What's cooking in git.git (May 2016, #04; Wed, 11)

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

What's cooking in git.git (May 2016, #04; Wed, 11)

Junio C Hamano
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.

The 'master' branch now has the eleventh batch of topics of this
cycle.  On the 'maint' front, 2.8.2 is out and fixes that have been
in 'master' accumulates on it for 2.8.3.

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"]

* ew/doc-split-pack-disables-bitmap (2016-04-28) 1 commit
  (merged to 'next' on 2016-05-06 at 6343d1e)
 + pack-objects: warn on split packs disabling bitmaps

 Doc update.


* ew/normal-to-e (2016-05-02) 1 commit
  (merged to 'next' on 2016-05-06 at 65a2c52)
 + .mailmap: update to my shorter email address


* js/close-packs-before-gc (2016-05-02) 1 commit
  (merged to 'next' on 2016-05-06 at bfd39bf)
 + t5510: run auto-gc in the foreground


* ls/p4-lfs (2016-04-28) 3 commits
  (merged to 'next' on 2016-05-06 at 3e1354d)
 + git-p4: fix Git LFS pointer parsing
 + travis-ci: express Linux/OS X dependency versions more clearly
 + travis-ci: update Git-LFS and P4 to the latest version

 Recent update to Git LFS broke "git p4" by changing the output from
 its "lfs pointer" subcommand.


* ls/travis-submitting-patches (2016-05-02) 1 commit
  (merged to 'next' on 2016-05-06 at 467930e)
 + Documentation: add setup instructions for Travis CI


* rn/glossary-typofix (2016-05-02) 1 commit
  (merged to 'next' on 2016-05-06 at 1e73e76)
 + Documentation: fix typo 'In such these cases'


* sb/clean-test-fix (2016-05-03) 1 commit
  (merged to 'next' on 2016-05-06 at d80c9c6)
 + t7300: mark test with SANITY


* sb/misc-cleanups (2016-04-28) 2 commits
  (merged to 'next' on 2016-05-06 at 87bc8a5)
 + submodule-config: don't shadow `cache`
 + config.c: drop local variable


* sk/gitweb-highlight-encoding (2016-05-03) 1 commit
  (merged to 'next' on 2016-05-06 at 441302c)
 + gitweb: apply fallback encoding before highlight

 Some multi-byte encoding can have a backslash byte as a later part
 of one letter, which would confuse "highlight" filter used in
 gitweb.

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

* es/t1500-modernize (2016-05-10) 7 commits
 - t1500: finish preparation upfront
 - t1500: be considerate to future potential tests
 - t1500: avoid setting environment variables outside of tests
 - t1500: avoid setting configuration options outside of tests
 - t1500: avoid changing working directory outside of tests
 - t1500: reduce dependence upon global state
 - t1500: test_rev_parse: facilitate future test enhancements

 test updates to make it more readable and maintainable.

 Will be rerolled.


* ls/travis-build-doc (2016-05-10) 1 commit
  (merged to 'next' on 2016-05-10 at 7f63497)
 + travis-ci: build documentation

 CI test was taught to build documentation pages.

 Will merge to 'master'.


* js/t3404-typofix (2016-05-10) 1 commit
  (merged to 'next' on 2016-05-10 at cbeabc0)
 + t3404: fix typo

 Will merge to 'master'.


* jk/rebase-interative-eval-fix (2016-05-10) 1 commit
  (merged to 'next' on 2016-05-11 at 4fdf387)
 + rebase--interactive: avoid empty list in shell for-loop

 Portability enhancement for "rebase -i" to help platforms whose
 shell does not like "for i in <empty>" (which is not POSIX-kosher).

 Will merge to 'master'.


* jk/test-send-sh-x-trace-elsewhere (2016-05-11) 1 commit
  (merged to 'next' on 2016-05-11 at 273a137)
 + test-lib: set BASH_XTRACEFD automatically

 Running tests with '-x' option to trace the individual command
 executions is a useful way to debug test scripts, but some tests
 that capture the standard error stream and check what the command
 said can be broken with the trace output mixed in.  When running
 our tests under "bash", however, we can redirect the trace output
 to another file descriptor to keep the standard error of programs
 being tested intact.

 Will merge to 'master'.


