The fsmonitor interface got out of sync after the in-core index
file gets discarded, which has been corrected.
* js/fsmonitor-refresh-after-discarding-index:
fsmonitor: force a refresh after the index was discarded
fsmonitor: demonstrate that it is not refreshed after discard_index()
To run tests for Git SVN, our scripts for CI used to install the
git-svn package (in the hope that it would bring in the right
dependencies). This has been updated to install the more direct
dependency, namely, libsvn-perl.
* sg/ci-libsvn-perl:
ci: install 'libsvn-perl' instead of 'git-svn'
On a filesystem like HFS+, the names of the refs stored as filesystem
entities may become different from what the end-user expects, just
like files in the working tree get "renamed". Work around the
mismatch by paying attention to the core.precomposeUnicode
configuration.
* en/unicode-in-refnames:
Honor core.precomposeUnicode in more places
Update "git difftool" and "git mergetool" so that the combinations
of {diff,merge}.{tool,guitool} configuration variables serve as
fallback settings of each other in a sensible order.
* dl/difftool-mergetool:
difftool: fallback on merge.guitool
difftool: make --gui, --tool and --extcmd mutually exclusive
mergetool: fallback to tool when guitool unavailable
mergetool--lib: create gui_mode function
mergetool: use get_merge_tool function
t7610: add mergetool --gui tests
t7610: unsuppress output
Attempt to use an abbreviated option in "git clone --recurs" is
responded by a request to disambiguate between --recursive and
--recurse-submodules, which is bad because these two are synonyms.
The parse-options API has been extended to define such synonyms
more easily and not produce an unnecessary failure.
* nd/parse-options-aliases:
parse-options: don't emit "ambiguous option" for aliases
"git branch new A...B" and "git checkout -b new A...B" have been
taught that in their contexts, the notation A...B means "the merge
base between these two commits", just like "git checkout A...B"
detaches HEAD at that commit.
* dl/branch-from-3dot-merge-base:
branch: make create_branch accept a merge base rev
t2018: cleanup in current test
Performance test framework has been broken and measured the version
of Git that happens to be on $PATH, not the specified one to
measure, for a while, which has been corrected.
* ab/perf-installed-fix:
perf-lib.sh: forbid the use of GIT_TEST_INSTALLED
perf tests: add "bindir" prefix to git tree test results
perf-lib.sh: remove GIT_TEST_INSTALLED from perf-lib.sh
perf-lib.sh: make "./run <revisions>" use the correct gits
perf aggregate: remove GIT_TEST_INSTALLED from --codespeed
perf README: correct docs for 3c8f12c96c regression
During the course of review for MyFirstContribution.txt, the suggestion
came up to include anchors to make it easier for veteran contributors to
link specific sections of this documents to newbies. To make life easier
for reviewers, add these anchors in their own commit. See review context
here: https://public-inbox.org/git/20190507195938.GD220818@google.com/
AsciiDoc does not support :sectanchors: and the anchors are not
discoverable, but they are referenceable. So a link to
"foo.com/MyFirstContribution.html#prerequisites" will still work if that
file was generated with AsciiDoc. The inclusion of :sectanchors: does
not create warnings or errors while compiling directly with `asciidoc -b
html5 Documentation/MyFirstContribution.txt` or while compiling with
`make doc`.
AsciiDoctor does support :sectanchors: and displays a paragraph link on
mouseover. When the anchor is included above or inline with a section
(as in this change), the link provided points to the custom ID contained
within [[]] instead of to an autogenerated ID. Practically speaking,
this means we have .../MyFirstContribution.html#summary instead of
.../MyFirstContribution.html#_summary. In addition to being prettier,
the custom IDs also enable anchor linking to work with
asciidoc-generated pages. This change compiles with no warnings using
`asciidoctor -b html5 Documentation/MyFirstContribution.txt`.
Signed-off-by: Emily Shaffer <emilyshaffer@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
documentation: add tutorial for first contribution
This tutorial covers how to add a new command to Git and, in the
process, everything from cloning git/git to getting reviewed on the
mailing list. It's meant for new contributors to go through
interactively, learning the techniques generally used by the git/git
development community.
Reviewed-by: Johannes Schindelin <johannes.schindelin@gmx.de> Reviewed-by: Jonathan Tan <jonathantanmy@google.com> Signed-off-by: Emily Shaffer <emilyshaffer@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
This allows to cancel the current merge without resetting worktree/index,
which is what --abort is for. Like other --quit(s), this is often used
when you forgot that you're in the middle of a merge and already
switched away, doing different things. By the time you've realized, you
can't even continue the merge anymore.
