fast-export: when using paths, avoid corrupt stream with non-existent mark
If file paths are specified to fast-export and multiple refs point to a
commit that does not touch any of the relevant file paths, then
fast-export can hit problems. fast-export has a list of additional refs
that it needs to explicitly set after exporting all blobs and commits,
and when it tries to get_object_mark() on the relevant commit, it can
get a mark of 0, i.e. "not found", because the commit in question did
not touch the relevant paths and thus was not exported. Trying to
import a stream with a mark corresponding to an unexported object will
cause fast-import to crash.
Avoid this problem by taking the commit the ref points to and finding an
ancestor of it that was exported, and make the ref point to that commit
instead.
Signed-off-by: Elijah Newren <newren@gmail.com> Acked-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
fast-export: avoid dying when filtering by paths and old tags exist
If --tag-of-filtered-object=rewrite is specified along with a set of
paths to limit what is exported, then any tags pointing to old commits
that do not contain any of those specified paths cause problems. Since
the old tagged commit is not exported, fast-export attempts to rewrite
such tags to an ancestor commit which was exported. If no such commit
exists, then fast-export currently die()s. Five years after the tag
rewriting logic was added to fast-export (see commit 2d8ad4691921,
"fast-export: Add a --tag-of-filtered-object option for newly dangling
tags", 2009-06-25), fast-import gained the ability to delete refs (see
commit 4ee1b225b99f, "fast-import: add support to delete refs",
2014-04-20), so now we do have a valid option to rewrite the tag to.
Delete these tags instead of dying.
Signed-off-by: Elijah Newren <newren@gmail.com> Acked-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
ABORT and ERROR happen to have the same value, but come from differnt
enums. Use the one from the correct enum, and while at it, rename the
values to avoid such problems.
Signed-off-by: Elijah Newren <newren@gmail.com> Acked-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Rename anonymize_sha1() to anonymize_oid(() and change its signature,
and switch from sha1_to_hex() to oid_to_hex() and from GIT_SHA1_RAWSZ to
the_hash_algo->rawsz. Also change a comment and a die string to mention
oid instead of sha1.
Signed-off-by: Elijah Newren <newren@gmail.com> Acked-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
bundle: dup() output descriptor closer to point-of-use
When writing a bundle to a file, the bundle code actually creates
"your.bundle.lock" using our lockfile interface. We feed that output
descriptor to a child git-pack-objects via run-command, which has the
quirk that it closes the output descriptor in the parent.
To avoid confusing the lockfile code (which still thinks the descriptor
is valid), we dup() it, and operate on the duplicate.
However, this has a confusing side effect: after the dup() but before we
call pack-objects, we have _two_ descriptors open to the lockfile. If we
call die() during that time, the lockfile code will try to clean up the
partially-written file. It knows to close() the file before unlinking,
since on some platforms (i.e., Windows) the open file would block the
deletion. But it doesn't know about the duplicate descriptor. On
Windows, triggering an error at the right part of the code will result
in the cleanup failing and the lockfile being left in the filesystem.
We can solve this by moving the dup() much closer to start_command(),
shrinking the window in which we have the second descriptor open. It's
easy to place this in such a way that no die() is possible. We could
still die due to a signal in the exact wrong moment, but we already
tolerate races there (e.g., a signal could come before we manage to put
the file on the cleanup list in the first place).
As a bonus, this shields create_bundle() itself from the duplicate-fd
trick, and we can simplify its error handling (note that the lock
rollback now happens unconditionally, but that's OK; it's a noop if we
didn't open the lock in the first place).
The included test uses an empty bundle to cause a failure at the right
spot in the code, because that's easy to trigger (the other likely
errors are write() problems like ENOSPC). Note that it would already
pass on non-Windows systems (because they are happy to unlink an
already-open file).
Based-on-a-patch-by: Gaël Lhez <gael.lhez@gmail.com> Signed-off-by: Jeff King <peff@peff.net> Tested-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
tests: add a special setup where rebase.useBuiltin is off
Add a GIT_TEST_REBASE_USE_BUILTIN=false test mode which is equivalent
to running with rebase.useBuiltin=false. This is needed to spot that
we're not introducing any regressions in the legacy rebase version
while we're carrying both it and the new builtin version.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The rebase.useBuiltin variable introduced in 55071ea248 ("rebase:
start implementing it as a builtin", 2018-08-07) was turned on by
default in 5541bd5b8f ("rebase: default to using the builtin rebase",
2018-08-08), 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[1], and make it clear that
needing to turn off builtin rebase means you've found a bug in git.
mingw: replace an obsolete link with the superseding one
The MSDN documentation has been superseded by Microsoft Docs (which is
backed by a repository on GitHub containing many, many files in Markdown
format).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
On Windows, when we refer to `/an/absolute/path/to/git`, it magically
resolves `git.exe` at that location. Except if something of the name
`git` exists next to that `git.exe`. So if we call `$BUILD_DIR/git`, it
will find `$BUILD_DIR/git.exe` *only* if there is not, say, a directory
called `$BUILD_DIR/git`.