* js/perf-rebase-i (2016-05-11) 3 commits
 - Add a perf test for rebase -i
 - perf: make the tests work in worktrees
 - perf: let's disable symlinks when they are not available

 Add perf test for "rebase -i"


* nd/worktree-cleanup-post-head-protection (2016-05-10) 7 commits
 - worktree: simplify prefixing paths
 - worktree: avoid 0{40}, too many zeroes, hard to read
 - worktree.c: add clear_worktree()
 - worktree.c: use is_dot_or_dotdot()
 - git-worktree.txt: keep subcommand listing in alphabetical order
 - worktree.c: rewrite mark_current_worktree() to avoid strbuf
 - completion: support git-worktree
 (this branch uses nd/worktree-various-heads.)


* va/mailinfo-doc-typofix (2016-05-11) 1 commit
  (merged to 'next' on 2016-05-11 at 7180176)
 + Documentation/git-mailinfo: fix typo

 Typofix.

 Will merge to 'master'.

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

* ep/http-curl-trace (2016-05-02) 2 commits
 . imap-send.c: introduce the GIT_TRACE_CURL environment variable
 . http.c: implement the GIT_TRACE_CURL environment variable

 HTTP transport gained an option to produce more detailed debugging
 trace.

 Expecting a reroll.
 ($gmane/292074, 293236)


* sb/bisect (2016-04-15) 22 commits
 - SQUASH???
 - bisect: get back halfway shortcut
 - bisect: compute best bisection in compute_relevant_weights()
 - bisect: use a bottom-up traversal to find relevant weights
 - bisect: prepare for different algorithms based on find_all
 - bisect: rename count_distance() to compute_weight()
 - bisect: make total number of commits global
 - bisect: introduce distance_direction()
 - bisect: extract get_distance() function from code duplication
 - bisect: use commit instead of commit list as arguments when appropriate
 - bisect: replace clear_distance() by unique markers
 - bisect: use struct node_data array instead of int array
 - bisect: get rid of recursion in count_distance()
 - bisect: make algorithm behavior independent of DEBUG_BISECT
 - bisect: make bisect compile if DEBUG_BISECT is set
 - bisect: plug the biggest memory leak
 - bisect: add test for the bisect algorithm
 - t6030: generalize test to not rely on current implementation
 - t: use test_cmp_rev() where appropriate
 - t/test-lib-functions.sh: generalize test_cmp_rev
 - bisect: allow 'bisect run' if no good commit is known
 - bisect: write about `bisect next` in documentation

 The internal algorithm used in "git bisect" to find the next commit
 to check has been optimized greatly.

 Expecting a reroll.
 ($gmane/291163)


* ar/diff-args-osx-precompose (2016-04-05) 1 commit
 - diff: run arguments through precompose_argv

 Many commands normalize command line arguments from NFD to NFC
 variant of UTF-8 on OSX, but commands in the "diff" family did
 not, causing "git diff $path" to complain that no such path is
 known to Git.  They have been taught to do the normalization.

 Will this be rerolled?
 ($gmane/290724)


* nd/shallow-deepen (2016-04-13) 26 commits
 - fetch, upload-pack: --deepen=N extends shallow boundary by N commits
 - upload-pack: add get_reachable_list()
 - upload-pack: split check_unreachable() in two, prep for get_reachable_list()
 - t5500, t5539: tests for shallow depth excluding a ref
 - clone: define shallow clone boundary with --shallow-exclude
 - fetch: define shallow boundary with --shallow-exclude
 - upload-pack: support define shallow boundary by excluding revisions
 - refs: add expand_ref()
 - t5500, t5539: tests for shallow depth since a specific date
 - clone: define shallow clone boundary based on time with --shallow-since
 - fetch: define shallow boundary with --shallow-since
 - upload-pack: add deepen-since to cut shallow repos based on time
 - shallow.c: implement a generic shallow boundary finder based on rev-list
 - fetch-pack: use a separate flag for fetch in deepening mode
 - fetch-pack.c: mark strings for translating
 - fetch-pack: use a common function for verbose printing
 - fetch-pack: use skip_prefix() instead of starts_with()
 - upload-pack: move rev-list code out of check_non_tip()
 - upload-pack: tighten number parsing at "deepen" lines
 - upload-pack: use skip_prefix() instead of starts_with()
 - upload-pack: move "unshallow" sending code out of deepen()
 - upload-pack: remove unused variable "backup"
 - upload-pack: move "shallow" sending code out of deepen()
 - upload-pack: move shallow deepen code out of receive_needs()
 - transport-helper.c: refactor set_helper_option()
 - remote-curl.c: convert fetch_git() to use argv_array

 The existing "git fetch --depth=<n>" option was hard to use
 correctly when making the history of an existing shallow clone
 deeper.  A new option, "--deepen=<n>", has been added to make this
 easier to use.  "git clone" also learned "--shallow-since=<date>"
 and "--shallow-exclude=<tag>" options to make it easier to specify
 "I am interested only in the recent N months worth of history" and
 "Give me only the history since that version".

