>> Would it be possible for this set to go in independently of\n>> nd/remove-ignore-env-field? I understand that some patches might be\n>> cleaner if ignore_env is first removed, but this patch set has already\n>> undergone several rounds of review and (I think) is an improvement to\n>> the codebase on its own.\n>\n> I thought the \"remove-ignore-env-field\" thing is a quite small and\n> more-or-less straightforward improvements that would serve as a good\n> preparatory change to give a solid foundation to the object-store\n> topic.\n>\n> I was hoping to hear quick Acks for remove-ignore-env (and also\n> Duy's reroll of their topics on it) from people involved in all the\n> related topics, so that we can advance them more-or-less at the same\n> time.\n\nI can go along with this for this series, but I want to explain why I\nam not thrilled with the process in general.\n\nThe series \"Moving global state into the repository object (part 1)\"\nhad gone through three revisions with relatively minor changes between\neach and we were in the middle of reviewing the fourth.  In response\nto a series that comes after it, \"(part 2)\", Duy created\nnd/remove-ignore-env, a series that I quite like.\n\nBut with this reroll, the result is:\n\n * The patches that have already been reviewed 3 times (in their\n   current incarnation; parts of this series have also appeared on\n   list earlier too) and we were in the middle of reviewing a fourth\n   time are now on a new base.\n\n * These three series, each of a manageable size, have been combined\n   into a larger series of 44 patches.\n\n * I have fears about when they're ever going to land and be part of\n   an API I can start making use of.\n\nSo even though I agree with the goal and I like the improved\ninitialization code, I am not happy.\n\nI would be happier with just the new initialization code being its own\nseries that we can fast-track, then deal with part 1 separately on\ntop, and then deal with part 2 after that.\n\nThanks,\nJonathan\n"},{"id":"341966","messageId":"xmqq1sgjzfn0.fsf@gitster-ct.c.googlers.com","threadId":"48046","inReplyTo":"20180316220824.GC151588@aiede.svl.corp.google.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Junio C Hamano","fromEmail":"gitster@pobox.com","sentAt":"2018-03-17T00:53:39Z","receivedAt":"2018-03-17T00:53:50Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"Jonathan Nieder <jrnieder@gmail.com> writes:\n\n>> I was hoping to hear quick Acks for remove-ignore-env (and also\n>> Duy's reroll of their topics on it) from people involved in all the\n>> related topics, so that we can advance them more-or-less at the same\n>> time.\n>\n> I can go along with this for this series, but I want to explain why I\n> am not thrilled with the process in general.\n>\n> The series \"Moving global state into the repository object (part 1)\"\n> had gone through three revisions with relatively minor changes between\n> each and we were in the middle of reviewing the fourth.  In response\n> to a series that comes after it, \"(part 2)\", Duy created\n> nd/remove-ignore-env, a series that I quite like.\n>\n> But with this reroll, the result is:\n> ...\n> So even though I agree with the goal and I like the improved\n> initialization code, I am not happy.\n\nWell, I do not think it was the \"process\" issue in particular,\nthough.  Yes, it was unfortunate that these three revisions were\nreviewed but they each resulted with only relative minor changes\nbetween iterations.  IOW, reviewers that lost fresh eyes, perhaps\nincluding me, all missed a larger wart in the design that made the\n\"ignore-env\" thing required in the structure, which was spotted by\nsomebody else.  It would have been nicer if such an improvement were\ndone during an earlier part of the cycle and did not require that\nother reviewer.\n\nBut we didn't spot it, and now we know a better structure of the\ntopic, luckily before all the things touch 'next'.  We not just got\na suggestion \"it would be better to rewrite it like so---now go\nahead to do that\", we got something even better and Duy did all the\nwork restructuring the series, reshuffling the patches.  We should\nbe thankful, not unhappy.\n\n"},{"id":"342253","messageId":"94484ca0-6331-90a7-41cb-dd44200d8d20@gmail.com","threadId":"48046","inReplyTo":"87r2olenw9.fsf@evledraar.gmail.com","subject":"Re: What's cooking in git.git (Mar 2018, #03; Wed, 14)","fromName":"Derrick Stolee","fromEmail":"stolee@gmail.com","sentAt":"2018-03-19T21:16:09Z","receivedAt":"2018-03-19T21:16:24Z","isPatch":false,"sender":{"key":"stolee@gmail.com","avatar":"https://avatars.githubusercontent.com/u/570044?v=4"},"body":"On 3/15/2018 4:36 AM, Ævar Arnfjörð Bjarmason wrote:\n> On Thu, Mar 15 2018, Junio C. Hamano jotted:\n>\n>> * nd/repack-keep-pack (2018-03-07) 6 commits\n>>   - SQUASH???