"git diff" had a confusion between taking data from a path in the
working tree and taking data from an object that happens to have
name 0{40} recorded in a tree.
* jk/maint-null-in-trees:
fsck: detect null sha1 in tree entries
do not write null sha1s to on-disk index
diff: do not use null sha1 as a sentinel value
"git send-email" did not unquote encoded words that appear on the
header correctly, and lost "_" from strings.
* tr/maint-send-email-2047:
send-email: improve RFC2047 quote parsing
The built-in "binary" attribute macro expands to "-diff -text", so
that textual diff is not produced, and the contents will not go
through any CR/LF conversion ever. During a merge, it should also
choose the "binary" low-level merge driver, but it didn't.
Make it expand to "-diff -merge -text".
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The (discouraged) -Xours/-Xtheirs modes of merge are supposed to
give a quick and dirty way to come up with a random mixture of
cleanly merged parts and punted conflict resolution to take contents
from one side in conflicting parts. These options however were only
passed down to the low level merge driver for text.
Teach the built-in binary merge driver to notice them as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When fetch is invoked with --all, we need to pass the tag-following
preference to each individual fetch; without this, we will always
auto-follow tags, preventing us from fetching the remote tags into a
remote-specific namespace, for example.
Reported-by: Oswald Buddenhagen <ossi@kde.org>
Signed-off-by: Dan Johnson <ComputerDruid@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
All remote subcommands are spelled out words except 'rm'. 'rm', being a
popular UNIX command name, may mislead users that there are also 'ls' or
'mv'. Use 'remove' to fit with the rest of subcommands.
'rm' is still supported and used in the test suite. It's just not
widely advertised.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When giving multiple individual revisions to cherry-pick or revert, as
in 'git cherry-pick A B' or 'git revert B A', one would expect them to
be picked/reverted in the order given on the command line. They are
instead ordered by their commit timestamp -- in chronological order
for "cherry-pick" and in reverse chronological order for
"revert". This matches the order in which one would usually give them
on the command line, making this bug somewhat hard to notice. Still,
it has been reported at least once before [1].
It seems like the chronological sorting happened by accident because
the revision walker has traditionally always sorted commits in reverse
chronological order when rev_info.no_walk was enabled. In the case of
'git revert B A' where B is newer than A, this sorting is a no-op. For
'git cherry-pick A B', the sorting would reverse the arguments, but
because the sequencer also flips the rev_info.reverse flag when
picking (as opposed to reverting), the end result is a chronological
order. The rev_info.reverse flag was probably flipped so that the
revision walker emits B before C in 'git cherry-pick A..C'; that it
happened to effectively undo the unexpected sorting done when not
walking, was probably a coincidence that allowed this bug to happen at
all.
Fix the bug by telling the revision walker not to sort the commits
when not walking. The only case we want to reverse the order is now
when cherry-picking and walking revisions (rev_info.no_walk = 0).
[1] http://thread.gmane.org/gmane.comp.version-control.git/164794
Signed-off-by: Martin von Zweigbergk <martinvonz@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Cherry-picking commits out of order (w.r.t. commit time stamp) doesn't
currently work. Add a test case to demonstrate it.
Signed-off-by: Martin von Zweigbergk <martinvonz@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When 'git log' is passed the --no-walk option, no revision walk takes
place, naturally. Perhaps somewhat surprisingly, however, the provided
revisions still get sorted by commit date. So e.g 'git log --no-walk
HEAD HEAD~1' and 'git log --no-walk HEAD~1 HEAD' give the same result
(unless the two revisions share the commit date, in which case they
will retain the order given on the command line). As the commit that
introduced --no-walk (8e64006 (Teach revision machinery about
--no-walk, 2007-07-24)) points out, the sorting is intentional, to
allow things like
git log --abbrev-commit --pretty=oneline --decorate --all --no-walk
to show all refs in order by commit date.
But there are also other cases where the sorting is not wanted, such
as
<command producing revisions in order> |
git log --oneline --no-walk --stdin
To accomodate both cases, leave the decision of whether or not to sort
up to the caller, by allowing --no-walk={sorted,unsorted}, defaulting
to 'sorted' for backward-compatibility reasons.
Signed-off-by: Martin von Zweigbergk <martinvonz@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Originally the "--quiet" option was parsed by the
diff-option parser into the internal QUICK option. This had
the effect of silencing diff output from the log (which was
not intended, but happened to work and people started to
use it). But it also had other odd side effects at the diff
level (for example, it would suppress the second commit in
"git show A B").
To fix this, commit 1c40c36 converted log to parse-options
and handled the "quiet" option separately, not passing it
on to the diff code. However, it simply ignored the option,
which was a regression for people using it as a synonym for
"-s". Commit 01771a8 then fixed that by interpreting the
option to add DIFF_FORMAT_NO_OUTPUT to the list of output
formats.
However, that commit did not fix it in all cases. It sets
the flag after setup_revisions is called. Naively, this
makes sense because you would expect the setup_revisions
parser to overwrite our output format flag if "-p" or
another output format flag is seen.
However, that is not how the NO_OUTPUT flag works. We
actually store it in the bit-field as just another format.
At the end of setup_revisions, we call diff_setup_done,
which post-processes the bitfield and clears any other
formats if we have set NO_OUTPUT. By setting the flag after
setup_revisions is done, diff_setup_done does not have a
chance to make this tweak, and we end up with other format
options still set.