Such a directory, however, exists in Git for Windows when building with
Visual Studio (our Visual Studio project generator defaults to putting
the build files into a directory whose name is the base name of the
corresponding `.exe`).
In the bin-wrappers/* scripts, we already take pains to use `git.exe`
rather than `git`, as this could pick up the wrong thing on Windows
(i.e. if there exists a `git` file or directory in the build directory).
Now we do the same in the tests' start-up code.
This also helps when testing an installed Git, as there might be even
more likely some stray file or directory in the way.
Note: the only way we can record whether the `.exe` suffix is by writing
it to the `GIT-BUILD-OPTIONS` file and sourcing it at the beginning of
`t/test-lib.sh`. This is not a requirement introduced by this patch, but
we move the call to be able to use the `$X` variable that holds the file
extension, if any.
Note also: the many, many calls to `git this` and `git that` are
unaffected, as the regular PATH search will find the `.exe` files on
Windows (and not be confused by a directory of the name `git` that is
in one of the directories listed in the `PATH` variable), while
`/path/to/git` would not, per se, know that it is looking for an
executable and happily prefer such a directory.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
doc: move extensions.worktreeConfig to the right place
All config extensions are described in technical/repository-version.txt.
I made a mistake of adding it in config.txt instead. This patch moves
it back to where it belongs.
Since repository-version.txt is not part of officially generated
documents (it's not even part of DOC_HTML target), it's only visible
to developers who read plain .txt files. Let's include it in
gitrepository-layout.5 for more visibility. Some minor asciidoc fixes
are required in repository-version.txt to make this happen.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
ref-filter: don't look for objects when outside of a repository
The command 'git ls-remote --sort=authordate <remote>' segfaults when
run outside of a repository, ever since the introduction of its
'--sort' option in 1fb20dfd8e (ls-remote: create '--sort' option,
2018-04-09).
While in general the 'git ls-remote' command can be run outside of a
repository just fine, its '--sort=<key>' option with certain keys does
require access to the referenced objects. This sorting is implemented
using the generic ref-filter sorting facility, which already handles
missing objects gracefully with the appropriate 'missing object deadbeef for HEAD' message. However, being generic means that it
checks replace refs while trying to retrieve an object, and while
doing so it accesses the 'git_replace_ref_base' variable, which has
not been initialized and is still a NULL pointer when outside of a
repository, thus causing the segfault.
Make ref-filter more careful upfront while parsing the format string,
and make it error out when encountering a format atom requiring object
access when we are not in a repository. Also add a test to ensure
that 'git ls-remote --sort' fails gracefully when executed outside of
a repository.
Reported-by: H.Merijn Brand <h.m.brand@xs4all.nl> Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Due to limitations in the current implementation, some configuration
variables specified via 'git clone -c var=val' (or 'git -c var=val
clone') are ignored during the initial fetch and checkout.
Let the users know which configuration variables are known to be
ignored ('remote.origin.mirror' and 'remote.origin.tagOpt') under the
documentation of 'git clone -c', along with hints to use the options
'--mirror' and '--no-tags' instead.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
clone: respect additional configured fetch refspecs during initial fetch
The initial fetch during a clone doesn't transfer refs matching
additional fetch refspecs given on the command line as configuration
variables, e.g. '-c remote.origin.fetch=<refspec>'. This contradicts
the documentation stating that configuration variables specified via
'git clone -c <key>=<value> ...' "take effect immediately after the
repository is initialized, but before the remote history is fetched"
and the given example specifically mentions "adding additional fetch
refspecs to the origin remote". Furthermore, one-shot configuration
variables specified via 'git -c <key>=<value> clone ...', though not
written to the newly created repository's config file, live during the
lifetime of the 'clone' command, including the initial fetch. All
this implies that any fetch refspecs specified this way should already
be taken into account during the initial fetch.
