git
4 years agoMerge branch 'jk/credential-sample-update'
Junio C Hamano [Fri, 8 May 2020 21:25:02 +0000 (14:25 -0700)] 
Merge branch 'jk/credential-sample-update'

The samples in the credential documentation has been updated to
make it clear that we depict what would appear in the .git/config
file, by adding appropriate quotes as needed..

* jk/credential-sample-update:
  gitcredentials(7): make shell-snippet example more realistic
  gitcredentials(7): clarify quoting of helper examples

4 years agoMerge branch 'ah/userdiff-markdown'
Junio C Hamano [Fri, 8 May 2020 21:25:01 +0000 (14:25 -0700)] 
Merge branch 'ah/userdiff-markdown'

The userdiff patterns for Markdown documents have been added.

* ah/userdiff-markdown:
  userdiff: support Markdown

4 years agoMerge branch 'cb/credential-store-ignore-bogus-lines'
Junio C Hamano [Fri, 8 May 2020 21:25:01 +0000 (14:25 -0700)] 
Merge branch 'cb/credential-store-ignore-bogus-lines'

With the recent tightening of the code that is used to parse
various parts of a URL for use in the credential subsystem, a
hand-edited credential-store file causes the credential helper to
die, which is a bit too harsh to the users.  Demote the error
behaviour to just ignore and keep using well-formed lines instead.

* cb/credential-store-ignore-bogus-lines:
  credential-store: ignore bogus lines from store file
  credential-store: document the file format a bit more

4 years agoMerge branch 'dl/switch-c-option-in-error-message'
Junio C Hamano [Fri, 8 May 2020 21:25:00 +0000 (14:25 -0700)] 
Merge branch 'dl/switch-c-option-in-error-message'

In error messages that "git switch" mentions its option to create a
new branch, "-b/-B" options were shown, where "-c/-C" options
should be, which has been corrected.

* dl/switch-c-option-in-error-message:
  switch: fix errors and comments related to -c and -C

4 years agoThe seventh batch
Junio C Hamano [Tue, 5 May 2020 21:51:29 +0000 (14:51 -0700)] 
The seventh batch

Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoMerge branch 'js/partial-urlmatch'
Junio C Hamano [Tue, 5 May 2020 21:54:29 +0000 (14:54 -0700)] 
Merge branch 'js/partial-urlmatch'

The same as js/partial-urlmatch-2.17, built on more recent codebase
to avoid unnecessary merge conflicts.

* js/partial-urlmatch:
  credential: handle `credential.<partial-URL>.<key>` again
  credential: optionally allow partial URLs in credential_from_url_gently()

4 years agoMerge branch 'js/partial-urlmatch-2.17'
Junio C Hamano [Tue, 5 May 2020 21:54:29 +0000 (14:54 -0700)] 
Merge branch 'js/partial-urlmatch-2.17'

Recent updates broke parsing of "credential.<url>.<key>" where
<url> is not a full URL (e.g. [credential "https://"] helper = ...)
stopped working, which has been corrected.

* js/partial-urlmatch-2.17:
  credential: handle `credential.<partial-URL>.<key>` again
  credential: optionally allow partial URLs in credential_from_url_gently()
  credential: fix grammar

4 years agoMerge branch 'tb/commit-graph-perm-bits'
Junio C Hamano [Tue, 5 May 2020 21:54:28 +0000 (14:54 -0700)] 
Merge branch 'tb/commit-graph-perm-bits'

Some of the files commit-graph subsystem keeps on disk did not
correctly honor the core.sharedRepository settings and some were
left read-write.

* tb/commit-graph-perm-bits:
  commit-graph.c: make 'commit-graph-chain's read-only
  commit-graph.c: ensure graph layers respect core.sharedRepository
  commit-graph.c: write non-split graphs as read-only
  lockfile.c: introduce 'hold_lock_file_for_update_mode'
  tempfile.c: introduce 'create_tempfile_mode'

4 years agoMerge branch 'dl/push-recurse-submodules-fix'
Junio C Hamano [Tue, 5 May 2020 21:54:28 +0000 (14:54 -0700)] 
Merge branch 'dl/push-recurse-submodules-fix'

Code cleanup.

* dl/push-recurse-submodules-fix:
  push: unset PARSE_OPT_OPTARG for --recurse-submodules

4 years agoMerge branch 'dl/opt-callback-cleanup'
Junio C Hamano [Tue, 5 May 2020 21:54:27 +0000 (14:54 -0700)] 
Merge branch 'dl/opt-callback-cleanup'

Code cleanup.

* dl/opt-callback-cleanup:
  Use OPT_CALLBACK and OPT_CALLBACK_F

4 years agoMerge branch 'jk/test-fail-prereqs-fix'
Junio C Hamano [Tue, 5 May 2020 21:54:27 +0000 (14:54 -0700)] 
Merge branch 'jk/test-fail-prereqs-fix'

Test update.

* jk/test-fail-prereqs-fix:
  t0000: disable GIT_TEST_FAIL_PREREQS in sub-tests

4 years agoMerge branch 'dd/iso-8601-updates'
Junio C Hamano [Tue, 5 May 2020 21:54:26 +0000 (14:54 -0700)] 
Merge branch 'dd/iso-8601-updates'

The approxidate parser learns to parse seconds with fraction.

* dd/iso-8601-updates:
  date.c: allow compact version of ISO-8601 datetime
  date.c: skip fractional second part of ISO-8601
  date.c: validate and set time in a helper function
  date.c: s/is_date/set_date/

4 years agoMerge branch 'bc/wildcard-credential'
Junio C Hamano [Tue, 5 May 2020 21:54:26 +0000 (14:54 -0700)] 
Merge branch 'bc/wildcard-credential'

Update the parser used for credential.<URL>.<variable>
configuration, to handle <URL>s with '/' in them correctly.

* bc/wildcard-credential:
  credential: fix matching URLs with multiple levels in path

4 years agocredential-store: ignore bogus lines from store file
Carlo Marcelo Arenas Belón [Sat, 2 May 2020 22:34:47 +0000 (15:34 -0700)] 
credential-store: ignore bogus lines from store file

With the added checks for invalid URLs in credentials, any locally
modified store files which might have empty lines or even comments
were reported[1] failing to parse as valid credentials.

Instead of doing a hard check for credentials, do a soft one and
therefore avoid the reported fatal error.

While at it add tests for all known corruptions that are currently
ignored to keep track of them and avoid the risk of regressions.

[1] https://stackoverflow.com/a/61420852/5005936

Reported-by: Dirk <dirk@ed4u.de>
Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Helped-by: Junio C Hamano <gitster@pobox.com>
Based-on-patch-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agouserdiff: support Markdown
Ash Holland [Sat, 2 May 2020 13:15:43 +0000 (14:15 +0100)] 
userdiff: support Markdown

It's typical to find Markdown documentation alongside source code, and
having better context for documentation changes is useful; see also
commit 69f9c87d4 (userdiff: add support for Fountain documents,
2015-07-21).

The pattern is based on the CommonMark specification 0.29, section 4.2
<https://spec.commonmark.org/> but doesn't match empty headings, as
seeing them in a hunk header is unlikely to be useful.

Only ATX headings are supported, as detecting setext headings would
require printing the line before a pattern matches, or matching a
multiline pattern. The word-diff pattern is the same as the pattern for
HTML, because many Markdown parsers accept inline HTML.