\n>>   - pack-objects: display progress in get_object_details()\n>>   - pack-objects: show some progress when counting kept objects\n>>   - gc --auto: exclude base pack if not enough mem to \"repack -ad\"\n>>   - repack: add --keep-pack option\n>>   - t7700: have closing quote of a test at the beginning of line\n>>\n>>   \"git gc\" in a large repository takes a lot of time as it considers\n>>   to repack all objects into one pack by default.  The command has\n>>   been taught to pretend as if the largest existing packfile is\n>>   marked with \".keep\" so that it is left untouched while objects in\n>>   other packs and loose ones are repacked.\n>>\n>>   Expecting a reroll.\n>>   cf. <CACsJy8BW_EtxQvgL=YrCXCQY7cEWCQxgfkeH=Gd=X=uVYhPJcw@mail.gmail.com>\n>>   Except for final finishing touches, this looked more-or-less ready\n>>   for 'next'.\n>>\n>>\n> As I noted in 87a7vdqegi.fsf@evledraar.gmail.com and\n> 877eqhq7ha.fsf@evledraar.gmail.com (both at:\n> https://public-inbox.org/git/?q=87a7vdqegi.fsf%40evledraar.gmail.com) I\n> think we should change the too-specific behavior here to be more generic\n> (and am happy to do the work pending feedback from Duy on what he thinks\n> about it).\n>\n> I'm also interested to know from those at Microsoft (CC'd some) if the\n> mechanism I've proposed is something closer to what they could\n> eventually use to gc windows.git.\n\nSorry that I couldn't get to this message sooner, I was traveling. While \nI was gone, the others who you CC'd volunteered me as the best person to \nrespond ;)\n\nIn the interest of full disclosure and hopefully starting an interesting \ndiscussion, I want to share as much detail as possible as well as a few \nfuture directions that can inform our actions now. Here are some rough \nideas that we are thinking about in this space:\n\n   I. Use the multi-pack index (MIDX) [1] to track \"packfile state\" so \nwe can do GC/repack incrementally.\n  II. Replace our prefetch packs model with partial clone [2].\nIII. Stop including all trees and focus the fetch down a cone of the \nworking directory.\n  IV. Provide a way to defer certain read-only commands to a remote when \nthe local repo doesn't have sufficient data.\n\n> I know that now it doesn't GC now, and they have some side-channel\n> mechanism for pre-deploying large (daily?) packs to clients, if it's\n> adjusted as I suggest gc could be told not to touch packs of that size,\n> leaving only stray small packs from \"git pull\" and loose objects to GC.\n>\n> I may also have entirely misunderstood how it works, this is from brief\n> in-person conversations at Git Merge.\n>\n> But as far as mainlining some of that eventually I think it would be\n> good to get feedback on whether the mechanism I proposed would get in\n> their way more or less than what Duy has, or be entirely irrelevant\n> because they need something else.\n>\n> Thanks!\n\nThe GVFS cache servers pre-compute daily packfiles filled with every \ncommit and tree introduced that day. When a client calls 'git fetch' the \nGVFS hook runs a \"prefetch\" command to get these daily packs from the \ncache servers and place them in an alternate we call the \"shared object \ncache\". GVFS also disables the \"receive-pack\" portion of the fetch. The \nMIDX is updated to cover the new packs.\n\nSomething we are going to enable soon is the addition of \"hourly packs\" \nwhere the cache servers keep a list of up to 24 hourly packs and on \nprefetch the client receives an extra pack that concatenates that day's \npack up to that point. We are doing this because the refs they are \nfetching are far enough ahead of the daily snapshots, which triggers a \ndecent amount of loose object downloads when those refs are checked out. \nSince the next day will create a new daily pack, we can delete that \nhourly pack (I don't think we do this currently). At the very least the \nMIDX prefers the duplicates in the new packfile.\n\nWe do not GC in GVFS-enabled repos. This doesn't destroy clients' disks \nso far because the prefetch packs do not contain blobs -- blobs are a \nlarge portion of the full repo size. As the repo grows, we are \nre-examining how to make GC and repack work in this environment. An \nimportant part of any solution will be to make it incremental: we cannot \nafford to create a second copy of the repo and we can't take the repo \noffline for a significant portion of time.