The reason for this is that the initial fetch is not a fully fledged
'git fetch' but a bunch of direct calls into the fetch/transport
machinery with clone's own refs-to-refspec matching logic, which
bypasses parts of 'git fetch' processing configured fetch refspecs.
This logic only considers a single default refspec, potentially
influenced by options like '--single-branch' and '--mirror'. The
configured refspecs are, however, already read and parsed properly
when clone calls remote.c:remote_get(), but it never looks at the
parsed refspecs in the resulting 'struct remote'.
Modify clone to take the remote's configured fetch refspecs into
account to retrieve all matching refs during the initial fetch. Note
that we have to explicitly add the default fetch refspec to the
remote's refspecs, because at that point the remote only includes the
fetch refspecs specified on the command line.
Add tests to check that refspecs given both via 'git clone -c ...' and
'git -c ... clone' retrieve all refs matching either the default or
the additional refspecs, and that it works even when the user
specifies an alternative remote name via '--origin=<name>'.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
clone: use a more appropriate variable name for the default refspec
cmd_clone() declares two strbufs 'key' and 'value' on the same line,
suggesting that they are used to contruct a config variable's name and
value. However, this is not the case: 'key' is used to construct the
names of multiple config variables, while 'value' is never used as a
value for any of those config variables, or for any other config
variable for that matter, but only to contruct the default fetch
refspec.
Let's rename 'value' to 'default_refspec' to make the intent clearer.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
config: report a bug if git_dir exists without commondir
This did happen at some stage, and was fixed relatively quickly. Make
sure that we detect very quickly, too, should that happen again.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
rebase: validate -C<n> and --whitespace=<mode> parameters early
It is a good idea to error out early upon seeing, say, `-Cbad`, rather
than starting the rebase only to have the `--am` backend complain later.
Let's do this.
The only options accepting parameters which we pass through to `git am`
(which may, or may not, forward them to `git apply`) are `-C` and
`--whitespace`. The other options we pass through do not accept
parameters, so we do not have to validate them here.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Currently, we parse the options intended for `git am` as if we wanted to
handle them in `git rebase`, and then reconstruct them painstakingly to
define the `git_am_opt` variable.
However, there is a much better way (that I was unaware of, at the time
when I mentored Pratik to implement these options): OPT_PASSTHRU_ARGV.
It is intended for exactly this use case, where command-line options
want to be parsed into a separate `argv_array`.
Let's use this feature.
Incidentally, this also allows us to address a bug discovered by Phillip
Wood, where the built-in rebase failed to understand that the `-C`
option takes an optional argument.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
object: parse_object to honor its repository argument
In 8e4b0b6047 (object.c: allow parse_object to handle
arbitrary repositories, 2018-06-28), we forgot to pass the
repository down to the read_object_file.
Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
object-store: prepare read_object_file to deal with any repo
As read_object_file is a widely used function (which is also regularly used
in new code in flight between master..pu), changing its signature is painful
is hard, as other series in flight rely on the original signature. It would
burden the maintainer if we'd just change the signature.
Introduce repo_read_object_file which takes the repository argument, and
hide the original read_object_file as a macro behind
NO_THE_REPOSITORY_COMPATIBILITY_MACROS, similar to e675765235 (diff.c: remove implicit dependency on the_index, 2018-09-21)
Add a coccinelle patch to convert existing callers, but do not apply
the resulting patch to keep the diff of this patch small.
Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
push: change needlessly ambiguous example in error
Change an example push added in b55e677522 ("push: introduce new
push.default mode "simple"", 2012-04-24) to always mean the same thing
whether the current setting happens to be "simple" or not.
This error is only emitted under "simple", but message is explaining
to the user that they can get two sorts of different behaviors by
these two invocations.
Let's use "git push <remote> HEAD" which always means push the current
branch name to that remote, instead of "git push <remote>
<current-branch-name>" which will do that under "simple", but is not
guaranteed to do under "upstream".
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
sha256: add an SHA-256 implementation using libgcrypt
Generally, one gets better performance out of cryptographic routines
written in assembly than C, and this is also true for SHA-256. In
addition, most Linux distributions cannot distribute Git linked against
OpenSSL for licensing reasons.
Most systems with GnuPG will also have libgcrypt, since it is a
dependency of GnuPG. libgcrypt is also faster than the SHA1DC
implementation for messages of a few KiB and larger.
For comparison, on a Core i7-6600U, this implementation processes 16 KiB
chunks at 355 MiB/s while SHA1DC processes equivalent chunks at 337
MiB/s.