 Needs review.


* sg/completion-updates (2016-02-28) 21 commits
 . completion: cache the path to the repository
 . completion: extract repository discovery from __gitdir()
 . completion: don't guard git executions with __gitdir()
 . completion: consolidate silencing errors from git commands
 . completion: don't use __gitdir() for git commands
 . completion: respect 'git -C <path>'
 . completion: fix completion after 'git -C <path>'
 . completion: don't offer commands when 'git --opt' needs an argument
 . rev-parse: add '--absolute-git-dir' option
 . completion: list short refs from a remote given as a URL
 . completion: don't list 'HEAD' when trying refs completion outside of a repo
 . completion: list refs from remote when remote's name matches a directory
 . completion: respect 'git --git-dir=<path>' when listing remote refs
 . completion: fix most spots not respecting 'git --git-dir=<path>'
 . completion: ensure that the repository path given on the command line exists
 . completion tests: add tests for the __git_refs() helper function
 . completion tests: check __gitdir()'s output in the error cases
 . completion tests: consolidate getting path of current working directory
 . completion tests: make the $cur variable local to the test helper functions
 . completion tests: don't add test cruft to the test repository
 . completion: improve __git_refs()'s in-code documentation

 Will be rerolled.
 ($gmane/287839)


* az/p4-bare-no-rebase (2016-02-19) 1 commit
 - git-p4.py: Don't try to rebase on submit from bare repository

 "git p4 submit" attempts to do a rebase, which would fail if done
 in a bare repository.  Not doing this rebase would paper over the
 failure, which is what this patch does, but it is unclear what the
 side effect of not rebasing is.

 Needs a better explanation.


* nd/icase (2016-02-15) 12 commits
 - grep.c: reuse "icase" variable
 - diffcore-pickaxe: support case insensitive match on non-ascii
 - diffcore-pickaxe: "share" regex error handling code
 - grep/pcre: support utf-8
 - gettext: add is_utf8_locale()
 - grep/pcre: prepare locale-dependent tables for icase matching
 - grep/icase: avoid kwsset when -F is specified
 - grep/icase: avoid kwsset on literal non-ascii strings
 - test-regex: expose full regcomp() to the command line
 - test-regex: isolate the bug test code
 - grep: break down an "if" stmt in preparation for next changes
 - grep: allow -F -i combination

 "git grep -i" has been taught to fold case in non-ascii locales.

 Needs review.
 ($gmane/286137)


* ec/annotate-deleted (2015-11-20) 1 commit
 - annotate: skip checking working tree if a revision is provided

 Usability fix for annotate-specific "<file> <rev>" syntax with deleted
 files.

 Waiting for review.


* dg/subtree-rebase-test (2016-01-19) 1 commit
 - contrib/subtree: Add a test for subtree rebase that loses commits

 Reviewed up to v5.
 Will be rerolled.
 ($gmane/284426)


* js/am-3-merge-recursive-direct (2015-10-12) 2 commits
 - am: make a direct call to merge_recursive
 - merge_recursive_options: introduce the "gently" flag

 The merge_recursive_generic() function has been made a bit safer to
 call from inside a process.  "git am -3" was taught to make a direct
 call to the function when falling back to three-way merge.

 Being able to make a direct call would be good in general, but as a
 performance thing, the change needs to be backed up by numbers.

 I haven't gone through the "gently" change with fine toothed comb;
 I can see that the change avoids calling die(), but I haven't made
 sure that the program states (e.g. what's in the in-core index) are
 adjusted sensibly when it returns to the caller instead of dying,
 or the codepaths that used to die() are free of resource leaks.
 The original code certainly did not care the program states at the
 point of dying exactly because it knew it is going to exit, but now
 they have to care, and they need to be audited.

 Will be rerolled.
 ($gmane/292205)


* dk/gc-more-wo-pack (2016-01-13) 4 commits
 - gc: clean garbage .bitmap files from pack dir
 - t5304: ensure non-garbage files are not deleted
 - t5304: test .bitmap garbage files
 - prepare_packed_git(): find more garbage

 Follow-on to dk/gc-idx-wo-pack topic, to clean up stale
 .bitmap and .keep files.

 Waiting for a reroll.
 ($gmane/284368).


* 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]

* ss/commit-dry-run-resolve-merge-to-no-op (2016-02-17) 1 commit
  (merged to 'next' on 2016-05-10 at 2ada404)
 + wt-status.c: set commitable bit if there is a meaningful merge.