Signed-off-by: Ash Holland <ash@sorrel.sh>
Acked-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoThe sixth batch
Junio C Hamano [Fri, 1 May 2020 20:37:35 +0000 (13:37 -0700)] 
The sixth batch

Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoMerge branch 'jt/v2-fetch-nego-fix'
Junio C Hamano [Fri, 1 May 2020 20:40:00 +0000 (13:40 -0700)] 
Merge branch 'jt/v2-fetch-nego-fix'

The upload-pack protocol v2 gave up too early before finding a
common ancestor, resulting in a wasteful fetch from a fork of a
project.  This has been corrected to match the behaviour of v0
protocol.

* jt/v2-fetch-nego-fix:
  fetch-pack: in protocol v2, reset in_vain upon ACK
  fetch-pack: in protocol v2, in_vain only after ACK
  fetch-pack: return enum from process_acks()

4 years agoMerge branch 'js/anonymise-push-url-in-errors'
Junio C Hamano [Fri, 1 May 2020 20:39:59 +0000 (13:39 -0700)] 
Merge branch 'js/anonymise-push-url-in-errors'

Error and verbose trace messages from "git push" did not redact
credential material embedded in URLs.

* js/anonymise-push-url-in-errors:
  push: anonymize URLs in error messages and warnings

4 years agoMerge branch 'es/bugreport'
Junio C Hamano [Fri, 1 May 2020 20:39:58 +0000 (13:39 -0700)] 
Merge branch 'es/bugreport'

The "bugreport" tool.

* es/bugreport:
  bugreport: drop extraneous includes
  bugreport: add compiler info
  bugreport: add uname info
  bugreport: gather git version and build info
  bugreport: add tool to generate debugging info
  help: move list_config_help to builtin/help

4 years agoMerge branch 'en/rebase-root-and-fork-point-are-incompatible'
Junio C Hamano [Fri, 1 May 2020 20:39:58 +0000 (13:39 -0700)] 
Merge branch 'en/rebase-root-and-fork-point-are-incompatible'

Incompatible options "--root" and "--fork-point" of "git rebase"
have been marked and documented as being incompatible.

* en/rebase-root-and-fork-point-are-incompatible:
  rebase: display an error if --root and --fork-point are both provided

4 years agoMerge branch 'ds/build-homebrew-gettext-fix'
Junio C Hamano [Fri, 1 May 2020 20:39:57 +0000 (13:39 -0700)] 
Merge branch 'ds/build-homebrew-gettext-fix'

Recent update to Homebrew used by macOS folks breaks build by
moving gettext library and necessary headers.

* ds/build-homebrew-gettext-fix:
  macOS/brew: let the build find gettext headers/libraries/msgfmt

4 years agoMerge branch 'dd/sparse-fixes'
Junio C Hamano [Fri, 1 May 2020 20:39:56 +0000 (13:39 -0700)] 
Merge branch 'dd/sparse-fixes'

Compilation fix.

* dd/sparse-fixes:
  progress.c: silence cgcc suggestion about internal linkage
  graph.c: limit linkage of internal variable
  compat/regex: move stdlib.h up in inclusion chain
  test-parse-pathspec-file.c: s/0/NULL/ for pointer type

4 years agoMerge branch 'mt/doc-worktree-ref'
Junio C Hamano [Fri, 1 May 2020 20:39:56 +0000 (13:39 -0700)] 
Merge branch 'mt/doc-worktree-ref'

Docfix.

* mt/doc-worktree-ref:
  config doc: fix reference to config.worktree info

4 years agoMerge branch 'eb/gitweb-more-trailers'
Junio C Hamano [Fri, 1 May 2020 20:39:55 +0000 (13:39 -0700)] 
Merge branch 'eb/gitweb-more-trailers'

Gitweb updates.

* eb/gitweb-more-trailers:
  gitweb: Recognize *-to and Closes/Fixes trailers

4 years agoMerge branch 'ds/multi-pack-index'
Junio C Hamano [Fri, 1 May 2020 20:39:55 +0000 (13:39 -0700)] 
Merge branch 'ds/multi-pack-index'

The multi-pack-index left mmapped file descriptors open when it
does not have to.

* ds/multi-pack-index:
  multi-pack-index: close file descriptor after mmap

4 years agoMerge branch 'ds/blame-on-bloom'
Junio C Hamano [Fri, 1 May 2020 20:39:54 +0000 (13:39 -0700)] 
Merge branch 'ds/blame-on-bloom'

"git blame" learns to take advantage of the "changed-paths" Bloom
filter stored in the commit-graph file.

* ds/blame-on-bloom:
  test-bloom: check that we have expected arguments
  test-bloom: fix some whitespace issues
  blame: drop unused parameter from maybe_changed_path
  blame: use changed-path Bloom filters
  tests: write commit-graph with Bloom filters
  revision: complicated pathspecs disable filters

4 years agoMerge branch 'gs/commit-graph-path-filter'
Junio C Hamano [Fri, 1 May 2020 20:39:53 +0000 (13:39 -0700)] 
Merge branch 'gs/commit-graph-path-filter'

Introduce an extension to the commit-graph to make it efficient to
check for the paths that were modified at each commit using Bloom
filters.

* gs/commit-graph-path-filter:
  bloom: ignore renames when computing changed paths
  commit-graph: add GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS test flag
  t4216: add end to end tests for git log with Bloom filters
  revision.c: add trace2 stats around Bloom filter usage
  revision.c: use Bloom filters to speed up path based revision walks
  commit-graph: add --changed-paths option to write subcommand
  commit-graph: reuse existing Bloom filters during write
  commit-graph: write Bloom filters to commit graph file
  commit-graph: examine commits by generation number
  commit-graph: examine changed-path objects in pack order
  commit-graph: compute Bloom filters for changed paths
  diff: halt tree-diff early after max_changes
  bloom.c: core Bloom filter implementation for changed paths.
  bloom.c: introduce core Bloom filter constructs
  bloom.c: add the murmur3 hash implementation
  commit-graph: define and use MAX_NUM_CHUNKS

4 years agoMerge branch 'tb/commit-graph-fd-exhaustion-fix'
Junio C Hamano [Fri, 1 May 2020 20:39:53 +0000 (13:39 -0700)] 
Merge branch 'tb/commit-graph-fd-exhaustion-fix'

The commit-graph code exhausted file descriptors easily when it
does not have to.

* tb/commit-graph-fd-exhaustion-fix:
  commit-graph: close descriptors after mmap
  commit-graph.c: gracefully handle file descriptor exhaustion
  t/test-lib.sh: make ULIMIT_FILE_DESCRIPTORS available to tests
  commit-graph.c: don't use discarded graph_name in error

4 years agoMerge branch 'tb/commit-graph-split-strategy'
Junio C Hamano [Fri, 1 May 2020 20:39:52 +0000 (13:39 -0700)] 
Merge branch 'tb/commit-graph-split-strategy'

"git commit-graph write" learned different ways to write out split
files.

* tb/commit-graph-split-strategy:
  Revert "commit-graph.c: introduce '--[no-]check-oids'"
  commit-graph.c: introduce '--[no-]check-oids'
  commit-graph.h: replace 'commit_hex' with 'commits'
  oidset: introduce 'oidset_size'
  builtin/commit-graph.c: introduce split strategy 'replace'
  builtin/commit-graph.c: introduce split strategy 'no-merge'
  builtin/commit-graph.c: support for '--split[=<strategy>]'
  t/helper/test-read-graph.c: support commit-graph chains

4 years agoMerge branch 'tb/reset-shallow'
Junio C Hamano [Fri, 1 May 2020 20:39:51 +0000 (13:39 -0700)] 
Merge branch 'tb/reset-shallow'

Fix in-core inconsistency after fetching into a shallow repository
that broke the code to write out commit-graph.