In addition, libgcrypt is licensed under the LGPL 2.1, which is
compatible with the GPL. Add an implementation of SHA-256 that uses
libgcrypt.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
SHA-1 is weak and we need to transition to a new hash function. For
some time, we have referred to this new function as NewHash. Recently,
we decided to pick SHA-256 as NewHash. The reasons behind the choice of
SHA-256 are outlined in the thread starting at [1] and in the commit
history for the hash function transition document.
Add a basic implementation of SHA-256 based off libtomcrypt, which is in
the public domain. Optimize it and restructure it to meet our coding
standards. Pull in the update and final functions from the SHA-1 block
implementation, as we know these function correctly with all compilers.
This implementation is slower than SHA-1, but more performant
implementations will be introduced in future commits.
Wire up SHA-256 in the list of hash algorithms, and add a test that the
algorithm works correctly.
Note that with this patch, it is still not possible to switch to using
SHA-256 in Git. Additional patches are needed to prepare the code to
handle a larger hash algorithm and further test fixes are needed.
Instead of using hard-coded constants for object sizes, use
the_hash_algo to look them up. In addition, use a function call to look
up the object ID version and produce the correct value. For now, we use
version 1, which means to use the default algorithm used in the rest of
the repository.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a utility (which is less for the testsuite and more for developers)
that can compute hash speeds for whatever hash algorithms are
implemented. This allows developers to test their personal systems to
determine the performance characteristics of various algorithms.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
There is one place we need the hash algorithm block size: the HMAC code
for push certs. Expose this constant in struct git_hash_algo and expose
values for SHA-1 and for the largest value of any hash.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since we're going to have multiple hash algorithms to test, it makes
sense to share as much of the test code as possible. Convert the sha1
helper for the test-tool to be generic and move it out into its own
module. This will allow us to share most of this code with our NewHash
implementation.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
We have in the past had some unfortunate endianness issues with some
SHA-1 implementations we ship, especially on big-endian machines. Add
an explicit test using the test helper to catch these issues and point
them out prominently. This test can also be used as a staging ground
for people testing additional algorithms to verify that their
implementations are working as expected.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
cache: make hashcmp and hasheq work with larger hashes
In 183a638b7d ("hashcmp: assert constant hash size", 2018-08-23), we
modified hashcmp to assert that the hash size was always 20 to help it
optimize and inline calls to memcmp. In a future series, we replaced
many calls to hashcmp and oidcmp with calls to hasheq and oideq to
improve inlining further.
However, we want to support hash algorithms other than SHA-1, namely
SHA-256. When doing so, we must handle the case where these values are
32 bytes long as well as 20. Adjust hashcmp to handle two cases:
20-byte matches, and maximum-size matches. Therefore, when we include
SHA-256, we'll automatically handle it properly, while at the same time
teaching the compiler that there are only two possible options to
consider. This will allow the compiler to write the most efficient
possible code.
Copy similar code into hasheq and perform an identical transformation.
At least with GCC 8.2.0, making hasheq defer to hashcmp when there are
two branches prevents the compiler from inlining the comparison, while
the code in this patch is inlined properly. Add a comment to avoid an
accidental performance regression from well-intentioned refactoring.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
hex: introduce functions to print arbitrary hashes
Currently, we have functions that turn an arbitrary SHA-1 value or an
object ID into hex format, either using a static buffer or with a
user-provided buffer. Add variants of these functions that can handle
an arbitrary hash algorithm, specified by constant. Update the
documentation as well.
While we're at it, remove the "extern" declaration from this family of
functions, since it's not needed and our style now recommends against
it.
We use the variant taking the algorithm structure pointer as the
internal variant, since taking an algorithm pointer is the easiest way
to handle all of the variants in use.
Note that we maintain these functions because there are hashes which
must change based on the hash algorithm in use but are not object IDs
(such as pack checksums).
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
diff: align move detection error handling with other options
This changes the error handling for the options --color-moved-ws
and --color-moved-ws to be like the rest of the options.
Move the die() call out of parse_color_moved_ws into the parsing
of command line options. As the function returns a bit field, change
its signature to return an unsigned instead of an int; add a new bit
to signal errors. Once the error is signaled, we discard the other
bits, such that it doesn't matter if the error bit overlaps with any
other bit.
Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
push doc: document the DWYM behavior pushing to unqualified <dst>
Document the DWYM behavior that kicks in when pushing to an
unqualified <dst> reference.