This also makes all in-progress commands, am, merge, rebase, revert and
cherry-pick, take all three --abort, --continue and --quit (bisect has a
different UI).
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Octave pattern is almost the same as matlab, except
that '%%%' and '##' can also be used to begin code sections,
in addition to '%%' that is understood by both. Octave
pattern is merged into Matlab pattern. Test cases for
the hunk header patterns of matlab and octave under
t/t4018 are added.
Signed-off-by: Boxuan Li <liboxuan@connect.hku.hk> Signed-off-by: Junio C Hamano <gitster@pobox.com>
If a merge can be fast-forwarded then make sure that we still edit the
commit message if the user specifies -c. The implementation follows the
same pattern that is used for ordinary rewords that are fast-forwarded.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Junio C Hamano <gitster@pobox.com>
I fixed a bug that had broken the reading of sendmail.transferEncoding
in 3494dfd3ee ("send-email: do defaults -> config -> getopt in that
order", 2019-05-09), but the test I added in that commit did nothing
to assert the bug had been fixed.
That issue originates in 8d81408435 ("git-send-email: add
--transfer-encoding option", 2014-11-25) which first added the
"sendemail.transferencoding=8bit".
That test has never done anything meaningful. It tested that the
"--transfer-encoding=8bit" option would turn on the 8bit
Transfer-Encoding, but that was the default at the time (and now). As
checking out 8d81408435 and editing the test to remove that option
will reveal, supplying it never did anything.
So when I copied it thinking it would work in 3494dfd3ee I copied a
previously broken test, although I was making sure it did the right
thing via da-hoc debugger inspection, so the bug was fixed.
So fix the test I added in 3494dfd3ee, as well as the long-standing
test added in 8d81408435. To test if we're actually setting the
Transfer-Encoding let's set it to 7bit, not 8bit, as 7bit will error
out on "email-using-8bit".
This means that we can remove the "sendemail.transferencoding=7bit
fails on 8bit data" test, since it was redundant, we now have other
tests that assert that that'll fail.
While I'm at it convert "git config <key> <value>" in the test setup
to just "-c <key>=<value>" on the command-line. Then we don't need to
cleanup after these tests, and there's no sense in asserting where
config values come from in these tests, we can take that as a given.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
send-email: remove cargo-culted multi-patch pattern in tests
Change test code added in f434c083a0 ("send-email: add --no-cc,
--no-to, and --no-bcc", 2010-03-07) which blindly copied a pattern
from an earlier test added in 32ae83194b ("add a test for
git-send-email for non-threaded mails", 2009-06-12) where the
"$patches" variable was supplied more than once.
As it turns out we didn't need more than one "$patches" for the test
added in 32ae83194b either. The only tests that actually needed this
sort of invocation were the tests added in 54aae5e1a0 ("t9001:
send-email interation with --in-reply-to and --chain-reply-to",
2010-10-19).
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change our default CC_LD_DYNPATH invocation to something GCC likes
these days. Since the GCC 4.6 release unknown flags haven't been
passed through to ld(1). Thus our previous default of CC_LD_DYNPATH=-R
would cause an error on modern GCC unless NO_R_TO_GCC_LINKER was set.
This CC_LD_DYNPATH flag is really obscure, and I don't expect anyone
except those working on git development ever use this.
It's not needed to simply link to libraries like say libpcre,
but *only* for those cases where we're linking to such a library not
present in the OS's library directories. See e.g. ldconfig(8) on Linux
for more details.
I use this to compile my git with a LIBPCREDIR=$HOME/g/pcre2/inst as
I'm building that from source, but someone maintaining an OS package
is almost certainly not going to use this. They're just going to set
USE_LIBPCRE=YesPlease after installing the libpcre dependency,
which'll point to OS libraries which ld(1) will find without the help
of CC_LD_DYNPATH.
Another thing that helps mitigate any potential breakage is that we
detect the right type of invocation in configure.ac, which e.g. HP/UX
uses[1], as does IBM's AIX package[2]. From what I can tell both AIX
and Solaris packagers are building git with GCC, so I'm not adding a
corresponding config.mak.uname default to cater to their OS-native
linkers.