* tb/reset-shallow:
  shallow.c: use '{commit,rollback}_shallow_file'
  t5537: use test_write_lines and indented heredocs for readability

4 years agoMerge branch 'dd/mailinfo-with-nul'
Junio C Hamano [Fri, 1 May 2020 20:39:51 +0000 (13:39 -0700)] 
Merge branch 'dd/mailinfo-with-nul'

Tighten "git mailinfo" to notice and error out when decoded result
contains NUL in it.

* dd/mailinfo-with-nul:
  mailinfo: disallow NUL character in mail's header
  mailinfo.c: avoid strlen on strings that can contains NUL
  t4254: merge 2 steps of a single test

4 years agoMerge branch 'dl/test-must-fail-fixes-4'
Junio C Hamano [Fri, 1 May 2020 20:39:50 +0000 (13:39 -0700)] 
Merge branch 'dl/test-must-fail-fixes-4'

Test clean-up.

* dl/test-must-fail-fixes-4:
  t9819: don't use test_must_fail with p4
  t9164: use test_must_fail only on git commands
  t9160: use test_path_is_missing()
  t9141: use test_path_is_missing()
  t7508: don't use `test_must_fail test_cmp`
  t7408: replace incorrect uses of test_must_fail
  t6030: use test_path_is_missing()

4 years agoMerge branch 'jk/build-with-right-curl'
Junio C Hamano [Fri, 1 May 2020 20:39:49 +0000 (13:39 -0700)] 
Merge branch 'jk/build-with-right-curl'

The build procedure did not use the libcurl library and its include
files correctly for a custom-built installation.

* jk/build-with-right-curl:
  Makefile: avoid running curl-config unnecessarily
  Makefile: use curl-config --cflags
  Makefile: avoid running curl-config multiple times

4 years agogitcredentials(7): make shell-snippet example more realistic
Jeff King [Fri, 1 May 2020 07:33:11 +0000 (03:33 -0400)] 
gitcredentials(7): make shell-snippet example more realistic

There's an example of using your own bit of shell to act as a credential
helper, but it's not very realistic:

 - It's stupid to hand out your secret password to _every_ host. In the
   real world you'd use the config-matcher to limit it to a particular
   host.

 - We never provided a username. We can easily do that in another config
   option (you can do it in the helper, too, but this is much more
   readable).

 - We were sending the secret even for store/erase operations. This
   is OK because Git would just ignore it, but a real system would
   probably be unlocking a password store, which you wouldn't want to do
   more than necessary.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agogitcredentials(7): clarify quoting of helper examples
Jeff King [Fri, 1 May 2020 07:33:05 +0000 (03:33 -0400)] 
gitcredentials(7): clarify quoting of helper examples

We give several helper config examples, but don't make clear that these
are raw values. It's up to the user to add the appropriate quoting to
put them into a config file (either by running with "git config" and
quoting against the shell, or by adding double-quotes as appropriate
within the git-config file).

Let's flesh them out as full config blocks, which makes the syntax more
clear (and makes it possible for people to just cut-and-paste them as a
starting point). I added double-quotes to any values larger than a
single word. That isn't strictly necessary in all cases, but it
sidesteps explaining the rules about exactly when you need to quote a
value.

The existing quotes can be converted to single-quotes in one instance,
and backslash-esccaped in the other. I also swapped out backticks for
our preferred $().

Reported-by: douglas.fuller@gmail.com
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoswitch: fix errors and comments related to -c and -C
Denton Liu [Thu, 30 Apr 2020 11:54:57 +0000 (07:54 -0400)] 
switch: fix errors and comments related to -c and -C

In d787d311db (checkout: split part of it to new command 'switch',
2019-03-29), the `git switch` command was created by extracting the
common functionality of cmd_checkout() in checkout_main(). However, in
b7b5fce270 (switch: better names for -b and -B, 2019-03-29), the branch
creation and force creation options for 'switch' were changed to -c and
-C, respectively. As a result of this, error messages and comments that
previously referred to `-b` and `-B` became invalid for `git switch`.

For error messages that refer to `-b` and `-B`, use a format string
instead so that `-c` and `-C` can be printed when `git switch` is
invoked.

Reported-by: Robert Simpson
Signed-off-by: Denton Liu <liu.denton@gmail.com>
Reviewed-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoThe fifth batch
Junio C Hamano [Wed, 29 Apr 2020 23:15:42 +0000 (16:15 -0700)] 
The fifth batch

Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoMerge branch 'ps/transactional-update-ref-stdin'
Junio C Hamano [Wed, 29 Apr 2020 23:15:31 +0000 (16:15 -0700)] 
Merge branch 'ps/transactional-update-ref-stdin'

"git update-ref --stdin" learned a handful of new verbs to let the
user control ref update transactions more explicitly, which helps
as an ingredient to implement two-phase commit-style atomic
ref-updates across multiple repositories.

* ps/transactional-update-ref-stdin:
  update-ref: implement interactive transaction handling
  update-ref: read commands in a line-wise fashion
  update-ref: move transaction handling into `update_refs_stdin()`
  update-ref: pass end pointer instead of strbuf
  update-ref: drop unused argument for `parse_refname`
  update-ref: organize commands in an array
  strbuf: provide function to append whole lines
  git-update-ref.txt: add missing word
  refs: fix segfault when aborting empty transaction

4 years agoMerge branch 'en/fill-directory-exponential'
Junio C Hamano [Wed, 29 Apr 2020 23:15:30 +0000 (16:15 -0700)] 
Merge branch 'en/fill-directory-exponential'

The directory traversal code had redundant recursive calls which
made its performance characteristics exponential with respect to
the depth of the tree, which was corrected.

* en/fill-directory-exponential:
  completion: fix 'git add' on paths under an untracked directory
  Fix error-prone fill_directory() API; make it only return matches
  dir: replace double pathspec matching with single in treat_directory()
  dir: include DIR_KEEP_UNTRACKED_CONTENTS handling in treat_directory()
  dir: replace exponential algorithm with a linear one
  dir: refactor treat_directory to clarify control flow
  dir: fix confusion based on variable tense
  dir: fix broken comment
  dir: consolidate treat_path() and treat_one_path()
  dir: fix simple typo in comment
  t3000: add more testcases testing a variety of ls-files issues
  t7063: more thorough status checking

4 years agoMerge branch 'en/sparse-checkout'
Junio C Hamano [Wed, 29 Apr 2020 23:15:30 +0000 (16:15 -0700)] 
Merge branch 'en/sparse-checkout'

"sparse-checkout" UI improvements.

* en/sparse-checkout:
  sparse-checkout: provide a new reapply subcommand
  unpack-trees: failure to set SKIP_WORKTREE bits always just a warning
  unpack-trees: provide warnings on sparse updates for unmerged paths too
  unpack-trees: make sparse path messages sound like warnings
  unpack-trees: split display_error_msgs() into two
  unpack-trees: rename ERROR_* fields meant for warnings to WARNING_*
  unpack-trees: move ERROR_WOULD_LOSE_SUBMODULE earlier
  sparse-checkout: use improved unpack_trees porcelain messages
  sparse-checkout: use new update_sparsity() function
  unpack-trees: add a new update_sparsity() function
  unpack-trees: pull sparse-checkout pattern reading into a new function
  unpack-trees: do not mark a dirty path with SKIP_WORKTREE
  unpack-trees: allow check_updates() to work on a different index
  t1091: make some tests a little more defensive against failures
  unpack-trees: simplify pattern_list freeing
  unpack-trees: simplify verify_absent_sparse()
  unpack-trees: remove unused error type
  unpack-trees: fix minor typo in comment

4 years agoMerge branch 'dd/ci-swap-azure-pipelines-with-github-actions'
Junio C Hamano [Wed, 29 Apr 2020 23:15:29 +0000 (16:15 -0700)] 
Merge branch 'dd/ci-swap-azure-pipelines-with-github-actions'

Update the CI configuration to use GitHub Actions, retiring the one
based on Azure Pipelines.