This behavior was added in f88395ac23 ("Renaming push.", 2005-08-03)
and f8aae12034 ("push: allow unqualified dest refspecs to DWIM",
2008-04-23), and somewhat documented in bb9fca80ce ("git-push: Update
description of refspecs and add examples", 2007-06-09), but has never
been fully documented.
The closest we got to having documented it was the description in the
commit message for f8aae12034, which I've borrowed from in writing
this documentation.
Let's also refer to this new documentation from the existing
documentation we had (added in bb9fca80ce).
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
push: test that <src> doesn't DWYM if <dst> is unqualified
Add a test asserting that "git push origin <src>:<dst>" where <src> is
a branch, tag, tree or blob in refs/remotes/* doesn't DWYM when <dst>
is unqualified. This has never been the case, but there haven't been
any tests for this behavior.
See f88395ac23 ("Renaming push.", 2005-08-03), bb9fca80ce ("git-push:
Update description of refspecs and add examples", 2007-06-09) and f8aae12034 ("push: allow unqualified dest refspecs to DWIM",
2008-04-23) which are most relevant commits that have changed or
documented the behavior of the DWYM feature in the past.
These tests were originally meant to lead up to a patch that made
refs/remotes/* on the LHS imply refs/heads/* on the RHS, see [1]. That
patch proved controversial and may not ever land in git.git, but we
should have the tests that remind us what the current behavior is in
case it's ever changed.
Add an advice to the recently improved error message added in f8aae12034 ("push: allow unqualified dest refspecs to DWIM",
2008-04-23).
Now with advice.pushUnqualifiedRefName=true (on by default) we show a
hint about how to proceed:
$ ./git-push avar v2.19.0^{commit}:newbranch -n
error: The destination you provided is not a full refname (i.e.,
starting with "refs/"). We tried to guess what you meant by:
- Looking for a ref that matches 'newbranch' on the remote side.
- Checking if the <src> being pushed ('v2.19.0^{commit}')
is a ref in "refs/{heads,tags}/". If so we add a corresponding
refs/{heads,tags}/ prefix on the remote side.
Neither worked, so we gave up. You must fully qualify the ref.
hint: The <src> part of the refspec is a commit object.
hint: Did you mean to create a new branch by pushing to
hint: 'v2.19.0^{commit}:refs/heads/newbranch'?
error: failed to push some refs to 'git@github.com:avar/git.git'
When trying to push a tag, tree or a blob we suggest that perhaps the
user meant to push them to refs/tags/ instead.
The if/else duplication for all of OBJ_{COMMIT,TAG,TREE,BLOB} is
unfortunate, but is required to correctly mark the messages for
translation. See the discussion in
<87r2gxebsi.fsf@evledraar.gmail.com> about that.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
push: move unqualified refname error into a function
A follow-up change will extend this error message with the advice
facility. Doing so would make the indentation too deeply nested for
comfort. So let's split this into a helper function.
There's no changes to the wording here. Just code moving &
re-indentation, and re-flowing the "TRANSLATORS" comment.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
push: improve the error shown on unqualified <dst> push
Improve the error message added in f8aae12034 ("push: allow
unqualified dest refspecs to DWIM", 2008-04-23), which before this
change looks like this:
$ git push avar v2.19.0^{commit}:newbranch -n
error: unable to push to unqualified destination: newbranch
The destination refspec neither matches an existing ref on the remote nor
begins with refs/, and we are unable to guess a prefix based on the source ref.
error: failed to push some refs to 'git@github.com:avar/git.git'
This message needed to be read very carefully to spot how to fix the
error, i.e. to push to refs/heads/newbranch. Now the message will look
like this instead:
$ ./git-push avar v2.19.0^{commit}:newbranch -n
error: The destination you provided is not a full refname (i.e.,
starting with "refs/"). We tried to guess what you meant by:
- Looking for a ref that matches 'newbranch' on the remote side.
- Checking if the <src> being pushed ('v2.19.0^{commit}')
is a ref in "refs/{heads,tags}/". If so we add a corresponding
refs/{heads,tags}/ prefix on the remote side.
Neither worked, so we gave up. You must fully qualify the ref.
error: failed to push some refs to 'git@github.com:avar/git.git'
This improvement is the result of on-list discussion in [1] and [2],
as well as my own fixes / reformatting / phrasing on top.