 "git commit --dry-run" reported "No, no, you cannot commit." in one
 case where "git commit" would have allowed you to commit, and this
 improves it a little bit ("git commit --dry-run --short" still does
 not give you the correct answer, for example).  This is a stop-gap
 measure in that "commit --short --dry-run" still gives an incorrect
 result.

 Will merge to 'master'.


* ak/t4151-ls-files-could-be-empty (2016-05-09) 1 commit
  (merged to 'next' on 2016-05-10 at 36ae38c)
 + t4151: make sure argument to 'test -z' is given

 Test fix.

 Will merge to 'master'.


* es/test-gpg-tags (2016-05-09) 1 commit
  (merged to 'next' on 2016-05-10 at 9fcb98b)
 + t6302: simplify non-gpg cases

 Test fix.

 Will merge to 'master'.


* jc/test-seq (2016-05-09) 2 commits
  (merged to 'next' on 2016-05-10 at 1512890)
 + test-lib-functions.sh: rewrite test_seq without Perl
 + test-lib-functions.sh: remove misleading comment on test_seq

 Test fix.

 Will merge to 'master'.


* js/windows-dotgit (2016-05-11) 2 commits
  (merged to 'next' on 2016-05-11 at d10caa2)
 + mingw: remove unnecessary definition
 + mingw: introduce the 'core.hideDotFiles' setting

 On Windows, .git and optionally any files whose name starts with a
 dot are now marked as hidden, with a core.hideDotFiles knob to
 customize this behaviour.

 Will merge to 'master'.


* nd/error-errno (2016-05-09) 41 commits
  (merged to 'next' on 2016-05-10 at 1cdeda8)
 + wrapper.c: use warning_errno()
 + vcs-svn: use error_errno()
 + upload-pack.c: use error_errno()
 + unpack-trees.c: use error_errno()
 + transport-helper.c: use error_errno()
 + sha1_file.c: use {error,die,warning}_errno()
 + server-info.c: use error_errno()
 + sequencer.c: use error_errno()
 + run-command.c: use error_errno()
 + rerere.c: use error_errno() and warning_errno()
 + reachable.c: use error_errno()
 + mailmap.c: use error_errno()
 + ident.c: use warning_errno()
 + http.c: use error_errno() and warning_errno()
 + grep.c: use error_errno()
 + gpg-interface.c: use error_errno()
 + fast-import.c: use error_errno()
 + entry.c: use error_errno()
 + editor.c: use error_errno()
 + diff-no-index.c: use error_errno()
 + credential-cache--daemon.c: use warning_errno()
 + copy.c: use error_errno()
 + connected.c: use error_errno()
 + config.c: use error_errno()
 + compat/win32/syslog.c: use warning_errno()
 + combine-diff.c: use error_errno()
 + check-racy.c: use error_errno()
 + builtin/worktree.c: use error_errno()
 + builtin/upload-archive.c: use error_errno()
 + builtin/update-index.c: prefer "err" to "errno" in process_lstat_error
 + builtin/rm.c: use warning_errno()
 + builtin/pack-objects.c: use die_errno() and warning_errno()
 + builtin/merge-file.c: use error_errno()
 + builtin/mailsplit.c: use error_errno()
 + builtin/help.c: use warning_errno()
 + builtin/fetch.c: use error_errno()
 + builtin/branch.c: use error_errno()
 + builtin/am.c: use error_errno()
 + bisect.c: use die_errno() and warning_errno()
 + usage.c: add warning_errno() and error_errno()
 + usage.c: move format processing out of die_errno()

 The code for warning_errno/die_errno has been refactored and a new
 error_errno() reporting helper is introduced.

 Will merge to 'master'.


* nd/remote-plural-ours-plus-theirs (2016-05-06) 1 commit
  (merged to 'next' on 2016-05-10 at aea08dc)
 + remote.c: specify correct plural form in "commit diverge" message

 Message fix.

 Will merge to 'master'.


* nd/test-helpers (2016-05-10) 1 commit
  (merged to 'next' on 2016-05-10 at e8ad58d)
 + wrap-for-bin.sh: regenerate bin-wrappers when switching branches

 Switching between 'master' and 'next', between which the paths to
 test helper binaries have changed, did not update bin-wrappers/*
 scripts used in tests, causing false test failures.

 Will merge to 'master'.