* dd/ci-swap-azure-pipelines-with-github-actions:
  ci: let GitHub Actions upload failed tests' directories
  ci: add a problem matcher for GitHub Actions
  tests: when run in Bash, annotate test failures with file name/line number
  ci: retire the Azure Pipelines definition
  README: add a build badge for the GitHub Actions runs
  ci: configure GitHub Actions for CI/PR
  ci: run gem with sudo to install asciidoctor
  ci: explicit install all required packages
  ci: fix the `jobname` of the `GETTEXT_POISON` job
  ci/lib: set TERM environment variable if not exist
  ci/lib: allow running in GitHub Actions
  ci/lib: if CI type is unknown, show the environment variables

4 years agoMerge branch 'dd/ci-musl-libc'
Junio C Hamano [Wed, 29 Apr 2020 23:15:28 +0000 (16:15 -0700)] 
Merge branch 'dd/ci-musl-libc'

A new CI job to build and run test suite on linux with musl libc
has been added.

* dd/ci-musl-libc:
  travis: build and test on Linux with musl libc and busybox
  ci/linux32: libify install-dependencies step
  ci: refactor docker runner script
  ci/linux32: parameterise command to switch arch
  ci/lib-docker: preserve required environment variables
  ci: make MAKEFLAGS available inside the Docker container in the Linux32 job

4 years agoMerge branch 'dl/merge-autostash-rebase-quit-fix'
Junio C Hamano [Wed, 29 Apr 2020 23:15:27 +0000 (16:15 -0700)] 
Merge branch 'dl/merge-autostash-rebase-quit-fix'

The stash entry created by "git rebase --autosquash" to keep the
initial dirty state were discarded by mistake upon "git rebase
--quit", which has been corrected.

* dl/merge-autostash-rebase-quit-fix:
  rebase: save autostash entry into stash reflog on --quit

4 years agoMerge branch 'dl/merge-autostash'
Junio C Hamano [Wed, 29 Apr 2020 23:15:27 +0000 (16:15 -0700)] 
Merge branch 'dl/merge-autostash'

"git merge" learns the "--autostash" option.

* dl/merge-autostash: (22 commits)
  pull: pass --autostash to merge
  t5520: make test_pull_autostash() accept expect_parent_num
  merge: teach --autostash option
  sequencer: implement apply_autostash_oid()
  sequencer: implement save_autostash()
  sequencer: unlink autostash in apply_autostash()
  sequencer: extract perform_autostash() from rebase
  rebase: generify create_autostash()
  rebase: extract create_autostash()
  reset: extract reset_head() from rebase
  rebase: generify reset_head()
  rebase: use apply_autostash() from sequencer.c
  sequencer: rename stash_sha1 to stash_oid
  sequencer: make apply_autostash() accept a path
  rebase: use read_oneliner()
  sequencer: make read_oneliner() extern
  sequencer: configurably warn on non-existent files
  sequencer: make read_oneliner() accept flags
  sequencer: make file exists check more efficient
  sequencer: stop leaking buf
  ...

4 years agoRevert "commit-graph.c: introduce '--[no-]check-oids'"
Junio C Hamano [Wed, 29 Apr 2020 19:44:40 +0000 (12:44 -0700)] 
Revert "commit-graph.c: introduce '--[no-]check-oids'"

This reverts commit 7a9ce0269bc0f4ef230f930b3910b70ac3142552,
which has not yet gained consensus.

4 years agocommit-graph.c: make 'commit-graph-chain's read-only
Taylor Blau [Wed, 29 Apr 2020 17:36:46 +0000 (11:36 -0600)] 
commit-graph.c: make 'commit-graph-chain's read-only

In a previous commit, we made incremental graph layers read-only by
using 'git_mkstemp_mode' with permissions '0444'.

There is no reason that 'commit-graph-chain's should be modifiable by
the user, since they are generated at a temporary location and then
atomically renamed into place.

To ensure that these files are read-only, too, use
'hold_lock_file_for_update_mode' with the same read-only permission
bits, and let the umask and 'adjust_shared_perm' take care of the rest.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agocommit-graph.c: ensure graph layers respect core.sharedRepository
Taylor Blau [Wed, 29 Apr 2020 17:36:42 +0000 (11:36 -0600)] 
commit-graph.c: ensure graph layers respect core.sharedRepository

Non-layered commit-graphs use 'adjust_shared_perm' to make the
commit-graph file readable (or not) to a combination of the user, group,
and others.

Call 'adjust_shared_perm' for split-graph layers to make sure that these
also respect 'core.sharedRepository'. The 'commit-graph-chain' file
already respects this configuration since it uses
'hold_lock_file_for_update' (which calls 'adjust_shared_perm' eventually
in 'create_tempfile_mode').

Suggested-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agocommit-graph.c: write non-split graphs as read-only
Taylor Blau [Wed, 29 Apr 2020 17:36:38 +0000 (11:36 -0600)] 
commit-graph.c: write non-split graphs as read-only

In the previous commit, Git learned 'hold_lock_file_for_update_mode' to
allow the caller to specify the permission bits (prior to further
adjustment by the umask and shared repository permissions) used when
acquiring a temporary file.

Use this in the commit-graph machinery for writing a non-split graph to
acquire an opened temporary file with permissions read-only permissions
to match the split behavior. (In the split case, Git uses
git_mkstemp_mode' for each of the commit-graph layers with permission
bits '0444').

One can notice this discrepancy when moving a non-split graph to be part
of a new chain. This causes a commit-graph chain where all layers have
read-only permission bits, except for the base layer, which is writable
for the current user.

Resolve this discrepancy by using the new
'hold_lock_file_for_update_mode' and passing the desired permission
bits.

Doing so causes some test fallout in t5318 and t6600. In t5318, this
occurs in tests that corrupt a commit-graph file by writing into it. For
these, 'chmod u+w'-ing the file beforehand resolves the issue. The
additional spot in 'corrupt_graph_verify' is necessary because of the
extra 'git commit-graph write' beforehand (which *does* rewrite the
commit-graph file). In t6600, this is caused by copying a read-only
commit-graph file into place and then trying to replace it. For these,
make these files writable.

Helped-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoSync with js/partial-urlmatch-2.17
Junio C Hamano [Wed, 29 Apr 2020 15:39:06 +0000 (08:39 -0700)] 
Sync with js/partial-urlmatch-2.17

4 years agocredential: handle `credential.<partial-URL>.<key>` again
Johannes Schindelin [Fri, 24 Apr 2020 11:49:52 +0000 (11:49 +0000)] 
credential: handle `credential.<partial-URL>.<key>` again

In the patches for CVE-2020-11008, the ability to specify credential
settings in the config for partial URLs got lost. For example, it used
to be possible to specify a credential helper for a specific protocol:

[credential "https://"]
helper = my-https-helper

Likewise, it used to be possible to configure settings for a specific
host, e.g.:

[credential "dev.azure.com"]
useHTTPPath = true

Let's reinstate this behavior.