Now for an overview of past development in this area:
Our use of "-R" dates back to 455a7f3275 ("More portability.",
2005-09-30). Soon after that in bbfc63dd78 ("gcc does not necessarily
pass runtime libpath with -R", 2006-12-27) the NO_R_TO_GCC flag was
added, allowing optional use of "-Wl,-rpath=".
Then in f5b904db6b ("Makefile: Allow CC_LD_DYNPATH to be overriden",
2008-08-16) the ability to override this flag to something else
entirely was added, as some linkers use neither "-Wl,-rpath," nor
"-R".
From what I can tell we should, with the benefit of hindsight, have
made this change back in 2006. GCC & ld supported this type of
invocation back then, or since at least binutils-gdb.git's[3] a1ad915dc4 ("[...]Add support for -rpath[...]", 1994-07-20).
Further reading and prior art can be found at [4][5][6][7]. Making a
plain "-R" an error seems from reading those reports to have been
introduced in GCC 4.6 released on March 25, 2011[8], but I couldn't
confirm this with absolute certainty, its release notes are ambiguous
on the subject, and I couldn't be bothered to try to build & bisect it
against GCC 4.5.
This adds xfuncname and word_regex patterns for Rust, a quite
popular programming language. It also includes test cases for the
xfuncname regex (t4018) and updated documentation.
The word_regex pattern finds identifiers, integers, floats and
operators, according to the Rust Reference Book.
Cc: Johannes Sixt <j6t@kdbg.org> Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
In git-format-patch, notes can be appended with the `--notes` option.
However, this must be specified by the user on an
invocation-by-invocation basis. If a user is not careful, it's possible
that they may forget to include it and generate a patch series without
notes.
Teach git-format-patch the `format.notes` config option. Its value is a
notes ref that will be automatically appended. The special value of
"standard" can be used to specify the standard notes. This option is
overridable with the `--no-notes` option in case a user wishes not to
append notes.
Signed-off-by: Denton Liu <liu.denton@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
dup() the input fd for fast-import used for remote helpers
When a remote helper exposes the "import" capability, stdout of the
helper is sent to stdin of a new fast-import process. This is done by
setting the corresponding child_process's in field to the value of the
out field of the helper child_process.
The child_process API is defined to close the file descriptors it's
given when calling start_command. This means when start_command is
called for the fast-import process, its input fd (the output fd of the
helper), is closed.
But when the transport helper is later destroyed, in disconnect_helper,
its input and output are closed, which means close() is called with
an invalid fd (since it was already closed as per above). Or worse, with
a valid fd owned by something else (since fd numbers can be reused).
Signed-off-by: Mike Hommey <mh@glandium.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
pkt-line: drop 'const'-ness of a param to set_packet_header()
The function's definition has a paramter of type "int" qualified as
"const". The fact that the incoming parameter is used as read-only
in the fuction is an implementation detail that the callers should
not have to be told in the prototype declaring it (and "const" there
has no effect, as C passes parameters by value).
The prototype defined for the function in pkt-line.h lacked the
matching "const" for this reason, but apparently some compilers
(e.g. MS Visual C 2017) complain about the parameter type mismatch.
Let's squelch it by removing the "const" that is pointless in the
definition of a small and trivial function like this, which would
not help optimizing compilers nor reading humans that much.
Noticed-by: Johannes Schindelin <johannes.schindelin@gmx.de> Helped-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
blame: optionally track line fingerprints during fill_blame_origin()
fill_blame_origin() is a convenient place to store data that we will use
throughout the lifetime of a blame_origin. Some heuristics for
ignoring commits during a blame session can make use of this storage.
In particular, we will calculate a fingerprint for each line of a file
for blame_origins involved in an ignored commit.
In this commit, we only calculate the line_starts, reusing the existing
code from the scoreboard's line_starts. In an upcoming commit, we will
actually compute the fingerprints.
This feature will be used when we attempt to pass blame entries to
parents when we "ignore" a commit. Most uses of fill_blame_origin()
will not require this feature, hence the flag parameter. Multiple calls
to fill_blame_origin() are idempotent, and any of them can request the
creation of the fingerprints structure.
Suggested-by: Michael Platings <michael@platin.gs> Signed-off-by: Barret Rhoden <brho@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
blame: add config options for the output of ignored or unblamable lines
When ignoring commits, the commit that is blamed might not be
responsible for the change, due to the inaccuracy of our heuristic.
Users might want to know when a particular line has a potentially
inaccurate blame.