* tb/core-eol-fix (2016-04-25) 4 commits
  (merged to 'next' on 2016-05-10 at fa8a200)
 + convert.c: ident + core.autocrlf didn't work
 + t0027: test cases for combined attributes
 + convert: allow core.autocrlf=input and core.eol=crlf
 + t0027: make commit_chk_wrnNNO() reliable

 A couple of bugs around core.autocrlf have been fixed.

 Will merge to 'master'.


* tb/t5601-sed-fix (2016-05-09) 1 commit
  (merged to 'next' on 2016-05-10 at d3e54e8)
 + t5601: Remove trailing space in sed expression

 Test fix.

 Will merge to 'master'.


* va/i18n-remote-comment-to-align (2016-05-09) 1 commit
  (merged to 'next' on 2016-05-10 at edbacbb)
 + i18n: remote: add comment for translators

 Message fix.

 Will merge to 'master'.


* jc/linkgit-fix (2016-05-09) 1 commit
  (merged to 'next' on 2016-05-10 at 0e5ba60)
 + Documentation: fix linkgit references

 Many 'linkgit:<git documentation page>' references were broken,
 which are all fixed with this.

 Will merge to 'master'.


* js/http-custom-headers (2016-05-10) 4 commits
  (merged to 'next' on 2016-05-10 at 7cf5cca)
 + submodule: ensure that -c http.extraheader is heeded
 + Merge branch 'jk/submodule-c-credential' into js/http-custom-headers
 + t5551: make the test for extra HTTP headers more robust
 + tests: adjust the configuration for Apache 2.2
 (this branch uses jk/submodule-c-credential.)

 Update tests for "http.extraHeaders=<header>" to be portable back
 to Apache 2.2 (the original depended on <RequireAll/> which is a
 more recent feature).

 Will merge to 'master'.


* sb/submodule-deinit-all (2016-05-05) 1 commit
  (merged to 'next' on 2016-05-09 at 0fd4518)
 + submodule deinit: require '--all' instead of '.' for all submodules

 Correct faulty recommendation to use "git submodule deinit ." when
 de-initialising all submodules, which would result in a strange
 error message in a pathological corner case.

 Will merge to 'master'.


* bn/config-doc-tt-varnames (2016-05-05) 1 commit
  (merged to 'next' on 2016-05-10 at aa7b834)
 + config: consistently format $variables in monospaced font
 (this branch uses jc/config-pathname-type.)

 Doc formatting fixes.

 Will merge to 'master'.


* lp/typofixes (2016-05-06) 1 commit
  (merged to 'next' on 2016-05-09 at 59683be)
 + typofix: assorted typofixes in comments, documentation and messages

 Will merge to 'master'.


* sb/z-is-gnutar-ism (2016-05-09) 2 commits
  (merged to 'next' on 2016-05-09 at 51d527d)
 + t6041: do not compress backup tar file
 + t3513: do not compress backup tar file

 Will merge to 'master'.


* jc/test-parse-options-expect (2016-05-10) 4 commits
  (merged to 'next' on 2016-05-10 at 3ca5783)
 + t0040: convert a few tests to use test-parse-options --expect
 + t0040: remove unused test helpers
 + test-parse-options: --expect=<string> option to simplify tests
 + test-parse-options: fix output when callback option fails
 (this branch uses pb/commit-verbose-config.)

 t0040 had too many unnecessary repetitions in its test data.  Teach
 test-parse-options program so that a caller can tell what it
 expects in its output, so that these repetitions can be cleaned up.

 Will merge to 'master'.


* jc/doc-lint (2016-05-10) 1 commit
 - ci: validate "linkgit:" in documentation

 Find common mistakes when writing gitlink: in our documentation and
 drive the check from "make check-docs".


* jc/commit-tree-ignore-commit-gpgsign (2016-05-03) 1 commit
  (merged to 'next' on 2016-05-06 at 08eccb2)
 + commit-tree: do not pay attention to commit.gpgsign

 "git commit-tree" plumbing command required the user to always sign
 its result when the user sets the commit.gpgsign configuration
 variable, which was an ancient mistake.  Rework "git rebase" that
 relied on this mistake so that it reads commit.gpgsign and pass (or
 not pass) the -S option to "git commit-tree" to keep the end-user
 expectation the same, while teaching "git commit-tree" to ignore
 the configuration variable.  This will stop requiring the users to
 sign commit objects used internally as an implementation detail of
 "git stash".

 Will merge to 'master'.


* jk/push-client-deadlock-fix (2016-05-11) 2 commits
  (merged to 'next' on 2016-05-11 at 8f4abf9)
 + Windows: only add a no-op pthread_sigmask() when needed
  (merged to 'next' on 2016-05-06 at e91626c)
 + Windows: add pthread_sigmask() that does nothing

 Some Windows SDK lacks pthread_sigmask() implementation and fails
 to compile the recently updated "git push" codepath that uses it.