As a result, the flag would have no effect in "git log -p
--quiet" or "git show --quiet". Fix it by setting the
format flag before the call to setup_revisions.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
All of the smart-http GET requests go through the http_get_*
functions, which will prompt for credentials and retry if we
see an HTTP 401.
POST requests, however, do not go through any central point.
Moreover, it is difficult to retry in the general case; we
cannot assume the request body fits in memory or is even
seekable, and we don't know how much of it was consumed
during the attempt.
Most of the time, this is not a big deal; for both fetching
and pushing, we make a GET request before doing any POSTs,
so typically we figure out the credentials during the first
request, then reuse them during the POST. However, some
servers may allow a client to get the list of refs from
receive-pack without authentication, and then require
authentication when the client actually tries to POST the
pack.
This is not ideal, as the client may do a non-trivial amount
of work to generate the pack (e.g., delta-compressing
objects). However, for a long time it has been the
recommended example configuration in git-http-backend(1) for
setting up a repository with anonymous fetch and
authenticated push. This setup has always been broken
without putting a username into the URL. Prior to commit
986bbc0, it did work with a username in the URL, because git
would prompt for credentials before making any requests at
all. However, post-986bbc0, it is totally broken. Since it
has been advertised in the manpage for some time, we should
make sure it works.
Unfortunately, it is not as easy as simply calling post_rpc
again when it fails, due to the input issue mentioned above.
However, we can still make this specific case work by
retrying in two specific instances:
1. If the request is large (bigger than LARGE_PACKET_MAX),
we will first send a probe request with a single flush
packet. Since this request is static, we can freely
retry it.
2. If the request is small and we are not using gzip, then
we have the whole thing in-core, and we can freely
retry.
That means we will not retry in some instances, including:
1. If we are using gzip. However, we only do so when
calling git-upload-pack, so it does not apply to
pushes.
2. If we have a large request, the probe succeeds, but
then the real POST wants authentication. This is an
extremely unlikely configuration and not worth worrying
about.
While it might be nice to cover those instances, doing so
would be significantly more complex for very little
real-world gain. In the long run, we will be much better off
when curl learns to internally handle authentication as a
callback, and we can cleanly handle all cases that way.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some sites set up http access to repositories such that
fetching is anonymous and unauthenticated, but pushing is
authenticated. While there are multiple ways to do this, the
technique advertised in the git-http-backend manpage is to
block access to locations matching "/git-receive-pack$".
Let's emulate that advice in our test setup, which makes it
clear that this advice does not actually work.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We do not currently test authentication over smart-http at
all. In theory, it should work exactly as it does for dumb
http (which we do test). It does indeed work for these
simple tests, but this patch lays the groundwork for more
complex tests in future patches.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We do not currently test authentication for smart-http repos
at all. Part of the infrastructure to do this is recognizing
that auth/smart is indeed a smart-http repo.
The current apache config recognizes only "^/smart/*" as
smart-http. Let's instead treat anything with /smart/ in the
URL as smart-http. This is obviously a stupid thing to do
for a real production site, but for our test suite we know
that our repositories will not have this magic string in the
name.
Note that we will route /foo/smart/bar.git directly to
git-http-backend/bar.git; in other words, everything before
the "/smart/" is irrelevant to finding the repo on disk (but
may impact apache config, for example by triggering auth
checks).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Our test apache config points all of auth/ directly to the
on-disk repositories via an Alias directive. This works fine
because everything authenticated is currently in auth/dumb,
which is a subset. However, this would conflict with a
ScriptAlias for auth/smart (which will come in future
patches), so let's narrow the Alias.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The t5550 script sets up a nice askpass helper for
simulating user input and checking what git prompted for.
Let's make it available to other http scripts by migrating
it to lib-httpd.
We can use this immediately in t5540 to make our tests more
robust (previously, we did not check at all that hitting the
password-protected repo actually involved a password).
Unfortunately, we end up failing the test because the
current code erroneously prompts twice (once for
git-remote-http, and then again when the former spawns
git-http-push).
More importantly, though, it will let us easily add
smart-http authentication tests in t5541 and t5551; we
currently do not test smart-http authentication at all.
As part of making it generic, let's always look for and
store auxiliary askpass files at the top-level trash
directory; this makes it compatible with t5540, which runs
some tests from sub-repositories. We can abstract away the
ugliness with a short helper function.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In most of our tests, we put repos to be accessed by dumb
protocols in /dumb, and repos to be accessed by smart
protocols in /smart. In our test apache setup, the whole
/auth hierarchy requires authentication. However, we don't
bother to split it by smart and dumb here because we are not
currently testing smart-http authentication at all.
That will change in future patches, so let's be explicit
that we are interested in testing dumb access here. This
also happens to match what t5540 does for the push tests.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Back when "git apply" was written, we made sure that the user can
skip more than the default number of path components (i.e. 1) by
giving "-p<n>", but the logic for doing so was built around the
notion of "we skip N slashes and stop". This obviously does not
work well when running under -p0 where we do not want to skip any,
but still want to skip SP/HT that separates the pathnames of
preimage and postimage and want to reject absolute pathnames.
Stop using "stop_at_slash()", and instead introduce a new helper
"skip_tree_prefix()" with similar logic but works correctly even for
the -p0 case.
This is an ancient bug, but has been masked for a long time because
most of the patches are text and have other clues to tell us the
name of the preimage and the postimage.
Noticed by Colin McCabe.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint-1.7.11:
Prepare for 1.7.11.6
Make the ciabot scripts completely self-configuring in the normal case.