While at it, increase the test coverage to document and verify the
behavior with a couple other categories of partial URLs.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Reviewed-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agocredential-store: document the file format a bit more
Junio C Hamano [Tue, 28 Apr 2020 18:15:15 +0000 (11:15 -0700)] 
credential-store: document the file format a bit more

Reading a malformed credential URL line and silently ignoring it
does not mean that we support empty lines and/or "# commented" lines
forever.  We should document it to avoid confusion.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoThe fourth batch
Junio C Hamano [Tue, 28 Apr 2020 22:50:33 +0000 (15:50 -0700)] 
The fourth batch

Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoMerge branch 'jn/demote-proto2-from-default'
Junio C Hamano [Tue, 28 Apr 2020 22:50:11 +0000 (15:50 -0700)] 
Merge branch 'jn/demote-proto2-from-default'

Those fetching over protocol v2 from linux-next and other kernel
repositories are reporting that v2 often fetches way too much than
needed.

* jn/demote-proto2-from-default:
  Revert "fetch: default to protocol version 2"

4 years agoMerge branch 'jc/gnu-hurd-lets-fread-read-dirs'
Junio C Hamano [Tue, 28 Apr 2020 22:50:11 +0000 (15:50 -0700)] 
Merge branch 'jc/gnu-hurd-lets-fread-read-dirs'

GNU/Hurd is also among the ones that need the fopen() wrapper.

* jc/gnu-hurd-lets-fread-read-dirs:
  config.mak.uname: Define FREAD_READS_DIRECTORIES for GNU/Hurd

4 years agoMerge branch 'ms/doc-revision-illustration-fix'
Junio C Hamano [Tue, 28 Apr 2020 22:50:10 +0000 (15:50 -0700)] 
Merge branch 'ms/doc-revision-illustration-fix'

Docfix.

* ms/doc-revision-illustration-fix:
  docs: fix minor glitch in illustration

4 years agoMerge branch 'tm/zsh-complete-switch-restore'
Junio C Hamano [Tue, 28 Apr 2020 22:50:09 +0000 (15:50 -0700)] 
Merge branch 'tm/zsh-complete-switch-restore'

zsh command line completion (in contrib/) update.

* tm/zsh-complete-switch-restore:
  complete: zsh: add missing sub cmd completion candidates

4 years agoMerge branch 'mt/grep-cquote-path'
Junio C Hamano [Tue, 28 Apr 2020 22:50:09 +0000 (15:50 -0700)] 
Merge branch 'mt/grep-cquote-path'

"git grep" did not quote a path with unusual character like other
commands (like "git diff", "git status") do, but did quote when run
from a subdirectory, both of which has been corrected.

* mt/grep-cquote-path:
  grep: follow conventions for printing paths w/ unusual chars

4 years agoMerge branch 'ds/log-exclude-decoration-config'
Junio C Hamano [Tue, 28 Apr 2020 22:50:08 +0000 (15:50 -0700)] 
Merge branch 'ds/log-exclude-decoration-config'

The "--decorate-refs" and "--decorate-refs-exclude" options "git
log" takes have learned a companion configuration variable
log.excludeDecoration that sits at the lowest priority in the
family.

* ds/log-exclude-decoration-config:
  log: add log.excludeDecoration config option
  log-tree: make ref_filter_match() a helper method

4 years agoMerge branch 'vd/range-diff-with-custom-pretty-format-fix'
Junio C Hamano [Tue, 28 Apr 2020 22:50:07 +0000 (15:50 -0700)] 
Merge branch 'vd/range-diff-with-custom-pretty-format-fix'

"git range-diff" fixes.

* vd/range-diff-with-custom-pretty-format-fix:
  range-diff: avoid negative string precision
  range-diff: fix a crash in parsing git-log output

4 years agoMerge branch 'tb/diff-tree-with-notes'
Junio C Hamano [Tue, 28 Apr 2020 22:50:07 +0000 (15:50 -0700)] 
Merge branch 'tb/diff-tree-with-notes'

"git diff-tree --pretty --notes" used to hit an assertion failure,
as it forgot to initialize the notes subsystem.

* tb/diff-tree-with-notes:
  diff-tree.c: load notes machinery when required

4 years agoMerge branch 'eb/mboxrd-doc'
Junio C Hamano [Tue, 28 Apr 2020 22:50:06 +0000 (15:50 -0700)] 
Merge branch 'eb/mboxrd-doc'

Doc update.

* eb/mboxrd-doc:
  Documentation: explain "mboxrd" pretty format

4 years agoMerge branch 'js/stash-p-fix'
Junio C Hamano [Tue, 28 Apr 2020 22:50:06 +0000 (15:50 -0700)] 
Merge branch 'js/stash-p-fix'

Allowing the user to split a patch hunk while "git stash -p" does
not work well; a band-aid has been added to make this (partially)
work better.

* js/stash-p-fix:
  stash -p: (partially) fix bug concerning split hunks
  t3904: fix incorrect demonstration of a bug

4 years agoMerge branch 'dl/libify-a-few'
Junio C Hamano [Tue, 28 Apr 2020 22:50:05 +0000 (15:50 -0700)] 
Merge branch 'dl/libify-a-few'

Code in builtin/*, i.e. those can only be called from within
built-in subcommands, that implements bulk of a couple of
subcommands have been moved to libgit.a so that they could be used
by others.

* dl/libify-a-few:
  Lib-ify prune-packed
  Lib-ify fmt-merge-msg

4 years agoMerge branch 'jx/atomic-push'
Junio C Hamano [Tue, 28 Apr 2020 22:50:04 +0000 (15:50 -0700)] 
Merge branch 'jx/atomic-push'

"git push --atomic" used to show failures for refs that weren't
even pushed, which has been corrected.

* jx/atomic-push:
  transport-helper: new method reject_atomic_push()
  transport-helper: mark failure for atomic push
  send-pack: mark failure of atomic push properly
  t5543: never report what we do not push
  send-pack: fix inconsistent porcelain output

4 years agoMerge branch 'jt/avoid-prefetch-when-able-in-diff'
Junio C Hamano [Tue, 28 Apr 2020 22:50:04 +0000 (15:50 -0700)] 
Merge branch 'jt/avoid-prefetch-when-able-in-diff'

"git diff" in a partial clone learned to avoid lazy loading blob
objects in more casese when they are not needed.

* jt/avoid-prefetch-when-able-in-diff:
  diff: restrict when prefetching occurs
  diff: refactor object read
  diff: make diff_populate_filespec_options struct
  promisor-remote: accept 0 as oid_nr in function

4 years agoMerge branch 'js/subtree-doc-update-to-asciidoctor-2'
Junio C Hamano [Tue, 28 Apr 2020 22:50:03 +0000 (15:50 -0700)] 
Merge branch 'js/subtree-doc-update-to-asciidoctor-2'

Doc markup update.

* js/subtree-doc-update-to-asciidoctor-2:
  subtree: fix build with AsciiDoctor 2

4 years agoMerge branch 'ds/t5319-touch-fix'
Junio C Hamano [Tue, 28 Apr 2020 22:50:02 +0000 (15:50 -0700)] 
Merge branch 'ds/t5319-touch-fix'

Tests update to use "test-chmtime" instead of "touch -t".

* ds/t5319-touch-fix:
  t5319: replace 'touch -m' with 'test-tool chmtime'

4 years agoMerge branch 'ds/commit-graph-expiry-fix'
Junio C Hamano [Tue, 28 Apr 2020 22:50:02 +0000 (15:50 -0700)] 
Merge branch 'ds/commit-graph-expiry-fix'

"git commit-graph write --expire-time=<timestamp>" did not use the
given timestamp correctly, which has been corrected.

* ds/commit-graph-expiry-fix:
  commit-graph: fix buggy --expire-time option

4 years agoMerge branch 'dr/doc-recurse-submodules'
Junio C Hamano [Tue, 28 Apr 2020 22:50:01 +0000 (15:50 -0700)] 
Merge branch 'dr/doc-recurse-submodules'

Documentation updates around the "--recurse-submodules" option.