Furthermore, guess_line_blames() may fail to find any parent commit for
a given line touched by an ignored commit. Those 'unblamable' lines
remain blamed on an ignored commit. Users might want to know if a line
is unblamable so that they do not spend time investigating a commit they
know is uninteresting.
This patch adds two config options to mark these two types of lines in
the output of blame.
The first option can identify ignored lines by specifying
blame.markIgnoredLines. When this option is set, each blame line that
was blamed on a commit other than the ignored commit is marked with a
'?'.
where the '?' is placed before the commit, and the hash has one fewer
characters.
Sometimes we are unable to even guess at what ancestor commit touched a
line. These lines are 'unblamable.' The second option,
blame.markUnblamableLines, will mark the line with '*'.
For example, say we ignore e5e8d36d04cbe, yet we are unable to blame
this line on another commit: e5e8d36d04cbe (Barret Rhoden 2016-04-11 13:57:54 -0400 26)
appears as:
*e5e8d36d04cb (Barret Rhoden 2016-04-11 13:57:54 -0400 26)
When these config options are used together, every line touched by an
ignored commit will be marked with either a '?' or a '*'.
Signed-off-by: Barret Rhoden <brho@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
blame: add the ability to ignore commits and their changes
Commits that make formatting changes or function renames are often not
interesting when blaming a file. A user may deem such a commit as 'not
interesting' and want to ignore and its changes it when assigning blame.
For example, say a file has the following git history / rev-list:
---O---A---X---B---C---D---Y---E---F
Commits X and Y both touch a particular line, and the other commits do
not:
X: "Take a third parameter"
-MyFunc(1, 2);
+MyFunc(1, 2, 3);
git-blame will blame Y for the change. I'd like to be able to ignore Y:
both the existence of the commit as well as any changes it made. This
differs from -S rev-list, which specifies the list of commits to
process for the blame. We would still process Y, but just don't let the
blame 'stick.'
This patch adds the ability for users to ignore a revision with
--ignore-rev=rev, which may be repeated. They can specify a set of
files of full object names of revs, e.g. SHA-1 hashes, one per line. A
single file may be specified with the blame.ignoreRevFile config option
or with --ignore-rev-file=file. Both the config option and the command
line option may be repeated multiple times. An empty file name "" will
clear the list of revs from previously processed files. Config options
are processed before command line options.
For a typical use case, projects will maintain the file containing
revisions for commits that perform mass reformatting, and their users
have the option to ignore all of the commits in that file.
Additionally, a user can use the --ignore-rev option for one-off
investigation. To go back to the example above, X was a substantive
change to the function, but not the change the user is interested in.
The user inspected X, but wanted to find the previous change to that
line - perhaps a commit that introduced that function call.
To make this work, we can't simply remove all ignored commits from the
rev-list. We need to diff the changes introduced by Y so that we can
ignore them. We let the blames get passed to Y, just like when
processing normally. When Y is the target, we make sure that Y does not
*keep* any blames. Any changes that Y is responsible for get passed to
its parent. Note we make one pass through all of the scapegoats
(parents) to attempt to pass blame normally; we don't know if we *need*
to ignore the commit until we've checked all of the parents.
The blame_entry will get passed up the tree until we find a commit that
has a diff chunk that affects those lines.
One issue is that the ignored commit *did* make some change, and there is
no general solution to finding the line in the parent commit that
corresponds to a given line in the ignored commit. That makes it hard
to attribute a particular line within an ignored commit's diff
correctly.
For example, the parent of an ignored commit has this, say at line 11:
We can pass that blame entry to the parent, but line 11 will be
attributed to commit A, even though "include b.h" came from commit B.
The blame mechanism will be looking at the parent's view of the file at
line number 11.
ignore_blame_entry() is set up to allow alternative algorithms for
guessing per-line blames. Any line that is not attributed to the parent
will continue to be blamed on the ignored commit as if that commit was
not ignored. Upcoming patches have the ability to detect these lines
and mark them in the blame output.
The existing algorithm is simple: blame each line on the corresponding
line in the parent's diff chunk. Any lines beyond that stay with the
target.
For example, the parent of an ignored commit has this, say at line 11:
Commit X nets two additionally lines: 13 and 14. The current
guess_line_blames() algorithm will not attribute these to the parent,
whose diff chunk is only two lines - not four.
When we ignore with the current algorithm, we get:
Note that line 12 was blamed on B, though B was the commit for
new_func_2(), not new_func_1(). Even when guess_line_blames() finds a
line in the parent, it may still be incorrect.