Improved documentation for the ciabot scripts.
man: git pull -r is a short for --rebase
gitcli: describe abbreviation of long options
rev-list docs: clarify --topo-order description
Documentation/CodingGuidelines: spell out more shell guidelines
Documentation: do not mention .git/refs/* directories
tests: Introduce test_seq
Currently, it will only do a checkout if the sha1 registered in the containing
repository doesn't match the HEAD of the submodule, regardless of whether the
submodule is dirty. As discussed on the mailing list, the '--force' flag is a
strong indicator that the state of the submodule is suspect, and should be reset
to HEAD.
Signed-off-by: Stefan Zager <szager@google.com>
Acked-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Either end of revision range operator can be omitted to default to HEAD,
as in "origin.." (what did I do since I forked) or "..origin" (what did
they do since I forked). But the current parser interprets ".." as an
empty range "HEAD..HEAD", and worse yet, because ".." does exist on the
filesystem, we get this annoying output:
$ cd Documentation/howto
$ git log .. ;# give me recent commits that touch Documentation/ area.
fatal: ambiguous argument '..': both revision and filename
Use '--' to separate filenames from revisions
Surely we could say "git log ../" or even "git log -- .." to disambiguate,
but we shouldn't have to.
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Besides reusing the new test prerequisite, this fixes also the issue
that the current output is not TAP compliant and produces the output "no
reason given" [for skipping].
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The linked list describing sort options was not correctly set up in
opt_parse_sort. In the result, contrary to the documentation, only the
last of multiple --sort options to git-for-each-ref was taken into
account. This commit fixes it.
Signed-off-by: Kacper Kornet <draenog@pld-linux.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Documentation of git-for-each-ref says that --sort=<key> option can be
used multiple times, in which case the last key becomes the primary key.
However this functionality was never checked in test suite and is
currently broken. This commit adds appropriate test in preparation for fix.
Signed-off-by: Kacper Kornet <draenog@pld-linux.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"stash apply" directly calls a backend merge function which does not
automatically invoke rerere. This confuses mergetool when leftover
rerere state is left behind from previous merges.
Invoke rerere explicitly when we encounter a conflict during stash
apply. This turns the test introduced by the previous commit to
succeed.
Signed-off-by: Phil Hord <hordp@cisco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a test to make sure that a conflicted "stash apply" invokes
rerere to record the conflicts and resolve the the files it can
(the current code doesn't, so the test is marked as failing).
Without correct state recorded for rerere, mergetool may be
confused, causing it to think no files have conflicts even though
they do. This condition is not verified by this test since a
subsequent commit will change the behavior to enable rerere for
stash conflicts.
Also, the next test expected us to finish up with a reset, which is
impossible to do if we fail (as we must) and it's an unreasonable
expectation anyway. Begin the next test with a reset of its own
instead.
Signed-off-by: Phil Hord <hordp@cisco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Various subcommands of the "git submodule" command exited with 0
status even though the path given by the user did not exist.
The reason behind that was that they all pipe the output of
module_list into the while loop which then does the action on the
paths specified by the commandline. Since the exit code of the
command on the upstream side of the pipe is ignored by the shell,
the status code of "ls-files --error-unmatch" nor "module_list" was
not propagated.
In case ls-files returns with an error code, we write a special
string that is not possible in non error situations, and no other
output, so that the downstream can detect the error and die with an
error code.
The error message that there is an unmatched pathspec comes through
stderr directly from ls-files. So the user still gets a hint whats going
on.
Signed-off-by: Heiko Voigt <hvoigt@hvoigt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since v1.7.0-rc2~11 (git-svn: persistent memoization, 2010-01-30),
git-svn has maintained some private per-repository caches in
.git/svn/.caches to avoid refetching and recalculating some
mergeinfo-related information with every 'git svn fetch'.
This memoization can cause problems, e.g consider the following case:
SVN repo:
... - a - b - c - m <- trunk
\ /
d - e <- branch1
The Git import of the above repo is at commit 'a' and doesn't know about
the branch1. In case of an 'git svn rebase', only the trunk of the
SVN repo is imported. During the creation of the git commit 'm', git svn
uses the svn:mergeinfo property and tries to find the corresponding git
commit 'e' to create 'm' with 'c' and 'e' as parents. But git svn rebase
only imports the current branch so commit 'e' is not imported.
Therefore git svn fails to create commit 'm' as a merge commit, because one
of its parents is not known to git. The imported history looks like this:
... - a - b - c - m <- trunk
A later 'git svn fetch' to import all branches can't rewrite the commit 'm'
to add 'e' as a parent and to make it a real git merge commit, because it
was already imported.
That's why the imported history misses the merge and looks like this:
... - a - b - c - m <- trunk
\
d - e <- branch1
Right now the only known workaround for importing 'm' as a merge is to
force reimporting 'm' again from SVN, e.g. via
$ git svn reset --revision $(git find-rev $c)
$ git svn fetch
Sadly, this is where the behavior has regressed: git svn reset doesn't
invalidate the old mergeinfo cache, which is no longer valid for the
reimport, which leads to 'm' beeing imprted with only 'c' as parent.
As solution to this problem, this commit invalidates the mergeinfo cache
to force correct recalculation of the parents.
During development of this patch, several ways for invalidating the cache
where considered. One of them is to use Memoize::flush_cache, which will
call the CLEAR method on the underlying Memoize persistency implementation.
Sadly, neither Memoize::Storable nor the newer Memoize::YAML module
introduced in 68f532f4ba could optionally be used implement the
CLEAR method, so this is not an option.