 Will merge to 'master'.


* sb/submodule-module-list-pathspec-fix (2016-05-03) 1 commit
  (merged to 'next' on 2016-05-06 at ab6dac3)
 + submodule deinit test: fix broken && chain in subshell

 Will merge to 'master'.


* ab/hooks (2016-05-04) 4 commits
  (merged to 'next' on 2016-05-09 at 23cf808)
 + hooks: allow customizing where the hook directory is
 + githooks.txt: minor improvements to the grammar & phrasing
 + githooks.txt: amend dangerous advice about 'update' hook ACL
 + githooks.txt: improve the intro section

 A new configuration variable core.hooksPath allows customizing
 where the hook directory is.

 Will merge to 'master'.


* jc/merge-impossible-no-commit (2016-04-26) 2 commits
 - merge: warn --no-commit merge when no new commit is created
 - merge: do not contaminate option_commit with --squash

 "git merge --no-commit" silently succeeded when there is no need to
 create any commit, either when you are more recent than the commit
 you tried to merge, or you can fast-forward to the commit you tried
 to merge.  The command gives a warning message in such cases.

 Just tying loose ends in a discussion.  Unless somebody else
 champions this topic, I'll drop it.


* mh/split-under-lock (2016-05-06) 33 commits
 - lock_ref_sha1_basic(): only handle REF_NODEREF mode
 - commit_ref_update(): remove the flags parameter
 - lock_ref_for_update(): don't resolve symrefs
 - lock_ref_for_update(): don't re-read non-symbolic references
 - refs: resolve symbolic refs first
 - ref_transaction_update(): check refname_is_safe() at a minimum
 - unlock_ref(): move definition higher in the file
 - lock_ref_for_update(): new function
 - add_update(): initialize the whole ref_update
 - verify_refname_available(): adjust constness in declaration
 - refs: don't dereference on rename
 - refs: allow log-only updates
 - delete_branches(): use resolve_refdup()
 - ref_transaction_commit(): correctly report close_ref() failure
 - ref_transaction_create(): disallow recursive pruning
 - refs: make error messages more consistent
 - lock_ref_sha1_basic(): remove unneeded local variable
 - read_raw_ref(): move docstring to header file
 - read_raw_ref(): improve docstring
 - read_raw_ref(): rename symref argument to referent
 - read_raw_ref(): clear *type at start of function
 - read_raw_ref(): rename flags argument to type
 - ref_transaction_commit(): remove local variable n
 - rename_ref(): remove unneeded local variable
 - commit_ref_update(): write error message to *err, not stderr
 - refname_is_safe(): insist that the refname already be normalized
 - refname_is_safe(): don't allow the empty string
 - refname_is_safe(): use skip_prefix()
 - remove_dir_recursively(): add docstring
 - safe_create_leading_directories(): improve docstring
 - read_raw_ref(): don't get confused by an empty directory
 - commit_ref(): if there is an empty dir in the way, delete it
 - t1404: demonstrate a bug resolving references

 Further preparatory work on the refs API before the pluggable
 backend series can land.

 Updated.  Will wait for comments for the last time, and then
 merge to 'next'.


* bn/http-cookiefile-config (2016-05-04) 2 commits
  (merged to 'next' on 2016-05-09 at d519b13)
 + http: expand http.cookieFile as a path
 + Documentation: config: improve word ordering for http.cookieFile

 "http.cookieFile" configuration variable clearly wants a pathname,
 but we forgot to treat it as such by e.g. applying tilde expansion.

 Will merge to 'master'.


* jc/config-pathname-type (2016-05-04) 1 commit
  (merged to 'next' on 2016-05-09 at 0876e55)
 + config: describe 'pathname' value type
 (this branch is used by bn/config-doc-tt-varnames.)

 Consolidate description of tilde-expansion that is done to
 configuration variables that take pathname to a single place.

 Will merge to 'master'.


* jk/submodule-c-credential (2016-05-06) 6 commits
  (merged to 'next' on 2016-05-10 at 4abe871)
 + submodule: stop sanitizing config options
 + submodule: use prepare_submodule_repo_env consistently
 + submodule--helper: move config-sanitizing to submodule.c
 + submodule: export sanitized GIT_CONFIG_PARAMETERS
 + t5550: break submodule config test into multiple sub-tests
 + t5550: fix typo in $HTTPD_URL
 (this branch is used by js/http-custom-headers.)