The suggestion by Jeff on-list was to make that second bullet point
"Looking at the refname of the local source.". The version being added
here is more verbose, but also more accurate. saying "local source"
could refer to any ref in the local refstore, including something in
refs/remotes/*. A later change will teach guess_ref() to infer a ref
type from remote-tracking refs, so let's not confuse the two.
While I'm at it, add a "TRANSLATORS" comment since the message has
gotten more complex and it's not as clear what the two %s's refer to.
i18n: remote.c: mark error(...) messages for translation
Mark up the error(...) messages in remote.c for translation. The likes
of "unable to push to unqualified destination" are relatively big
parts of the UI, i.e. error messages shown when "git push" fails.
I don't think any of these are plumbing, an the entire test suite
passes when running the tests with GIT_GETTEXT_POISON=1 (after
building with GETTEXT_POISON).
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
remote.c: add braces in anticipation of a follow-up change
The CodingGuidelines say "When there are multiple arms to a
conditional and some of them require braces, enclose even a single
line block in braces for consistency.". Fix the code in
match_explicit() to conform.
While I'm at it change the if/else if/else in guess_ref() to use
braces. This is not currently needed, but a follow-up change will add
a new multi-line condition to that logic.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
range-diff: make diff option behavior (e.g. --stat) consistent
Make the behavior when diff options (e.g. "--stat") are passed
consistent with how "diff" behaves.
Before 73a834e9e2 ("range-diff: relieve callers of low-level
configuration burden", 2018-07-22) running range-diff with "--stat"
would produce stat output and the diff output, as opposed to how
"diff" behaves where once "--stat" is specified "--patch" also needs
to be provided to emit the patch output.
As noted in a previous change ("range-diff doc: add a section about
output stability", 2018-11-07) the "--stat" output with "range-diff"
is useless at the moment.
But we should behave consistently with "diff" in anticipation of such
output being useful in the future, because it would make for confusing
UI if "diff" and "range-diff" behaved differently when it came to how
they interpret diff options.
The new behavior is also consistent with the existing documentation
added in ba931edd28 ("range-diff: populate the man page",
2018-08-13). See "[...]also accepts the regular diff options[...]" in
git-range-diff(1).
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
win32: replace pthread_cond_*() with much simpler code
The Win32 CONDITION_VARIABLE has better performance and is easier to
maintain, as the code is a lot shorter now (the semantics of the
CONDITION_VARIABLE matches the pthread_cond_t very well).
Note: CONDITION_VARIABLE is not available in Windows XP and below,
but the declared minimal Windows version required to build and run
Git for Windows is Windows Vista (which is also beyond its
end-of-life, but for less long than Windows XP), so that's okay.
Signed-off-by: Loo Rong Jie <loorongjie@gmail.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
One of the problems with "git checkout" is that it does so many
different things and could confuse people specially when we fail to
handle ambiguation correctly.
One way to help with that is tell the user what sort of operation is
actually carried out. When switching branches, we always print
something unless --quiet, either
- "HEAD is now at ..."
- "Reset branch ..."
- "Already on ..."
- "Switched to and reset ..."
- "Switched to a new branch ..."
- "Switched to branch ..."
Checking out paths however is silent. Print something so that if we
got the user intention wrong, they won't waste too much time to find
that out. For the remaining cases of checkout we now print either
- "Checked out ... paths out of the index"
- "Checked out ... paths out of <abbrev hash>"
Since the purpose of printing this is to help disambiguate. Only do it
when "--" is missing.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
checkout: disambiguate dwim tracking branches and local files
When checkout dwim is added in [1], it is restricted to only dwim when
certain conditions are met and fall back to default checkout behavior
otherwise. It turns out falling back could be confusing. One of the
conditions to turn
git checkout frotz
to
git checkout -b frotz origin/frotz
is that frotz must not exist as a file. But when the user comes to
expect "git checkout frotz" to create the branch "frotz" and there
happens to be a file named "frotz", git's silently reverting "frotz"
file content is not helping. This is reported in Git mailing list [2]
and even used as an example of "Git is bad" elsewhere [3].
We normally try to do the right thing, but when there are multiple
"right things" to do, it's best to leave it to the user to decide.
Check this case, ask the user to to disambiguate:
- "git checkout -- foo" will check out path "foo"
- "git checkout foo --" will dwim and create branch "foo" [4]
For users who do not want dwim, use --no-guess. It's useless in this
particular case because "git checkout --no-guess foo --" will just
fail. But it could be used by scripts.