Reseting the internal hash used to store the memoized values has the same
problem, because it calls the non-existing CLEAR method of the
underlying persistency layer, too.
Considering this and taking into account the different implementations
of the memoization modules, where Memoize::Storable is not in our control,
implementing the missing CLEAR method is not an option, at least not if
Memoize::Storable is still used.
Therefore the easiest solution to clear the cache is to delete the files
on disk in 'git svn reset'. Normally, deleting the files behind the back
of the memoization module would be problematic, because the in-memory
representation would still exist and contain wrong data. Fortunately, the
memoization is active in memory only for a small portion of the code.
Invalidating the cache by deleting the files on disk if it isn't active
should be safe.
Signed-off-by: Peter Baumann <waste.manager@gmx.de>
Signed-off-by: Steven Walter <stevenrwalter@gmail.com>
Signed-off-by: Eric Wong <normalperson@yhbt.net>
dcommit didn't handle errors returned by SVN and coped very
poorly with concurrent commits that appear in SVN repository
while dcommit was running. In both cases it left git repository
in inconsistent state: index (which was reset with `git reset
--mixed' after a successful commit to SVN) no longer matched the
checkouted tree, when the following commit failed or needed to be
rebased. See http://bugs.debian.org/676904 for examples.
This patch fixes the issues by:
- introducing error handler for dcommit. The handler will try
to rebase or reset working tree before returning error to the
end user. dcommit_rebase function was extracted out of cmd_dcommit
to ensure consistency between cmd_dcommit and the error handler.
- calling `git reset --mixed' only once after all patches are
successfully committed to SVN. This ensures index is not touched
for most of the time of dcommit run.
Signed-off-by: Eric Wong <normalperson@yhbt.net>
Add test cases for 'git rebase --keep-empty' with and without an
"empty" commit already in upstream. The empty commit that is about to
be rebased should be kept in both cases.
Signed-off-by: Martin von Zweigbergk <martin.von.zweigbergk@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The standard output channel of receive-pack is a structured protocol
channel, and subprocesses must never be allowed to leak anything
into it by writing to their standard output.
Use RUN_COMMAND_STDOUT_TO_STDERR option to run_command_v_opt() just
like we do when running hooks to prevent output from "gc" leaking to
the standard output.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When receive-pack triggers 'git gc --auto' and 'git prune' is called to
remove a stale temporary object, 'git prune' prints an informational
message to stdout about the file that it will remove. Since this message
is written to stdout, it is sent back over the transport channel to the git
client which tries to interpret it as part of the pack protocol and then
promptly terminates with a complaint about a protocol error.
Introduce a test which exercises the auto-gc functionality of receive-pack
and demonstrates this breakage.
Signed-off-by: Brandon Casey <drafnel@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a file that ends with an incomplete line is expressed as a
complete rewrite with the -B option, git diff incorrectly
appends the incomplete line indicator "\ No newline at end of
file" after such a line, rather than writing it on a line of its
own (the output codepath for normal output without -B does not
have this problem). Add a LF after the incomplete line before
writing the "\ No newline ..." out to fix this.
Add a couple of tests to confirm that the indicator comment is
generated on its own line in both plain diff and rewrite mode.
Signed-off-by: Adam Butcher <dev.lists@jessamine.co.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jeff King wrote:
The seq command is GNU-ism, and is missing at least in older BSD
releases and their derivatives, not to mention antique
commercial Unixes.
We already purged it in b3431bc (Don't use seq in tests, not
everyone has it, 2007-05-02), but a few new instances have crept
in. They went unnoticed because they are in scripts that are not
run by default.
Replace them with test_seq that is implemented with a Perl snippet
(proposed by Jeff). This is better than inlining this snippet
everywhere it's needed because it's easier to read and it's easier
to change the implementation (e.g. to C) if we ever decide to remove
Perl from the test suite.
Note that test_seq is not a complete replacement for seq(1). It
just has what we need now, in addition that it makes it possible for
us to do something like "test_seq a m" if we wanted to in the
future.
There are also many places that do `for i in 1 2 3 ...` but I'm not sure
if it's worth converting them to test_seq. That would introduce running
more processes of Perl.
Signed-off-by: Michał Kiedrowicz <michal.kiedrowicz@gmail.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The RFC2047 unquoting, used to parse email addresses in From and Cc
headers, is broken in several ways:
* It erroneously substitutes ' ' for '_' in *the whole* header, even
outside the quoted field. [Noticed by Christoph.]
* It is too liberal in its matching, and happily matches the start
of one quoted chunk against the end of another, or even just
something that looks like such an end. [Noticed by Junio.]
* It fundamentally cannot cope with encodings that are not a
superset of ASCII, nor several (incompatible) encodings in the
same header.
This patch fixes the first two by doing a more careful decoding of
the outer quoting (e.g. "=AB" to represent an octet whose value is
0xAB). Fixing the fundamental issues is left for a future, more
intrusive, patch.
Noticed-by: Christoph Miebach <christoph.miebach@web.de>
Helped-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add i18n support for scripted Porcelains, and mark strings in
merge(-recursive), am, and rebase for i18n.
* jx/i18n-1.7.11:
i18n: merge-recursive: mark strings for translation
Remove dead code which contains bad gettext block
i18n: am: mark more strings for translation
rebase: remove obsolete and unused LONG_USAGE which breaks xgettext
i18n: Rewrite gettext messages start with dash
i18n: rebase: mark messages for translation
i18n: New keywords for xgettext extraction from sh
"git commit-tree" learned a more natural "-p <parent> <tree>" order
of arguments long time ago, but recently forgot it by mistake.