Signed-off-by: Barret Rhoden <brho@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
init_skiplist() took a file consisting of SHA-1s and comments and added
the objects to an oidset. This functionality is useful for other
commands and will be moved to oidset.c in a future commit.
In preparation for that move, this commit renames it to
oidset_parse_file() to reflect its more generic usage and cleans up a
few of the names.
Signed-off-by: Barret Rhoden <brho@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Doc: git.txt: remove backticks from link and add git-scm.com/docs
While checking the html formatted git(1) manual page, it was noted
that the link to https://git.github.io/htmldocs/git.html was formatted
as code. Remove the backticks.
While at it, add the https://git-scm.com/docs link which one reviewer
noted had linkable section headings.
Helped-by: Jeff King <peff@peff.net> Signed-off-by: Philip Oakley <philipoakley@iee.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git.c: show usage for accessing the git(1) help page
It is not immediately obvious how to use the `git help` system to show
the git(1) page, with its overview and its background and coordinating
material, such as environment variables.
Let's simply list it as the last few words of the last usage line.
Signed-off-by: Philip Oakley <philipoakley@iee.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The JGIT prereq uses `type jgit` to determine whether jgit is present.
While this is usually sufficient, it won't help if the jgit found is
badly broken. This wastes time running tests which fail due to no fault
of our own.
Use `jgit --version` instead, to guard against cases where jgit is
present on the system, but will fail to run, e.g. because of some JRE
issue, or missing Java dependencies. Checking that it gets far enough
to process the '--version' argument isn't perfect, but seems to be good
enough in practice. It's also consistent with how we detect some other
dependencies, see e.g. the CURL and UNZIP prerequisites.
Signed-off-by: Todd Zullinger <tmz@pobox.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
find_worktree() can die() unexpectedly because it uses real_path()
instead of the gentler version. When it's used in 'git worktree add' [1]
and there's a bad worktree, this die() could prevent people from adding
new worktrees.
The "bad" condition to trigger this is when a parent of the worktree's
location is deleted. Then real_path() will complain.
Use the other version so that bad worktrees won't affect 'worktree
add'. The bad ones will eventually be pruned, we just have to tolerate
them for a bit.
[1] added in cb56f55c16 (worktree: disallow adding same path multiple
times, 2018-08-28), or since v2.20.0. Though the real bug in
find_worktree() is much older.
Reported-by: Shaheed Haque <shaheedhaque@gmail.com> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When get_oid() and its helpers see an index name like ":.gitmodules",
they try to load the index on demand, like:
if (repo->index->cache)
repo_read_index(repo);
However, that misses the case when "repo->index" itself is NULL; we'll
segfault in the conditional.
This never happens with the_repository; there we always point its index
field to &the_index. But a submodule repository may have a NULL index
field until somebody calls repo_read_index().
This bug is triggered by t7411, but it was hard to notice because it's
in an expect_failure block. That test was added by 2b1257e463 (t/helper:
add test-submodule-nested-repo-config, 2018-10-25). Back then we had no
easy way to access the .gitmodules blob of a submodule repo, so we
expected (and got) an error message to that effect. Later, d9b8b8f896
(submodule-config.c: use repo_get_oid for reading .gitmodules,
2019-04-16) started looking in the correct repo, which is when we
started triggering the segfault.
With this fix, the test starts passing (once we clean it up as its
comment instructs).
Note that as far as I know, this bug could not be triggered outside of
the test suite. It requires resolving an index name in a submodule, and
all of the code paths (aside from test-tool) which do that either load
the index themselves, or always pass the_repository.
Ultimately it comes from 3a7a698e93 (sha1-name.c: remove implicit
dependency on the_index, 2019-01-12), which replaced a check of
"the_index.cache" with "repo->index->cache". So even if there is another
way to trigger it, it wouldn't affect any versions before then.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Do not change the existing info/refs and objects/info/packs
files if they match the existing content on the filesystem.
This is intended to preserve mtime and make it easier for dumb
HTTP pollers to rely on the If-Modified-Since header.
Combined with stdio and kernel buffering; the kernel should be
able to avoid block layer writes and reduce wear for small files.
As a result, the --force option is no longer needed. So stop
documenting it, but let it remain for compatibility (and
debugging, if necessary).
v3: perform incremental comparison while generating to avoid
OOM with giant files. Remove documentation for --force.