\n\nI'm not sure that we can use Duy's solution out-of-the-box, in \nparticular because we haven't integrated our prefetch packs with the \n\"promisor\" concept from partial clone. Hence, GC will fail with missing \nobjects. I think the core idea is good: perform GC incrementally and \ndon't touch previously-GC'd packs. Since it is rare for an object to be \nreachable for a long time and then become unreachable, performing GC and \nthen keeping the resulting pack forever is a good idea. Especially when \npaired with the MIDX so you care less about the number of packfiles.\n\nI.\n\nOne of our thoughts is to use the MIDX to mark the packfiles with \nmetadata. For instance, we could mark new packfiles as \"raw\" and after \nenough time we could run a GC on just one packfile (or a batch of \npackfiles) to remove unreachable objects and mark it as \"clean\". In our \nsituation, we don't want to immediately GC the daily prefetch pack \nbecause a large portion of those objects will be reachable in a few days \ndue to branch integrations.\n\nII.\n\nOur long-term plan is to remove these GVFS-specific fetch patterns and \nreplace them with partial clone logic. We still see the cache servers as \nan important part of this process, but they are essentially read-only \ncopies of the main server object database (no refs). We haven't done the \nwork to see how expensive it will be to replace the precomputed prefetch \npacks with a partial clone negotiation. Then we will need to see how to \nmake the shared object cache be transparent to the user.\n\nAn aside: it would be good to investigate how Git could provide \nreplication like a cache server with minimal setup, including client \nconfiguration. The GVFS protocol has a \"gvfs/config\" endpoint that can \nprovide a list of cache server URLs, including a default. (For the \nWindows repo, these URLs point to load balancers that pass-through to an \narray of machines. I imagine most users will only need one server per \nlocation.) Does Git already have a way to redirect the call to upload-pack?\n\nIII.\n\nPartial clone has options for filtering trees by paths [3]. We want to \ntake advantage of this functionality to reduce the number of objects \nstored on disk. In a perfect world, we would use the virtualization \nlayer in GVFS to auto-detect the paths that are important to the user \nand create a sparse filter for them that way. Partial clone doesn't have \na mechanism for that right now (we require a sparse-checkout \nspecification to be present on the remote) but perhaps it could be part \nof a verb in protocol v2. We are working to collect sparse-checkout \nfiles from our users to see if it is possible to cluster their usage \npatterns into a reasonably small list of sparse-checkout specifications. \nAlternatively, how large would it be to dynamically send a list of paths \nbased on the current virtual projection? In either case, we expect some \nextra trees by including some amount of sibling paths as we expect users \nto be interested in reading nearby files or inspecting history for those \npaths.\n\nIV.\n\nOne big reason we started computing daily prefetch packs for GVFS is \nthat certain important Git commands became too slow if all objects were \nobtained from the loose-object download. Simple commands like \"git log\" \nwould take forever. With all commits and trees, we can run \"git log -- \npath\" at the same speed as a vanilla repo. \"git diff\" can still be slow \nbecause it needs blob contents, but hopefully the diff was for a single \ncommit and there are not too many calls. \"git grep\" could be a disaster.\n\nWhen we start restricting to sparse-checkout specifications, not all \n\"git log -- deep/paths/in/particular\" commands will have enough data \nlocally to run successfully.\n\nOne approach is to have a way to pass-through to a remote. Certain \ncommands could be requested (via a protocol v2 verb) to the server. The \nserver can compute the answer and send a (paged) response faster than \nthe client can download objects and then compute locally. This also \nprevents users from getting a bloated object database full of loose \nblobs they don't need.\n\nAt the very least, we will want to detect that we are missing many \nobjects and ask the user something like \"Did you really want to do that? \nIt will cause 1000+ objects to be downloaded. [Y/n]\"\n\n\nI hope this combo of information and half-baked ideas is helpful. We are \ndefinitely approaching a lot of interesting scale limits with many \ndifferent possible solutions.\n\nThanks,\n-Stolee\n\n[1] \nhttps://public-inbox.org/git/20180107181459.222909-1-dstolee@microsoft.com/T/#u\n     [RFC PATCH 00/18] Multi-pack index (MIDX)\n\n[2] \nhttps://public-inbox.org/git/20171214152404.35708-1-git@jeffhostetler.com/T/#u\n     [PATCH v2] Partial clone design document\n\n[3] \nhttps://github.com/git/git/blob/0afbf6caa5b16dcfa3074982e5b48e27d452dbbb/Documentation/rev-list-options.txt#L727-L733\n     rev-list-options.txt Documentation for \"--filter=sparse:*\"\n"}]}