* dr/doc-recurse-submodules:
  doc: --recurse-submodules mostly applies to active submodules
  doc: be more precise on (fetch|push).recurseSubmodules
  doc: explain how to deactivate submodule.recurse completely
  doc: document --recurse-submodules for reset and restore
  doc: list all commands affected by submodule.recurse

4 years agoMerge branch 'jc/log-no-mailmap'
Junio C Hamano [Tue, 28 Apr 2020 22:50:00 +0000 (15:50 -0700)] 
Merge branch 'jc/log-no-mailmap'

"git log" learns "--[no-]mailmap" as a synonym to "--[no-]use-mailmap"

* jc/log-no-mailmap:
  log: give --[no-]use-mailmap a more sensible synonym --[no-]mailmap
  clone: reorder --recursive/--recurse-submodules
  parse-options: teach "git cmd -h" to show alias as alias

4 years agoMerge branch 'ma/doc-discard-docbook-xsl-1.73'
Junio C Hamano [Tue, 28 Apr 2020 22:50:00 +0000 (15:50 -0700)] 
Merge branch 'ma/doc-discard-docbook-xsl-1.73'

Raise the minimum required version of docbook-xsl package to 1.74,
as 1.74.0 was from late 2008, which is more than 10 years old, and
drop compatibility cruft from our documentation suite.

* ma/doc-discard-docbook-xsl-1.73:
  user-manual.conf: don't specify [listingblock]
  INSTALL: drop support for docbook-xsl before 1.74
  manpage-normal.xsl: fold in manpage-base.xsl
  manpage-bold-literal.xsl: stop using git.docbook.backslash
  Doc: drop support for docbook-xsl before 1.73.0
  Doc: drop support for docbook-xsl before 1.72.0
  Doc: drop support for docbook-xsl before 1.71.1

4 years agoMerge branch 'lx/submodule-clear-variables'
Junio C Hamano [Tue, 28 Apr 2020 22:49:59 +0000 (15:49 -0700)] 
Merge branch 'lx/submodule-clear-variables'

The "git submodule" command did not initialize a few variables it
internally uses and was affected by variable settings leaked from
the environment.

* lx/submodule-clear-variables:
  git-submodule.sh: setup uninitialized variables

4 years agoMerge branch 'jk/fast-import-use-hashmap'
Junio C Hamano [Tue, 28 Apr 2020 22:49:58 +0000 (15:49 -0700)] 
Merge branch 'jk/fast-import-use-hashmap'

The custom hash function used by "git fast-import" has been
replaced with the one from hashmap.c, which gave us a nice
performance boost.

* jk/fast-import-use-hashmap:
  fast-import: replace custom hash with hashmap.c

4 years agoMerge branch 'jk/config-use-size-t'
Junio C Hamano [Tue, 28 Apr 2020 22:49:58 +0000 (15:49 -0700)] 
Merge branch 'jk/config-use-size-t'

The config API made mixed uses of int and size_t types to represent
length of various pieces of text it parsed, which has been updated
to use the correct type (i.e. size_t) throughout.

* jk/config-use-size-t:
  config: reject parsing of files over INT_MAX
  config: use size_t to store parsed variable baselen
  git_config_parse_key(): return baselen as size_t
  config: drop useless length variable in write_pair()
  parse_config_key(): return subsection len as size_t
  remote: drop auto-strlen behavior of make_branch() and make_rewrite()

4 years agoMerge branch 'bc/constant-memequal'
Junio C Hamano [Tue, 28 Apr 2020 22:49:57 +0000 (15:49 -0700)] 
Merge branch 'bc/constant-memequal'

Validation of push certificate has been made more robust against
timing attacks.

* bc/constant-memequal:
  receive-pack: compilation fix
  builtin/receive-pack: use constant-time comparison for HMAC value

4 years agoMerge branch 'lr/freshen-file-fix'
Junio C Hamano [Tue, 28 Apr 2020 22:49:56 +0000 (15:49 -0700)] 
Merge branch 'lr/freshen-file-fix'

The code that refreshes the last access and modified time of
on-disk packfiles and loose object files have been updated.

* lr/freshen-file-fix:
  freshen_file(): use NULL `times' for implicit current-time

4 years agoMerge branch 'en/rebase-doc-hooks-called-by-accident'
Junio C Hamano [Tue, 28 Apr 2020 22:49:56 +0000 (15:49 -0700)] 
Merge branch 'en/rebase-doc-hooks-called-by-accident'

"git rebase" happens to call some hooks meant for "checkout" and
"commit" by this was not a designed behaviour than historical
accident.  This has been documented.

* en/rebase-doc-hooks-called-by-accident:
  git-rebase.txt: add another hook to the hooks section, and explain more

4 years agoMerge branch 'jc/doc-test-leaving-early'
Junio C Hamano [Tue, 28 Apr 2020 22:49:55 +0000 (15:49 -0700)] 
Merge branch 'jc/doc-test-leaving-early'

Document the recommended way to abort a failing test early (e.g. by
exiting a loop), which is to say "return 1".

* jc/doc-test-leaving-early:
  t/README: suggest how to leave test early with failure

4 years agoMerge branch 'dd/test-with-busybox'
Junio C Hamano [Tue, 28 Apr 2020 22:49:54 +0000 (15:49 -0700)] 
Merge branch 'dd/test-with-busybox'

Various tests have been updated to work around issues found with
shell utilities that come with busybox etc.

* dd/test-with-busybox:
  t5703: feed raw data into test-tool unpack-sideband
  t4124: tweak test so that non-compliant diff(1) can also be used
  t7063: drop non-POSIX argument "-ls" from find(1)
  t5616: use rev-parse instead to get HEAD's object_id
  t5003: skip conversion test if unzip -a is unavailable
  t5003: drop the subshell in test_lazy_prereq
  test-lib-functions: test_cmp: eval $GIT_TEST_CMP
  t4061: use POSIX compliant regex(7)

4 years agopush: anonymize URLs in error messages and warnings
Johannes Schindelin [Fri, 24 Apr 2020 14:20:08 +0000 (14:20 +0000)] 
push: anonymize URLs in error messages and warnings

Just like 47abd85ba0 (fetch: Strip usernames from url's before storing
them, 2009-04-17) and later 882d49ca5c (push: anonymize URL in status
output, 2016-07-13), and even later c1284b21f243 (curl: anonymize URLs
in error messages and warnings, 2019-03-04) this change anonymizes URLs
(read: strips them of user names and especially passwords) in
user-facing error messages and warnings.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Reviewed-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agorebase: save autostash entry into stash reflog on --quit
Denton Liu [Tue, 28 Apr 2020 09:31:31 +0000 (05:31 -0400)] 
rebase: save autostash entry into stash reflog on --quit

In a03b55530a (merge: teach --autostash option, 2020-04-07), the
--autostash option was introduced for `git merge`. Notably, when
`git merge --quit` is run with an autostash entry present, it is saved
into the stash reflog. This is contrasted with the current behaviour of
`git rebase --quit` where the autostash entry is simply just dropped out
of existence.

Adopt the behaviour of `git merge --quit` in `git rebase --quit` and
save the autostash entry into the stash reflog instead of just deleting
it.