* kk/maint-commit-tree:
Revert "git-commit-tree(1): update synopsis"
commit-tree: resurrect command line parsing updates
"git diff --no-ext-diff" did not output anything for a typechange
filepair when GIT_EXTERNAL_DIFF is in effect.
* jv/maint-no-ext-diff:
diff: test precedence of external diff drivers
diff: correctly disable external_diff with --no-ext-diff
When "git submodule add" clones a submodule repository, it can get
confused where to store the resulting submodule repository in the
superproject's .git/ directory when there is a symbolic link in the
path to the current directory.
* jl/maint-1.7.10-recurse-submodules-with-symlink:
submodules: don't stumble over symbolic links when cloning recursively
In 1.7.9 era, we taught "git rebase" about the raw timestamp format
but we did not teach the same trick to "filter-branch", which rolled
a similar logic on its own.
* jc/maint-filter-branch-epoch-date:
t7003: add test to filter a branch with a commit at epoch
date.c: Fix off by one error in object-header date parsing
filter-branch: do not forget the '@' prefix to force git-timestamp
"git grep" stopped spawning an external "grep" long time ago, but a
duplicated test to check internal and external "grep" was left
behind.
* rj/maint-grep-remove-redundant-test:
t7810-*.sh: Remove redundant test
Finishing touches to the new test script.
* dg/submodule-in-dismembered-working-tree:
t7409: make sure submodule is initialized and updated in more detail
The earlier test did not even make sure that the correct commit is
checked out in the submodule directory. Inspect the result in a bit
more detail.
Signed-off-by: Daniel Graña <dangra@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The test happened to use "rev-parse --max-count=1 HEAD" consistently
to prepare the expected output and the actual output, so the
comparison between them gave us a correct success/failure because
both output had irrelevant "--max-count=1" in it.
But that is not an excuse to keep it broken. Replace it a more
meaningful construct "rev-parse --verify HEAD".
Noticed by Daniel Graña while working on his submodule tests.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since commit bbc09c22 ("grep: rip out support for external grep",
12-01-2010), test number 60 ("grep -C1 hunk mark between files") is
essentially the same as test number 59.
Test 59 was intended to verify the behaviour of git-grep resulting
from multiple invocations of an external grep. As part of the test,
it creates and adds 1024 files to the index, which is now wasted
effort.
Remove test 59, since it is now redundant.
Signed-off-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In particular, the final test ('flags and then non flags') fails
intermittently, depending on how much time elapsed between the
invocations of "git commit-tree" when creating the commits which
later have their commit id's compared. For example, if the commits
for childid-3 and childid-4 are created 1 or more seconds apart,
then the commits, which would otherwise be identical, will have
different commit id's.
In order to make the test reproducible, we remove the variability
by setting the author and committer times to a well defined state.
We accomplish this with a single call to 'test_tick' at the start
of the test.
Signed-off-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Short of somebody happening to beat the 1 in 2^160 odds of
actually generating content that hashes to the null sha1, we
should never see this value in a tree entry. So let's have
fsck warn if it it seen.
As in the previous commit, we test both blob and submodule
entries to future-proof the test suite against the
implementation depending on connectivity to notice the
error.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We should never need to write the null sha1 into an index
entry (short of the 1 in 2^160 chance that somebody actually
has content that hashes to it). If we attempt to do so, it
is much more likely that it is a bug, since we use the null
sha1 as a sentinel value to mean "not valid".
The presence of null sha1s in the index (which can come
from, among other things, "update-index --cacheinfo", or by
reading a corrupted tree) can cause problems for later
readers, because they cannot distinguish the literal null
sha1 from its use a sentinel value. For example, "git
diff-files" on such an entry would make it appear as if it
is stat-dirty, and until recently, the diff code assumed
such an entry meant that we should be diffing a working tree
file rather than a blob.
Ideally, we would stop such entries from entering even our
in-core index. However, we do sometimes legitimately add
entries with null sha1s in order to represent these sentinel
situations; simply forbidding them in add_index_entry breaks
a lot of the existing code. However, we can at least make
sure that our in-core sentinel representation never makes it
to disk.
To be thorough, we will test an attempt to add both a blob
and a submodule entry. In the former case, we might run into
problems anyway because we will be missing the blob object.
But in the latter case, we do not enforce connectivity
across gitlink entries, making this our only point of
enforcement. The current implementation does not care which
type of entry we are seeing, but testing both cases helps
future-proof the test suite in case that changes.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The diff code represents paths using the diff_filespec
struct. This struct has a sha1 to represent the sha1 of the
content at that path, as well as a sha1_valid member which
indicates whether its sha1 field is actually useful. If
sha1_valid is not true, then the filespec represents a
working tree file (e.g., for the no-index case, or for when
the index is not up-to-date).
The diff_filespec is only used internally, though. At the
interfaces to the diff subsystem, callers feed the sha1
directly, and we create a diff_filespec from it. It's at
that point that we look at the sha1 and decide whether it is
valid or not; callers may pass the null sha1 as a sentinel
value to indicate that it is not.
We should not typically see the null sha1 coming from any
other source (e.g., in the index itself, or from a tree).
However, a corrupt tree might have a null sha1, which would
cause "diff --patch" to accidentally diff the working tree
version of a file instead of treating it as a blob.
This patch extends the edges of the diff interface to accept
a "sha1_valid" flag whenever we accept a sha1, and to use
that flag when creating a filespec. In some cases, this
means passing the flag through several layers, making the
code change larger than would be desirable.