 An earlier addition of "sanitize_submodule_env" with 14111fc4 (git:
 submodule honor -c credential.* from command line, 2016-02-29)
 turned out to be a convoluted no-op; implement what it wanted to do
 correctly, and stop filtering settings given via "git -c var=val".

 Will merge to 'master'.


* mh/connect-leak (2016-04-28) 1 commit
 - git_connect(): fix memory leak with CONNECT_DIAG_URL

 Is already made obsolete with a patch in flight under discussion.
 ($gmane/292962)

 Will discard.


* ew/fast-import-unpack-limit (2016-05-11) 1 commit
  (merged to 'next' on 2016-05-11 at ffd4efb)
 + fast-import: implement unpack limit

 "git fast-import" learned the same performance trick to avoid
 creating too small a packfile as "git fetch" and "git push" have,
 using *.unpackLimit configuration.

 Will merge to 'master'.


* nd/worktree-various-heads (2016-04-22) 13 commits
  (merged to 'next' on 2016-05-10 at 61d3415)
 + branch: do not rename a branch under bisect or rebase
 + worktree.c: check whether branch is bisected in another worktree
 + wt-status.c: split bisect detection out of wt_status_get_state()
 + worktree.c: check whether branch is rebased in another worktree
 + worktree.c: avoid referencing to worktrees[i] multiple times
 + wt-status.c: make wt_status_check_rebase() work on any worktree
 + wt-status.c: split rebase detection out of wt_status_get_state()
 + path.c: refactor and add worktree_git_path()
 + worktree.c: mark current worktree
 + worktree.c: make find_shared_symref() return struct worktree *
 + worktree.c: store "id" instead of "git_dir"
 + path.c: add git_common_path() and strbuf_git_common_path()
 + dir.c: rename str(n)cmp_icase to fspath(n)cmp
 (this branch is used by nd/worktree-cleanup-post-head-protection.)

 The experimental "multiple worktree" feature gains more safety to
 forbid operations on a branch that is checked out or being actively
 worked on elsewhere, by noticing that e.g. it is being rebased.

 Will merge to 'master'.


* pb/commit-verbose-config (2016-05-10) 7 commits
 + commit: add a commit.verbose config variable
 + t7507-commit-verbose: improve test coverage by testing number of diffs
 + parse-options.c: make OPTION_COUNTUP respect "unspecified" values
 + t/t7507: improve test coverage
 + t0040-parse-options: improve test coverage
 + test-parse-options: print quiet as integer
 + t0040-test-parse-options.sh: fix style issues
 (this branch is used by jc/test-parse-options-expect.)

 "git commit" learned to pay attention to "commit.verbose"
 configuration variable and act as if "--verbose" option was
 given from the command line.

 Will merge to 'master'.


* jc/fsck-nul-in-commit (2016-05-10) 2 commits
  (merged to 'next' on 2016-05-10 at 3bc3ca3)
 + fsck: detect and warn a commit with embedded NUL
 + fsck_commit_buffer(): do not special case the last validation

 "git fsck" learned to catch NUL byte in a commit object as
 potential error and warn.

 Will merge to 'master'.


* jc/ll-merge-internal (2016-05-09) 3 commits
  (merged to 'next' on 2016-05-10 at a6bf1d0)
 + t6036: remove pointless test that expects failure
 + ll-merge: use a longer conflict marker for internal merge
 + ll-merge: fix typo in comment

 "git rerere" can get confused by conflict markers deliberately left
 by the inner merge step, because they are indistinguishable from
 the real conflict markers left by the outermost merge which are
 what the end user and "rerere" need to look at.  This was fixed by
 making the conflict markers left by the inner merges a bit longer.

 Will merge to 'master'.


* sb/submodule-init (2016-05-03) 7 commits
  (merged to 'next' on 2016-05-03 at 8a5fce4)
 + submodule init: redirect stdout to stderr
  (merged to 'next' on 2016-04-29 at 3e81ee88)
 + submodule--helper update-clone: abort gracefully on missing .gitmodules
 + submodule init: fail gracefully with a missing .gitmodules file
  (merged to 'next' on 2016-04-27 at afaad81)
 + submodule: port init from shell to C
 + submodule: port resolve_relative_url from shell to C
 + Merge branch 'sb/submodule-path-misc-bugs' into sb/submodule-init
 + Merge branch 'sb/submodule-helper-clone-regression-fix' into sb/submodule-init

 Update of "git submodule" to move pieces of logic to C continues.

 Will cook for a bit more in 'next'.