Signed-off-by: Eric Wong <e@80x24.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Worktree names are based on $(basename $GIT_WORK_TREE). They aren't
significant until 3a3b9d8cde (refs: new ref types to make per-worktree
refs visible to all worktrees - 2018-10-21), where worktree name could
be part of a refname and must follow refname rules.
Update 'worktree add' code to remove special characters to follow
these rules. In the future the user will be able to specify the
worktree name by themselves if they're not happy with this dumb
character substitution.
Reported-by: Konstantin Kharlamov <hi-angel@yandex.ru> Helped-by: Jeff King <peff@peff.net> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
http-push: prevent format overflow warning with gcc >= 9
In function 'finish_request',
inlined from 'process_response' at http-push.c:248:2:
http-push.c:587:4: warning: '%s' directive argument is null [-Wformat-overflow=]
587 | fprintf(stderr, "Unable to get pack file %s\n%s",
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
588 | request->url, curl_errorstr);
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
request->url is needed for the error message if there was a failure
during fetch but was being cleared unnecessarily earlier.
note that the leak is prevented by calling release_request unconditionally
at the end.
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com> Suggested-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When fetching, the client sends "have" commit IDs indicating that the
server does not need to send any object referenced by those commits,
reducing network I/O. When the client is a partial clone, the client
still sends "have"s in this way, even if it does not have every object
referenced by a commit it sent as "have".
If a server omits such an object, it is fine: the client could lazily
fetch that object before this fetch, and it can still do so after.
The issue is when the server sends a thin pack containing an object that
is a REF_DELTA against such a missing object: index-pack fails to fix
the thin pack. When support for lazily fetching missing objects was
added in 8b4c0103a9 ("sha1_file: support lazily fetching missing
objects", 2017-12-08), support in index-pack was turned off in the
belief that it accesses the repo only to do hash collision checks.
However, this is not true: it also needs to access the repo to resolve
REF_DELTA bases.
Support for lazy fetching should still generally be turned off in
index-pack because it is used as part of the lazy fetching process
itself (if not, infinite loops may occur), but we do need to fetch the
REF_DELTA bases. (When fetching REF_DELTA bases, it is unlikely that
those are REF_DELTA themselves, because we do not send "have" when
making such fetches.)
To resolve this, prefetch all missing REF_DELTA bases before attempting
to resolve them. This both ensures that all bases are attempted to be
fetched, and ensures that we make only one request per index-pack
invocation, and not one request per missing object.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
A subsequent patch will perform the same packfile replacement that is
already done twice, so refactor it into its own function. Also, the same
subsequent patch will use, in another way, part of the packfile
replacement functionality, so extract those out too.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When `rebase -r` finishes it removes any refs under refs/rewritten that
it has created. However if the user aborts or quits the rebase refs are
not removed. This can cause problems for future rebases. For example I
recently wanted to merge a updated version of a topic branch into an
integration branch so ran `rebase -ir` and removed the picks and label
for the topic branch from the todo list so that
merge -C <old-merge> topic
would pick up the new version of topic. Unfortunately
refs/rewritten/topic already existed from a previous rebase that had
been aborted so the rebase just used the old topic, not the new one.
The logic for the non-interactive quit case is changed to ensure
`buf` is always freed.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Junio C Hamano <gitster@pobox.com>
If rebase --quit cannot remove the state directory then it dies. However
when rebase finishes normally or the user runs rebase --abort any errors
that occur when removing the state directory are ignored. That is fixed
by this commit.
All of the callers of finish_rebase() except the code
that handles --abort are careful to make sure they get a postive return
value, do the same for --abort.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the user asks to merge "foo" and we suggest "origin/foo" instead,
we do so by simply chopping off "refs/remotes/" from the front of the
suggested ref. This is usually fine, but it's possible that the
resulting name is ambiguous (e.g., you have "refs/heads/origin/foo",
too).
Let's use shorten_unambiguous_ref() to do this the right way, which
should usually yield the same "origin/foo", but "remotes/origin/foo" if
necessary.
Note that in this situation there may be other options (e.g., we could
suggest "heads/origin/foo" as well). I'll leave that up for debate; the
focus here is just to avoid giving advice that does not actually do what
we expect.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "git merge" sees an unknown refname, we iterate through the refs to
try to suggest some possible alternates. We do so with for_each_ref(),
and in the callback we add some of the refnames we get to a
string_list that is declared with NODUP, directly adding a pointer into
the refname string our callback received.
But the for_each_ref() machinery does not promise that the refname
string will remain valid, and as a result we may print garbage memory.