One alternative would be to simply die() upon seeing
corrupted trees with null sha1s. However, this fix more
directly addresses the problem (while bogus sha1s in a tree
are probably a bad thing, it is really the sentinel
confusion sending us down the wrong code path that is what
makes it devastating). And it means that git is more capable
of examining and debugging these corrupted trees. For
example, you can still "diff --raw" such a tree to find out
when the bogus entry was introduced; you just cannot do a
"--patch" diff (just as you could not with any other
corrupted tree, as we do not have any content to diff).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* 'extract-remaining' of git://git.bogomips.org/git-svn:
Extract Git::SVN::GlobSpec from git-svn.
Move Git::IndexInfo into its own file.
Load all the modules in one place and before running code.
Extract Git::SVN::Migration from git-svn.
Prepare Git::SVN::Migration for extraction from git-svn.
Extract Git::SVN::Log from git-svn.
Prepare Git::SVN::Log for extraction from git-svn.
* git://git.bogomips.org/git-svn:
Move initialization of Git::SVN variables into Git::SVN.
Extract Git::SVN from git-svn into its own .pm file.
Prepare Git::SVN for extraction into its own file.
Extract some utilities from git-svn to allow extracting Git::SVN.
perl: detect new files in MakeMaker builds
The Makefile.PL will now find .pm files itself.
Don't lose Error.pm if $@ gets clobbered.
Quiet warning if Makefile.PL is run with -w and no --localedir
Fix test breakages by a builder who does not have a valid user name
in his /etc/password entry.
* jk/autoident-test:
t7502: test early quit from commit with bad ident
t7502: handle systems where auto-identity is broken
t7502: drop confusing test_might_fail call
t7502: narrow checks for author/committer name in template
t7502: properly quote GIT_EDITOR
t7502: clean up fake_editor tests
In a superproject that has repository outside of its working tree,
"git submodule add" failed to clone a new submodule, as GIT_DIR and
GIT_WORK_TREE environment variables necessary to work in such a
superproject interfered with access to the submodule repository.
* dg/submodule-in-dismembered-working-tree:
git-submodule: work with GIT_DIR/GIT_WORK_TREE
Straight cut & paste. That's the last class.
* Make Git::SVN load it on its own, its the only thing that needs it.
Signed-off-by: Eric Wong <normalperson@yhbt.net>
Put them in a new module called Git::SVN::Utils. Yeah, not terribly
original and it will be a dumping ground. But its better than having
them in the main git-svn program. At least they can be documented
and tested.
* fatal() is used by many classes.
* Change the $can_compress lexical into a function.
This should be enough to extract Git::SVN.
Signed-off-by: Michael G. Schwern <schwern@pobox.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Eric Wong <normalperson@yhbt.net>
UTF8 behaviour of the filesystem (conversion from nfd to nfc) plays a
role in several tests and is tested in several tests. Therefore, move
the test from t0050 into the test lib and use the prerequisite in t0050.
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Case insensitivity plays a role in several tests and is tested in several
tests. Therefore, move the test from t003 into the test lib and use the
prerequisite in t0003.
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The test prerequisite mechanism is a useful way to allow some tests
in a test script to be skipped in environments that do not support
certain features (e.g. it is pointless to attempt checking how well
symbolic links are handled by Git on filesystems that do not support
them). It is OK for commonly used prerequisites to be always tested
during start-up of a test script by having a codeblock that tests a
feature and calls test_set_prereq, but for an uncommon feature,
forcing 90% of scripts to pay the same probing overhead for
prerequisite they do not care about is wasteful.
Introduce a mechanism to probe the prerequiste lazily. Changes are:
- test_lazy_prereq () function, which takes the name of the
prerequisite it probes and the script to probe for it, is
added. This only registers the name of the prerequiste that can
be lazily probed and the script to eval (without running).
- test_have_prereq() function (which is used by test_expect_success
and also can be called directly by test scripts) learns to look
at the list of prerequisites that can be lazily probed, and the
prerequisites that have already been probed that way. When asked
for a prerequiste that can be but haven't been probed, the script
registered with an earlier call to test_lazy_prereq is evaluated
and the prerequisite is set.
- test_run_lazy_prereq_() function is a helper to run the probe
script with the same kind of sandbox as regular tests, helped by
Jeff King.
Update the codeblock to probe and set SYMLINKS prerequisite using
the new mechanism as an example.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
All other shell variables that are used to globally keep track of
states related to prerequisite have "prereq" somewhere in their
names. Be consistent and avoid potential name crashes with other
kinds of satisfaction in the future.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Mark strings in merge-recursive for translation.
Some tests would start to fail with GETTEXT_POISON turned on after
this update. Use test_i18ncmp and test_i18ngrep where appropriate
to mark strings that should only be checked in the C locale output
to avoid such issues.
Signed-off-by: Jiang Xin <worldhello.net@gmail.com>
Reviewed-by: Stefano Lattarini <stefano.lattarini@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Gettext message in a shell script should not start with '-', one
workaround is adding '--' between gettext and the message, like:
gettext -- "--exec option ..."
But due to a bug in the xgettext extraction, xgettext can not
extract the actual message for this case. Rewriting the message
is a simpler and better solution.