The function `CreateHardLink()` is available in all supported Windows
versions (even since Windows XP), so there is no more need to resolve it
at runtime.
Helped-by: Max Kirillov <max@max630.net> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
tests: respect GIT_TEST_INSTALLED when initializing repositories
It really makes very, very little sense to use a different git
executable than the one the caller indicated via setting the environment
variable GIT_TEST_INSTALLED.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Teach `make coccicheck` to avoid patches named "*.pending.cocci" and
handle them separately in a new `make coccicheck-pending` instead.
This means that we can separate "critical" patches from "FYI" patches.
The former target can continue causing Travis to fail its static
analysis job, while the latter can let us keep an eye on ongoing
(pending) transitions without them causing too much fallout.
Document the intended use-cases around these two targets.
As the process around the pending patches is not yet fully explored,
leave that out.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Based-on-work-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The procedure to install dependencies before testing at Travis CI
is getting revamped for both simplicity and flexibility, taking
advantage of the recent move to the vm-based environment.
* sg/travis-install-dependencies:
travis-ci: install packages in 'ci/install-dependencies.sh'
"git add" needs to internally run "diff-files" equivalent, and the
codepath learned the same optimization as "diff-files" has to run
lstat(2) in parallel to find which paths have been updated in the
working tree.
* bp/add-diff-files-optim:
add: speed up cmd_add() by utilizing read_cache_preload()
The interface into "xdiff" library used to discover the offset and
size of a generated patch hunk by first formatting it into the
textual hunk header "@@ -n,m +k,l @@" and then parsing the numbers
out. A new interface has been introduced to allow callers a more
direct access to them.
* jk/xdiff-interface:
xdiff-interface: drop parse_hunk_header()
range-diff: use a hunk callback
diff: convert --check to use a hunk callback
combine-diff: use an xdiff hunk callback
diff: use hunk callback for word-diff
diff: discard hunk headers for patch-ids earlier
diff: avoid generating unused hunk header lines
xdiff-interface: provide a separate consume callback for hunks
xdiff: provide a separate emit callback for hunks
The code to traverse objects for reachability, used to decide what
objects are unreferenced and expendable, have been taught to also
consider per-worktree refs of other worktrees as starting points to
prevent data loss.
* nd/per-worktree-ref-iteration:
git-worktree.txt: correct linkgit command name
reflog expire: cover reflog from all worktrees
fsck: check HEAD and reflog from other worktrees
fsck: move fsck_head_link() to get_default_heads() to avoid some globals
revision.c: better error reporting on ref from different worktrees
revision.c: correct a parameter name
refs: new ref types to make per-worktree refs visible to all worktrees
Add a place for (not) sharing stuff between worktrees
refs.c: indent with tabs, not spaces
Assorted fixes for bugs found while auditing -Wunused-parameter
warnings.
* jk/misc-unused-fixes:
approxidate: fix NULL dereference in date_time()
pathspec: handle non-terminated strings with :(attr)
approxidate: handle pending number for "specials"
rev-list: handle flags for --indexed-objects
The helper function to refresh the cached stat information in the
in-core index has learned to perform the lstat() part of the
operation in parallel on multi-core platforms.
* bp/refresh-index-using-preload:
refresh_index: remove unnecessary calls to preload_index()
speed up refresh_index() by utilizing preload_index()
"git send-email --transfer-encoding=..." in recent versions of Git
sometimes produced an empty "Content-Transfer-Encoding:" header,
which has been corrected.
* al/send-email-auto-cte-fixup:
send-email: avoid empty transfer encoding header
In preparation to the day when we can deprecate and remove the
"rebase -p", make sure we can skip and later remove tests for
it.
* js/rebase-p-tests:
tests: optionally skip `git rebase -p` tests
t3418: decouple test cases from a previous `rebase -p` test case
t3404: decouple some test cases from outcomes of previous test cases
"git rev-parse --exclude=* --branches --branches" (i.e. first
saying "add only things that do not match '*' out of all branches"
and then adding all branches, without any exclusion this time")
worked as expected, but "--exclude=* --all --all" did not work the
same way, which has been fixed.
* ag/rev-parse-all-exclude-fix:
rev-parse: clear --exclude list after 'git rev-parse --all'
The submodule support has been updated to read from the blob at
HEAD:.gitmodules when the .gitmodules file is missing from the
working tree.