The code in question dates back to its inception in e56181060e (help:
add help_unknown_ref(), 2013-05-04). But back then, the refname strings
generally did remain stable, at least immediately after the
for_each_ref() call. Later, in d1cf15516f (packed_ref_iterator_begin():
iterate using `mmapped_ref_iterator`, 2017-09-25), we started
consistently re-using a separate buffer for packed refs.
The fix is simple: duplicate the strings we intend to collect. We
already call string_list_clear(), so the memory is correctly freed.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
rebase: fold git-rebase--common into the -p backend
The only remaining scripted part of `git rebase` is the
`--preserve-merges` backend. Meaning: there is little reason to keep the
"library of common rebase functions" as a separate file.
While moving the functions to `git-rebase--preserve-merges.sh`, we also
drop the `move_to_original_branch` function that is no longer used.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since 21853626ea (built-in rebase: call `git am` directly, 2019-01-18),
the built-in rebase already uses the built-in `git am` directly.
Now that d03ebd411c (rebase: remove the rebase.useBuiltin setting,
2019-03-18) even removed the scripted rebase, there is no longer any
user of `git-rebase--am.sh`, so let's just remove it.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The stash.useBuiltin variable introduced in 90a462725e ("stash:
optionally use the scripted version again", 2019-02-25) was turned on by
default, but had no documentation.
Let's document it so that users who run into any stability issues with
the C rewrite know there's an escape hatch, and spell out that the
user should please report the bug when they have to turn off the
built-in stash.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
fast-export: do automatic reencoding of commit messages only if requested
Automatic re-encoding of commit messages (and dropping of the encoding
header) hurts attempts to do reversible history rewrites (e.g. sha1sum
<-> sha256sum transitions, some subtree rewrites), and seems
inconsistent with the general principle followed elsewhere in
fast-export of requiring explicit user requests to modify the output
(e.g. --signed-tags=strip, --tag-of-filtered-object=rewrite). Add a
--reencode flag that the user can use to specify, and like other
fast-export flags, default it to 'abort'.
Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
fast-export: differentiate between explicitly UTF-8 and implicitly UTF-8
The find_encoding() function returned the encoding used by a commit
message, returning a default of git_commit_encoding (usually UTF-8).
Although the current code does not differentiate between a commit which
explicitly requested UTF-8 and one where we just assume UTF-8 because no
encoding is set, it will become important when we try to preserve the
encoding header. Since is_encoding_utf8() returns true when passed
NULL, we can just return NULL from find_encoding() instead of returning
git_commit_encoding.
Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
fast-export: avoid stripping encoding header if we cannot reencode
When fast-export encounters a commit with an 'encoding' header, it tries
to reencode in UTF-8 and then drops the encoding header. However, if it
fails to reencode in UTF-8 because e.g. one of the characters in the
commit message was invalid in the old encoding, then we need to retain
the original encoding or otherwise we lose information needed to
understand all the other (valid) characters in the original commit
message.
Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since git supports commit messages with an encoding other than UTF-8,
allow fast-import to import such commits. This may be useful for folks
who do not want to reencode commit messages from an external system, and
may also be useful to achieve reversible history rewrites (e.g. sha1sum
<-> sha256sum transitions or subtree work) with git repositories that
have used specialized encodings in their commit history.
Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
t9350: fix encoding test to actually test reencoding
This test used an author with non-ascii characters in the name, but no
special commit message. It then grep'ed for those non-ascii characters,
but those are guaranteed to exist regardless of the reencoding process
since the reencoding only affects the commit message, not the author or
committer names. As such, the test would work even if the re-encoding
process simply stripped the commit message entirely. Modify the test to
actually check that the reencoding into UTF-8 worked.
Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
tests: add a special setup where prerequisites fail
As discussed in [1] there's a regression in the "pu" branch now
because a new test implicitly assumed that a previous test guarded by
a prerequisite had been run. Add a "GIT_TEST_FAIL_PREREQS" special
test setup where we'll skip (nearly) all tests guarded by
prerequisites, allowing us to easily emulate those platform where we
don't run these tests.
As noted in the documentation I'm adding I'm whitelisting the SYMLINKS
prerequisite for now. A lot of tests started failing if we lied about
not supporting symlinks. It's also unlikely that we'll have a failing
test due to a hard dependency on symlinks without that being the
obvious cause, so for now it's not worth the effort to make it work.
Update sha1dc from the latest version by the upstream
maintainer[1]. See 07a20f569b ("Makefile: fix unaligned loads in
sha1dc with UBSan", 2019-03-12) for the last update.