Signed-off-by: Denton Liu <liu.denton@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agopush: unset PARSE_OPT_OPTARG for --recurse-submodules
Denton Liu [Mon, 27 Apr 2020 06:44:08 +0000 (02:44 -0400)] 
push: unset PARSE_OPT_OPTARG for --recurse-submodules

When the usage for `git push` is shown, it includes the following
lines

--recurse-submodules[=(check|on-demand|no)]
      control recursive pushing of submodules

which seem to indicate that the argument for --recurse-submodules is
optional. However, we cannot actually run that optiion without an
argument:

$ git push --recurse-submodules
fatal: recurse-submodules missing parameter

Unset PARSE_OPT_OPTARG so that it is clear that this option requires an
argument. Since the parse-options machinery guarantees that an argument
is present now, assume that `arg` is set in the else of
option_parse_recurse_submodules().

Reported-by: Andrew White <andrew.white@audinate.com>
Signed-off-by: Denton Liu <liu.denton@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoUse OPT_CALLBACK and OPT_CALLBACK_F
Denton Liu [Tue, 28 Apr 2020 08:36:28 +0000 (04:36 -0400)] 
Use OPT_CALLBACK and OPT_CALLBACK_F

In the codebase, there are many options which use OPTION_CALLBACK in a
plain ol' struct definition. However, we have the OPT_CALLBACK and
OPT_CALLBACK_F macros which are meant to abstract these plain struct
definitions away. These macros are useful as they semantically signal to
developers that these are just normal callback option with nothing fancy
happening.

Replace plain struct definitions of OPTION_CALLBACK with OPT_CALLBACK or
OPT_CALLBACK_F where applicable. The heavy lifting was done using the
following (disgusting) shell script:

#!/bin/sh

do_replacement () {
tr '\n' '\r' |
sed -e 's/{\s*OPTION_CALLBACK,\s*\([^,]*\),\([^,]*\),\([^,]*\),\([^,]*\),\([^,]*\),\s*0,\(\s*[^[:space:]}]*\)\s*}/OPT_CALLBACK(\1,\2,\3,\4,\5,\6)/g' |
sed -e 's/{\s*OPTION_CALLBACK,\s*\([^,]*\),\([^,]*\),\([^,]*\),\([^,]*\),\([^,]*\),\([^,]*\),\(\s*[^[:space:]}]*\)\s*}/OPT_CALLBACK_F(\1,\2,\3,\4,\5,\6,\7)/g' |
tr '\r' '\n'
}

for f in $(git ls-files \*.c)
do
do_replacement <"$f" >"$f.tmp"
mv "$f.tmp" "$f"
done

The result was manually inspected and then reformatted to match the
style of the surrounding code. Finally, using
`git grep OPTION_CALLBACK \*.c`, leftover results which were not handled
by the script were manually transformed.

Signed-off-by: Denton Liu <liu.denton@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agot0000: disable GIT_TEST_FAIL_PREREQS in sub-tests
Jeff King [Tue, 28 Apr 2020 08:14:21 +0000 (04:14 -0400)] 
t0000: disable GIT_TEST_FAIL_PREREQS in sub-tests

The test added by 477dcaddb6 (tests: do not let lazy prereqs inside
`test_expect_*` turn off tracing, 2020-03-26) runs a sub-test script
that traces a test with a lazy prereq, like:

  test_have_prereq LAZY && echo trace

That won't work if GIT_TEST_FAIL_PREREQS is set in the environment,
because our have_prereq will report failure, and we won't run the echo
at all.

We could work around this by avoiding the &&-chain, but we can
fix this and any future tests at once by unsetting that variable for our
sub-tests. These are meant to be controlled environments where we test
the test-suite itself; the outer test snippet should be in charge of the
sub-test environment, not whatever mode the user happens to be running
in.

Reported-by: Son Luong Ngoc <sluongng@gmail.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agofetch-pack: in protocol v2, reset in_vain upon ACK
Jonathan Tan [Tue, 28 Apr 2020 00:01:10 +0000 (17:01 -0700)] 
fetch-pack: in protocol v2, reset in_vain upon ACK

In the function process_acks() in fetch-pack.c, the variable
received_ack is meant to track that an ACK was received, but it was
never set. This results in negotiation terminating prematurely through
the in_vain counter, when the counter should have been reset upon every
ACK.

Therefore, reset the in_vain counter upon every ACK.

Helped-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agofetch-pack: in protocol v2, in_vain only after ACK
Jonathan Tan [Tue, 28 Apr 2020 00:01:09 +0000 (17:01 -0700)] 
fetch-pack: in protocol v2, in_vain only after ACK

When fetching, Git stops negotiation when it has sent at least
MAX_IN_VAIN (which is 256) "have" lines without having any of them
ACK-ed. But this is supposed to trigger only after the first ACK, as
pack-protocol.txt says:

  However, the 256 limit *only* turns on in the canonical client
  implementation if we have received at least one "ACK %s continue"
  during a prior round.  This helps to ensure that at least one common
  ancestor is found before we give up entirely.

The code path for protocol v0 observes this, but not protocol v2,
resulting in shorter negotiation rounds but significantly larger
packfiles. Teach the code path for protocol v2 to check this criterion
only after at least one ACK was received.

Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agofetch-pack: return enum from process_acks()
Jonathan Tan [Tue, 28 Apr 2020 00:01:08 +0000 (17:01 -0700)] 
fetch-pack: return enum from process_acks()

process_acks() returns 0, 1, or 2, depending on whether "ready" was
received and if not, whether at least one commit was found to be common.
Replace these magic numbers with a documented enum.

Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agobugreport: drop extraneous includes
Emily Shaffer [Mon, 27 Apr 2020 23:42:31 +0000 (16:42 -0700)] 
bugreport: drop extraneous includes

In the generic parts of the source files, system headers like
<time.h> and <stdio.h> are supposed to be included indirectly
by including "git-compat-util.h", which manages portability issues.

Drop our explicit inclusions and rely on "cache.h", which includes
"git-compat-util.h".

Signed-off-by: Emily Shaffer <emilyshaffer@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agorebase: display an error if --root and --fork-point are both provided
Elijah Newren [Mon, 27 Apr 2020 17:59:49 +0000 (17:59 +0000)] 
rebase: display an error if --root and --fork-point are both provided

--root implies we want to rebase all commits since the beginning of
history.  --fork-point means we want to use the reflog of the specified
upstream to find the best common ancestor between <upstream> and
<branch> and only rebase commits since that common ancestor.  These
options are clearly contradictory, so throw an error (instead of
segfaulting on a NULL pointer) if both are specified.

Reported-by: Alexander Berg <alexander.berg@atos.net>
Documentation-by: Alban Gruin <alban.gruin@gmail.com>
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agocredential: fix matching URLs with multiple levels in path
brian m. carlson [Mon, 27 Apr 2020 01:18:08 +0000 (01:18 +0000)] 
credential: fix matching URLs with multiple levels in path

46fd7b3900 ("credential: allow wildcard patterns when matching config",
2020-02-20) introduced support for matching credential helpers using
urlmatch.  In doing so, it introduced code to percent-encode the paths
we get from the credential helper so that they could be effectively
matched by the urlmatch code.

Unfortunately, that code had a bug: it percent-encoded the slashes in
the path, resulting in any URL path that contained multiple levels
(i.e., a directory component) not matching.

We are currently the only caller of the percent-encoding code and could
simply change it not to encode slashes.  However, we still want to
encode slashes in the username component, so we need to have both
behaviors available.

So instead, let's add a flag to control encoding slashes, which is the
behavior we want here, and use it when calling the code in this case.

Add a test for credential helper URLs using multiple slashes in the
path, which our test suite previously lacked, as well as one ensuring
that we handle usernames with slashes gracefully.  Since we're testing
other percent-encoding handling, let's add one for non-ASCII UTF-8
characters as well.