* ao/submodule-wo-gitmodules-checked-out:
t/helper: add test-submodule-nested-repo-config
submodule: support reading .gitmodules when it's not in the working tree
submodule: add a helper to check if it is safe to write to .gitmodules
t7506: clean up .gitmodules properly before setting up new scenario
submodule: use the 'submodule--helper config' command
submodule--helper: add a new 'config' subcommand
t7411: be nicer to future tests and really clean things up
t7411: merge tests 5 and 6
submodule: factor out a config_set_in_gitmodules_file_gently function
submodule: add a print_config_from_gitmodules() helper
* nb/worktree-api-doc:
worktree: rename is_worktree_locked to worktree_lock_reason
worktree: update documentation for lock_reason and lock_reason_valid
A couple of tests used to leave the repository in a state that is
deliberately corrupt, which have been corrected.
* ab/pack-tests-cleanup:
index-pack tests: don't leave test repo dirty at end
pack-objects tests: don't leave test .git corrupt at end
pack-objects test: modernize style
Tests for the recently introduced multi-pack index machinery.
* ds/test-multi-pack-index:
packfile: close multi-pack-index in close_all_packs
multi-pack-index: define GIT_TEST_MULTI_PACK_INDEX
midx: close multi-pack-index on repack
midx: fix broken free() in close_midx()
A pattern with '**' that does not have a slash on either side used
to be an invalid one, but the code now treats such double-asterisks
the same way as two normal asterisks that happen to be adjacent to
each other.
* nd/wildmatch-double-asterisk:
wildmatch: change behavior of "foo**bar" in WM_PATHNAME mode
A fourth class of configuration files (in addition to the
traditional "system wide", "per user in the $HOME directory" and
"per repository in the $GIT_DIR/config") has been introduced so
that different worktrees that share the same repository (hence the
same $GIT_DIR/config file) can use different customization.
* nd/per-worktree-config:
worktree: add per-worktree config files
t1300: extract and use test_cmp_config()
"git ls-remote $there foo" was broken by recent update for the
protocol v2 and stopped showing refs that match 'foo' that are not
refs/{heads,tags}/foo, which has been fixed.
* jk/proto-v2-ref-prefix-fix:
ls-remote: pass heads/tags prefixes to transport
ls-remote: do not send ref prefixes for patterns
Split the overly large Documentation/config.txt file into million
little pieces. This potentially allows each individual piece
included into the manual page of the command it affects more easily.
* nd/config-split: (81 commits)
config.txt: remove config/dummy.txt
config.txt: move worktree.* to a separate file
config.txt: move web.* to a separate file
config.txt: move versionsort.* to a separate file
config.txt: move user.* to a separate file
config.txt: move url.* to a separate file
config.txt: move uploadpack.* to a separate file
config.txt: move uploadarchive.* to a separate file
config.txt: move transfer.* to a separate file
config.txt: move tag.* to a separate file
config.txt: move submodule.* to a separate file
config.txt: move stash.* to a separate file
config.txt: move status.* to a separate file
config.txt: move splitIndex.* to a separate file
config.txt: move showBranch.* to a separate file
config.txt: move sequencer.* to a separate file
config.txt: move sendemail-config.txt to config/
config.txt: move reset.* to a separate file
config.txt: move rerere.* to a separate file
config.txt: move repack.* to a separate file
...
format-patch: respect --stat in cover letter's diffstat
Commit 43662b23ab (format-patch: keep cover-letter diffstat wrapped in
72 columns - 2018-01-24) uncondtionally sets stat width to 72 when
generating diffstat for the cover letter, ignoring --stat from command
line. But it should only do so when stat width is still default
(i.e. stat_width == 0).
In order to fix this, we should only set stat_width if stat_width is
zero. But it will never be. Commit 071dd0ba43 (format-patch: reduce
patch diffstat width to 72 - 2018-02-01) makes sure that default stat
width will be 72 (ignoring $COLUMNS, but could still be overriden by
--stat). So all we need to do here is drop the assignment.
Reported-by: Laszlo Ersek <lersek@redhat.com> Helped-by: Leif Lindholm <leif.lindholm@linaro.org> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
built-in rebase: reinstate `checkout -q` behavior where appropriate
When we converted a `git checkout -q $onto^0` call to use
`reset_head()`, we inadvertently incurred a change from a twoway_merge
to a oneway_merge, as if we wanted a `git reset --hard` instead.
This has performance ramifications under certain, though, as the
oneway_merge needs to lstat() every single index entry whereas
twoway_merge does not.
So let's go back to the old behavior.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>