This fixes an issue where HP-UX IA64 was wrongly detected as a
Little-endian instead of a Big-endian system, see [2] and [3].
parse-options: adjust `parse_opt_unknown_cb()`s declared return type
In f41179f16ba2 (parse-options: avoid magic return codes, 2019-01-27),
the signature of the low-level parse-opt callback function was changed
to return an `enum`.
And while the implementations were changed, one declaration was left
unchanged, still claiming to return `int`.
This can potentially lead to problems, as compilers are free to choose
any integral type for an `enum` as long as it can represent all declared
values.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git chery-pick" (and "revert" that shares the same runtime engine)
that deals with multiple commits got confused when the final step
gets stopped with a conflict and the user concluded the sequence
with "git commit". Attempt to fix it by cleaning up the state
files used by these commands in such a situation.
* pw/clean-sequencer-state-upon-final-commit:
fix cherry-pick/revert status after commit
commit/reset: try to clean up sequencer state
The internal implementation of "git rebase -i" has been updated to
avoid forking a separate "rebase--interactive" process.
* pw/rebase-i-internal:
rebase -i: run without forking rebase--interactive
rebase: use a common action enum
rebase -i: use struct rebase_options in do_interactive_rebase()
rebase -i: use struct rebase_options to parse args
rebase -i: use struct object_id for squash_onto
rebase -i: use struct commit when parsing options
rebase -i: remove duplication
rebase -i: combine rebase--interactive.c with rebase.c
rebase: use OPT_RERERE_AUTOUPDATE()
rebase: rename write_basic_state()
rebase: don't translate trace strings
sequencer: always discard index after checkout
* dl/rev-tilde-doc-clarify:
revisions.txt: remove ambibuity between <rev>:<path> and :<path>
revisions.txt: mention <rev>~ form
revisions.txt: mark optional rev arguments with []
revisions.txt: change "rev" to "<rev>"
A "ls-files" that emulates "find" to enumerate files in the working
tree resulted in duplicated Makefile rules that caused the build to
issue an unnecessary warning during a trial build after merge
conflicts are resolved in working tree *.h files but before the
resolved results are added to the index. This has been corrected.
* jc/make-dedup-ls-files-output:
Makefile: dedup list of files obtained from ls-files
Mechanically and systematically drop "extern" from function
declarlation.
* dl/no-extern-in-func-decl:
*.[ch]: manually align parameter lists
*.[ch]: remove extern from function declarations using sed
*.[ch]: remove extern from function declarations using spatch
The connectivity bitmaps are created by default in bare
repositories now; also the pathname hash-cache is created by
default to avoid making crappy deltas when repacking.
* ew/repack-with-bitmaps-by-default:
pack-objects: default to writing bitmap hash-cache
t5310: correctly remove bitmaps for jgit test
repack: enable bitmaps by default on bare repos
During an initial "git clone --depth=..." partial clone, it is
pointless to spend cycles for a large portion of the connectivity
check that enumerates and skips promisor objects (which by
definition is all objects fetched from the other side). This has
been optimized out.
* js/partial-clone-connectivity-check:
t/perf: add perf script for partial clones
clone: do faster object check for partial clones
Polishing of the new trace2 facility continues. The system-level
configuration can specify site-wide trace2 settings, which can be
overridden with per-user configuration and environment variables.
* jh/trace2-sid-fix:
trace2: fixup access problem on /etc/gitconfig in read_very_early_config
trace2: update docs to describe system/global config settings
trace2: make SIDs more unique
trace2: clarify UTC datetime formatting
trace2: report peak memory usage of the process
trace2: use system/global config for default trace2 settings
config: add read_very_early_config()
trace2: find exec-dir before trace2 initialization
trace2: add absolute elapsed time to start event
trace2: refactor setting process starting time
config: initialize opts structure in repo_read_config()
'git difftool' falls back to 'git mergetool' config variables when the
difftool equivalents have not been defined.
However, when `diff.guitool` is missing, it doesn't fallback to
anything. Make git-difftool fallback to `merge.guitool` when `diff.guitool` is
missing.
Signed-off-by: Denton Liu <liu.denton@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
difftool: make --gui, --tool and --extcmd mutually exclusive
In git-difftool, these options specify which tool to ultimately run. As
a result, they are logically conflicting. Explicitly disallow these
options from being used together.
Signed-off-by: Denton Liu <liu.denton@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>