* jc/send-email-skip-backup (2016-04-12) 1 commit
 - send-email: detect and offer to skip backup files

 A careless invocation of "git send-email directory/" after editing
 0001-change.patch with an editor often ends up sending both
 0001-change.patch and its backup file, 0001-change.patch~, causing
 embarrassment and a minor confusion.  Detect such an input and
 offer to skip the backup files when sending the patches out.

 Needs review.


* va/i18n-misc-updates (2016-04-19) 9 commits
  (merged to 'next' on 2016-05-10 at b5dbd0d)
 + i18n: builtin/pull.c: split strings marked for translation
 + i18n: builtin/pull.c: mark placeholders for translation
 + i18n: git-parse-remote.sh: mark strings for translation
 + i18n: branch: move comment for translators
 + i18n: branch: unmark string for translation
 + i18n: builtin/rm.c: remove a comma ',' from string
 + i18n: unpack-trees: mark strings for translation
 + i18n: builtin/branch.c: mark option for translation
 + i18n: index-pack: use plural string instead of normal one

 Mark several messages for translation.

 Will merge to 'master'.


* kn/ref-filter-branch-list (2016-04-25) 17 commits
 - branch: implement '--format' option
 - branch: use ref-filter printing APIs
 - branch, tag: use porcelain output
 - ref-filter: allow porcelain to translate messages in the output
 - ref-filter: add `:dir` and `:base` options for ref printing atoms
 - ref-filter: make remote_ref_atom_parser() use refname_atom_parser_internal()
 - ref-filter: introduce symref_atom_parser() and refname_atom_parser()
 - ref-filter: introduce refname_atom_parser_internal()
 - ref-filter: make "%(symref)" atom work with the ':short' modifier
 - ref-filter: add support for %(upstream:track,nobracket)
 - ref-filter: make %(upstream:track) prints "[gone]" for invalid upstreams
 - ref-filter: introduce format_ref_array_item()
 - ref-filter: move get_head_description() from branch.c
 - ref-filter: modify "%(objectname:short)" to take length
 - ref-filter: implement %(if:equals=<string>) and %(if:notequals=<string>)
 - ref-filter: include reference to 'used_atom' within 'atom_value'
 - ref-filter: implement %(if), %(then), and %(else) atoms

 The code to list branches in "git branch" has been consolidated
 with the more generic ref-filter API.

 Needs review.


* xy/format-patch-base (2016-04-26) 4 commits
  (merged to 'next' on 2016-05-10 at dd19e0a)
 + format-patch: introduce format.useAutoBase configuration
 + format-patch: introduce --base=auto option
 + format-patch: add '--base' option to record base tree info
 + patch-ids: make commit_patch_id() a public helper function

 "git format-patch" learned a new "--base" option to record what
 (public, well-known) commit the original series was built on in
 its output.

 Will merge to 'master'.


* dt/index-helper (2016-05-09) 19 commits
 - untracked-cache: config option
 - trace: measure where the time is spent in the index-heavy operations
 - index-helper: optionally automatically run
 - index-helper: autorun mode
 - index-helper: don't run if already running
 - index-helper: kill mode
 - watchman: add a config option to enable the extension
 - unpack-trees: preserve index extensions
 - update-index: enable/disable watchman support
 - index-helper: use watchman to avoid refreshing index with lstat()
 - watchman: support watchman to reduce index refresh cost
 - read-cache: add watchman 'WAMA' extension
 - index-helper: add --detach
 - daemonize(): set a flag before exiting the main process
 - index-helper: log warnings
 - index-helper: add --strict
 - index-helper: new daemon for caching index and related stuff
 - read-cache: allow to keep mmap'd memory after reading
 - read-cache.c: fix constness of verify_hdr()

 Needs review.
 ($gmane/294056).


* 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.


* jc/merge-drop-old-syntax (2015-04-29) 1 commit
 - merge: drop 'git merge <message> HEAD <commit>' syntax

 Stop supporting "git merge <message> HEAD <commit>" syntax that has
 been deprecated since October 2007, and issues a deprecation
 warning message since v2.5.0.

 It has been reported that git-gui still uses the deprecated syntax,
 which needs to be fixed before this final step can proceed.
 ($gmane/282594)

--------------------------------------------------
[Discarded]

* jc/diff-compact-always-use-blank-heuristics (2016-04-29) 1 commit
 . diff: enable "compaction heuristics" and lose experimentation knob

 Superseded by the tip commit on the jk/diff-compact-heuristic topic.
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to [hidden email]
More majordomo info at  http://vger.kernel.org/majordomo-info.html