Reported-by: Vincent van Ravesteijn <vfr@lyx.org>
Signed-off-by: Jiang Xin <worldhello.net@gmail.com>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Reviewed-by: Stefano Lattarini <stefano.lattarini@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In commit f20f387, "git commit" notices and dies much
earlier when we have a bogus commit identity. That commit
did not add a test because we cannot do so reliably (namely,
we can only trigger the behavior on a system where the
automatically generated identity is bogus). However, now
that we have a prerequisite check for this feature, we can
add a test that will at least run on systems that produce
such a bogus identity.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Test t7502.21 checks whether we write the committer name
into COMMIT_EDITMSG when it has been automatically
determined. However, not all systems can produce valid
automatic identities.
Prior to f20f387 (commit: check committer identity more
strictly), this test worked even when we did not have a
valid automatic identity, since it did not run the strict
test until after we had generated the template. That commit
tightened the check to fail early (since we would fail
later, anyway), meaning that systems without a valid GECOS
name or hostname would fail the test.
We cannot just work around this, because it depends on
configuration outside the control of the test script.
Therefore we introduce a new test_prerequisite to run this
test only on systems where automatic ident works at all.
As a result, we can drop the confusing test_must_fail bit
from the test. The intent was that by giving "git commit"
invalid input (namely, nothing to commit), that it would
stop at a predictable point, whether we had a valid identity
or not, from which we could view the contents of
COMMIT_EDITMSG. Since that assumption no longer holds, and
we can only run the test when we have a valid identity,
there is no reason not to let commit run to completion. That
lets us be more robust to other unforeseen failures.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In t7502.20, we run "git commit" and check that it warns us
that the author and committer identity are not the same
(this is always the case in the test environment, since we
set up the idents differently).
Instead of actually making a commit, we have a clean index,
so the "git commit" we run will fail. This is marked as
might_fail, which is not really correct; it will always fail
since there is nothing to commit.
However, the only reason not to do a complete commit would
be to see the intermediate state of the COMMIT_EDITMSG file
when the commit is not completed. We don't need to care
about this, though; even a complete commit will leave
COMMIT_EDITMSG for us to view. By doing a real commit and
dropping the might_fail, we are more robust against other
unforeseen failures of "git commit" that might influence our
test result.
It might seem less robust to depend on the fact that "git
commit" leaves COMMIT_EDITMSG in place after a successful
commit. However, that brings this test in line with others
parts of the script, which make the same assumption.
Furthermore, if that ever does change, the right solution is
not to prevent commit from completing, but to set EDITOR to
a script that will record the contents we see. After all,
the point of these tests is to check what the user sees in
their EDITOR, so that would be the most direct test. For
now, though, we can continue to use the "shortcut" that
COMMIT_EDITMSG is left intact.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
t7502.20 and t7502.21 check that the author and committer
name are mentioned in the commit message template under
certain circumstances. However, they end up checking a much
larger and unnecessary portion of the template. Let's narrow
their checks to the specific lines.
While we're at it, let's give these tests more descriptive
names, so their purposes are more obvious.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
One of the tests tries to ensure that editor is not run due
to an early failure. However, it needs to quote the pathname
of the trash directory used in $GIT_EDITOR, since git will
pass it along to the shell. In other words, the test would
pass whether the code was correct or not, since the unquoted
editor specification would never run.
We never noticed the problem because the code is indeed
correct, so git-commit never even tried to run the editor.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Using write_script saves us a few lines of code, and means
we consistently use $SHELL_PATH.
We can also drop the setting of the $pwd variable from
$(pwd). In the first instance, there is no reason to use it
(we can just use $(pwd) directly two lines later, since we
are interpolating the here-document). In the second
instance, it is totally pointless and probably just a
cut-and-paste from the first instance.
Finally, we can use a non-interpolating here document for
the final script, which saves some quoting.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Reorders t/test-lib.sh so that we dot-source GIT-BUILD-OPTIONS that
records the shell and Perl the user told us to use with Git a lot
early, so that test-lib.sh script itself can use "$PERL_PATH" in
one of its early operations.
* jc/test-lib-source-build-options-early:
test-lib: reorder and include GIT-BUILD-OPTIONS a lot earlier
Finishing touches to the XDG support (new feature for 1.7.12) and
tests.
* mm/config-xdg:
t1306: check that XDG_CONFIG_HOME works
ignore: make sure we have an xdg path before using it
attr: make sure we have an xdg path before using it
test-lib.sh: unset XDG_CONFIG_HOME
The combination of GIT_DIR and GIT_WORK_TREE can be used to manage
files in one directory hierarchy while keeping the repository that
keeps track of them outside the directory hierarchy. For example:
git init --bare /path/to/there
alias dotfiles="GIT_DIR=/path/to/there GIT_WORK_TREE=/path/to/here git"
cd /path/to/here
dotfiles add file
dotfiles commit -a -m "add /path/to/here/file"
...
lets you manage files under /path/to/here/ in the repository located
at /path/to/there.
git-submodule however fails to add submodules, as it is confused by
GIT_DIR and GIT_WORK_TREE environment variables when it tries to
work in the submodule, like so:
dotfiles submodule add http://path.to/submodule
fatal: working tree '/path/to/here' already exists.
Simply unsetting the environment where the command works on the
submodule is sufficient to fix this, as it has set things up so
that GIT_DIR and GIT_WORK_TREE do not even have to point at the
repository and the working tree of the submodule.
Signed-off-by: Daniel Graña <dangra@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Mark messages in git-rebase.sh for translation. While doing this
Jonathan noticed that the comma usage and sentence structure of the
resolvemsg was not quite right, so correct that and its cousins in
git-am.sh and t/t0201-gettext-fallbacks.sh at the same time.