Reported-by: Ilya Tretyakov <it@it3xl.ru>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agomacOS/brew: let the build find gettext headers/libraries/msgfmt
Johannes Schindelin [Sun, 26 Apr 2020 20:09:32 +0000 (22:09 +0200)] 
macOS/brew: let the build find gettext headers/libraries/msgfmt

Apparently a recent Homebrew update now installs `gettext` into the
subdirectory /usr/local/opt/gettext/[lib/include].

Sometimes the ci job succeeds:
 brew link --force gettext
 Linking /usr/local/Cellar/gettext/0.20.1... 179 symlinks created

And sometimes installing the package "gettext" with force-link fails:
 brew link --force gettext
 Warning: Refusing to link macOS provided/shadowed software: gettext
 If you need to have gettext first in your PATH run:
  echo 'export PATH="/usr/local/opt/gettext/bin:$PATH"' >> ~/.bash_profile

(And the is not the final word either, since macOS itself says:
 The default interactive shell is now zsh.)

Anyway, The latter requires CFLAGS to include /usr/local/opt/gettext/include
and LDFLAGS to include /usr/local/opt/gettext/lib.

Likewise, the `msgfmt` tool is no longer in the `PATH`.

While it is unclear which change is responsible for this breakage (that
most notably only occurs on CI build agents that updated very recently),
https://github.com/Homebrew/homebrew-core/pull/53489 has fixed it.

Nevertheless, let's work around this issue, as there are still quite a
few build agents out there that need some help in this regard: we
explicitly do not call `brew update` in our CI/PR builds anymore.

Helped-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agolockfile.c: introduce 'hold_lock_file_for_update_mode'
Taylor Blau [Mon, 27 Apr 2020 16:27:58 +0000 (10:27 -0600)] 
lockfile.c: introduce 'hold_lock_file_for_update_mode'

We use 'hold_lock_file_for_update' (and the '_timeout') variant to
acquire a lock when updating references, the commit-graph file, and so
on.

In particular, the commit-graph machinery uses this to acquire a
temporary file that is used to write a non-split commit-graph. In a
subsequent commit, an issue in the commit-graph machinery produces
graph files that have a different permission based on whether or not
they are part of a multi-layer graph will be addressed.

To do so, the commit-graph machinery will need a version of
'hold_lock_file_for_update' that takes the permission bits from the
caller.

Introduce such a function in this patch for both the
'hold_lock_file_for_update' and 'hold_lock_file_for_update_timeout'
functions, and leave the existing functions alone by inlining their
definitions in terms of the new mode variants.

Note that, like in the previous commit, 'hold_lock_file_for_update_mode'
is not guarenteed to set the given mode, since it may be modified by
both the umask and 'core.sharedRepository'.

Note also that even though the commit-graph machinery only calls
'hold_lock_file_for_update', that this is defined in terms of
'hold_lock_file_for_update_timeout', and so both need an additional mode
parameter here.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agotempfile.c: introduce 'create_tempfile_mode'
Taylor Blau [Mon, 27 Apr 2020 16:27:54 +0000 (10:27 -0600)] 
tempfile.c: introduce 'create_tempfile_mode'

In the next patch, 'hold_lock_file_for_update' will gain an additional
'mode' parameter to specify permissions for the associated temporary
file.

Since the lockfile.c machinery uses 'create_tempfile' which always
creates a temporary file with global read-write permissions, introduce a
variant here that allows specifying the mode.

Note that the mode given to 'create_tempfile_mode' is not guaranteed to
be written to disk, since it is subject to both the umask and
'core.sharedRepository'.

Arguably, all temporary files should have permission 0444, since they
are likely to be renamed into place and then not written to again. This
is a much larger change than we may want to take on in this otherwise
small patch, so for the time being, make 'create_tempfile' behave as it
has always done by inlining it to 'create_tempfile_mode' with mode set
to '0666'.

Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoprogress.c: silence cgcc suggestion about internal linkage
Đoàn Trần Công Danh [Mon, 27 Apr 2020 14:22:37 +0000 (21:22 +0700)] 
progress.c: silence cgcc suggestion about internal linkage

Signed-off-by: Đoàn Trần Công Danh <congdanhqx@gmail.com>
Reviewed-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agograph.c: limit linkage of internal variable
Đoàn Trần Công Danh [Mon, 27 Apr 2020 14:22:36 +0000 (21:22 +0700)] 
graph.c: limit linkage of internal variable

Signed-off-by: Đoàn Trần Công Danh <congdanhqx@gmail.com>
Reviewed-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agocompat/regex: move stdlib.h up in inclusion chain
Đoàn Trần Công Danh [Mon, 27 Apr 2020 14:22:35 +0000 (21:22 +0700)] 
compat/regex: move stdlib.h up in inclusion chain

In Linux with musl libc, we have this inclusion chain:

compat/regex/regex.c:69
`-> compat/regex/regex_internal.h
   `-> /usr/include/stdlib.h
      `-> /usr/include/features.h
      `-> /usr/include/alloca.h

In that inclusion chain, `<features.h>` claims it's _BSD_SOURCE
compatible when it's NOT asked to be either
{_POSIX,_GNU,_XOPEN,_BSD}_SOURCE, or __STRICT_ANSI__.
And, `<stdlib.h>` will include `<alloca.h>` to be compatible with
software written for GNU and BSD. Thus, redefine `alloca` macro,
which was defined before at compat/regex/regex.c:66.

Considering this is only compat code, we've taken from other project,
it's not our business to decide which source should we adhere to.

Include `<stdlib.h>` early to prevent the redefinition of alloca.
This also remove a potential warning about alloca not defined on:
#undef alloca

Helped-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Đoàn Trần Công Danh <congdanhqx@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agotest-parse-pathspec-file.c: s/0/NULL/ for pointer type
Đoàn Trần Công Danh [Mon, 27 Apr 2020 14:22:34 +0000 (21:22 +0700)] 
test-parse-pathspec-file.c: s/0/NULL/ for pointer type

Signed-off-by: Đoàn Trần Công Danh <congdanhqx@gmail.com>
Reviewed-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agocommit-graph: close descriptors after mmap
Jeff King [Thu, 23 Apr 2020 21:41:13 +0000 (15:41 -0600)] 
commit-graph: close descriptors after mmap

We don't ever refer to the descriptor after mmap-ing it. And keeping it
open means we can run out of descriptors in degenerate cases (e.g.,
thousands of split chain files). Let's close it as soon as possible.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agogitweb: Recognize *-to and Closes/Fixes trailers
Emma Brooks [Sat, 25 Apr 2020 02:17:23 +0000 (02:17 +0000)] 
gitweb: Recognize *-to and Closes/Fixes trailers

Commit trailers like "Thanks-to:", "Fixes:", and "Closes:" are fairly
common, but gitweb didn't highlight them like other trailers.

Signed-off-by: Emma Brooks <me@pluvano.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years agoconfig doc: fix reference to config.worktree info
Matheus Tavares [Sat, 25 Apr 2020 01:05:38 +0000 (22:05 -0300)] 
config doc: fix reference to config.worktree info

356aea6 ("doc: move extensions.worktreeConfig to the right place",
2018-11-14) moved the explanation of extension.worktreeConfig from
config.txt to technical/repository-version.txt. However, the former
still contains a reference to the removed paragraph. We could fix it
referencing the gitrepository-layout man page, which contains the moved
explanation. But the git-worktree man page has additional information
and recommendations for the worktree config file, so let's reference it
instead.

Signed-off-by: Matheus Tavares <matheus.bernardino@usp.br>
Signed-off-by: Junio C Hamano <gitster@pobox.com>