Some tests would start to fail with GETTEXT_POISON turned on after
this update. Use test_i18ncmp and test_i18ngrep where appropriate
to mark strings that should only be checked in the C locale output
to avoid such issues.
Signed-off-by: Jiang Xin <worldhello.net@gmail.com>
Reviewed-by: Stefano Lattarini <stefano.lattarini@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Running the http tests with valgrind does not work for two
reasons:
1. Apache complains about following the symbolic link from
git-http-backend to valgrind.sh.
2. Apache does not pass through the GIT_VALGRIND variable
to the backend CGI.
This patch fixes both problems. Unfortunately, there is a
slight hack we need to handle passing environment variables
through Apache. If we just tell it:
PassEnv GIT_VALGRIND
then Apache will complain when GIT_VALGRIND is not set. If
we try:
SetEnv GIT_VALGRIND ${GIT_VALGRIND}
then when GIT_VALGRIND is not set, it will pass through the
literal "${GIT_VALGRIND}". Instead, we now unconditionally
pass through GIT_VALGRIND from lib-httpd.sh into apache,
even if it is empty.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There is a bug with git rebase -i --root when a fixup or squash line is
applied to the new root. We attempt to amend the commit onto which they
apply with git reset --soft HEAD^ followed by a normal commit. Unlike a
real commit --amend, this sequence will fail against a root commit as it
has no parent.
Fix rebase -i to use commit --amend for fixup and squash instead, and
add a test for the case of a fixup of the root commit.
Signed-off-by: Chris Webb <chris@arachsys.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This should override $HOME/.config, but we never actually tested it.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit e3ebc35 (config: fix several access(NULL) calls, 2012-07-12) was
fixing access(NULL) calls when trying to access $HOME/.config/git/config,
but missed the ones when trying to access $HOME/.config/git/ignore. Fix
and test this.
Signed-off-by: Matthieu Moy <Matthieu.Moy@imag.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If we don't have a core.attributesfile configured, we fall
back to checking XDG config, which is usually
$HOME/.config/git/attributes.
However, if $HOME is unset, then home_config_paths will return
NULL, and we end up calling fopen(NULL).
Depending on your system, this may or may not cause the
accompanying test to fail (e.g., on Linux and glibc, the
address will go straight to open, which will return EFAULT).
However, valgrind will reliably notice the error.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Now that git respects XDG_CONFIG_HOME for some lookups, we
must be sure to cleanse the test environment. Otherwise, the
user's XDG_CONFIG_HOME could influence the test results.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git diff --no-ext-diff" did not output anything for a typechange
filepair when GIT_EXTERNAL_DIFF is in effect.
* jv/maint-no-ext-diff:
diff: test precedence of external diff drivers
diff: correctly disable external_diff with --no-ext-diff
"commit --amend" used to refuse amending a commit with an empty log
message, with or without "--allow-empty-message".
* cw/amend-commit-without-message:
Allow edit of empty message with commit --amend
"git commit --amend --only --" was meant to allow "Clever" people to
rewrite the commit message without making any change even when they
have already changes for the next commit added to their index, but
it never worked as advertised since it was introduced in 1.3.0 era.
* jk/maint-commit-amend-only-no-paths:
commit: fix "--amend --only" with no pathspec
Even though the index can record pathnames longer than 1<<12 bytes,
in some places we were not comparing them in full, potentially
replacing index entries instead of adding.
* tg/maint-cache-name-compare:
cache_name_compare(): do not truncate while comparing paths
"git show"'s auto-walking behaviour was an unreliable and
unpredictable hack; it now behaves just like "git log" does when it
walks.
* tr/maint-show-walk:
show: fix "range implies walking"
Demonstrate git-show is broken with ranges
"git diff", "git status" and anything that internally uses the
comparison machinery was utterly broken when the difference
involved a file with "-" as its name. This was due to the way "git
diff --no-index" was incorrectly bolted on to the system, making
any comparison that involves a file "-" at the root level
incorrectly read from the standard input.
* jc/refactor-diff-stdin:
diff-index.c: "git diff" has no need to read blob from the standard input
diff-index.c: unify handling of command line paths
diff-index.c: do not pretend paths are pathspecs
We did not have test to make sure "git rebase" without extra options
filters out an empty commit in the original history.
* mz/empty-rebase-test:
add test case for rebase of empty commit
"git fast-export" produced an input stream for fast-import without
properly quoting pathnames when they contain SPs in them.
* js/fast-export-paths-with-spaces:
fast-export: quote paths with spaces
"git checkout --detach", when you are still on an unborn branch,
should be forbidden, but it wasn't.
* cw/no-detaching-an-unborn:
git-checkout: disallow --detach on unborn branch
Some implementations of Perl terminates "lines" with CRLF even when
the script is operating on just a sequence of bytes. Make sure to
use "$PERL_PATH", the version of Perl the user told Git to use, in
our tests to avoid unnecessary breakages in tests.
* vr/use-our-perl-in-tests:
t/README: add a bit more Don'ts
tests: enclose $PERL_PATH in double quotes
t/test-lib.sh: export PERL_PATH for use in scripts
t: Replace 'perl' by $PERL_PATH
* as/t4012-style-updates:
t4012: Use test_must_fail instead of if-else
t4012: use 'printf' instead of 'dd' to generate a binary file
t4012: Re-indent test snippets
t4012: Make --shortstat test more robust
t4012: Break up pipe into serial redirections
t4012: Actually quote the sed script
t4012: Unquote git command fragment in test title
t4012: modernize style for quoting
