Previous D/F fixes I submitted (5a2580d and ae74548) had caused merge to
become excessively spammy, which was fixed in 96ecac6 (merge-recursive:
Avoid excessive output for and reprocessing of renames 2010-08-20). Add a
new test to avoid repeating that mistake with my several upcoming changes.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
d5af510 (RE: [PATCH] Avoid rename/add conflict when contents are identical
2010-09-01) avoided erroring out in a rename/add conflict when the contents
were identical. A simpler fix could have handled that particular testcase,
but it would not correctly handle the case where a symlink is involved.
Add another testcase using symlinks, to avoid breaking that case.
Signed-off-by: Ken Schalk <ken.schalk@intel.com>
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We need to get the correct mode when blame reads the source from the
working tree, the index, or trees. This allows us to omit running
textconv filters on symbolic links.
Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru>
Reviewed-by: Matthieu Moy <Matthieu.Moy@grenoble-inp.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git blame --textconv is wrongly calling the textconv filter on
symlinks: symlinks are stored as blobs whose content is the target of
the link, and blame calls the textconv filter on a temporary file
filled-in with the content of this blob.
For example:
$ git blame -C -C regular-file.pdf
Error: May not be a PDF file (continuing anyway)
Error: PDF file is damaged - attempting to reconstruct xref table...
Error: Couldn't find trailer dictionary
Error: Couldn't read xref table
Warning: program returned non-zero exit code #1
fatal: unable to read files to diff
That errors come from pdftotext run on symlink.pdf being extracted to
/tmp/ with one-line plain-text content pointing to link destination.
So several failures are demonstrated here:
- git cat-file --textconv :symlink.bin # also HEAD:symlink.bin
- git blame --textconv symlink.bin
- git blame -C -C --textconv regular-file # but also looks on symlink.bin
At present they all fail with something like.
E: /tmp/j3ELEs_symlink.bin is not "binary" file
NOTE: git diff doesn't try to textconv the pathnames, it runs the
textual diff without textconv, which is the expected behavior.
Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru>
Reviewed-by: Matthieu Moy <Matthieu.Moy@grenoble-inp.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The textconv filter is sometimes incorrectly ran on a temporary file
whose content is the target of a symbolic link, instead of actual file
content. Prepare to test this by marking the content of the file to
convert with "bin:", and let the helper die if "bin:" is not found in
the file content.
NOTE: I've changed $@ to $1 in helper becase textconv program "should
take a single argument" (see Documentation/gitattributes.txt), so
making this more explicit makes sense and also helps to avoid
problems with feeding arguments to echo.
Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru>
Reviewed-by: Matthieu Moy <Matthieu.Moy@grenoble-inp.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jc/grep-header-all-match-fix:
log --author: take union of multiple "author" requests
grep: move logic to compile header pattern into a separate helper
* jc/pickaxe-grep:
diff/log -G<pattern>: tests
git log/diff: add -G<regexp> that greps in the patch text
diff: pass the entire diff-options to diffcore_pickaxe()
gitdiffcore doc: update pickaxe description
* rr/fmt-merge-msg:
t6200-fmt-merge-msg: Exercise '--log' to configure shortlog length
t6200-fmt-merge-msg: Exercise 'merge.log' to configure shortlog length
merge: Make 'merge.log' an integer or boolean option
merge: Make '--log' an integer option for number of shortlog entries
fmt_merge_msg: Change fmt_merge_msg API to accept shortlog_len
Conflicts:
builtin/merge.c
It's a minor annoyance when you take the painstaking time to setup To:
headers for each patch in a large series, and then go out to send the
series with git-send-email and watch git ignore the To: headers in the
patch files.
Therefore, always add To: headers from a patch file to the To: headers
for that message. Keep the prompt for the blanket To: header so as to
not break scripts (and user expectations). This means even if a patch
has a To: header, git will prompt for the To: address. Otherwise, we'll
need to introduce interface breakage to either request the header for
each patch missing a To: header or default the header to whatever To:
address is found first (be it in a patch or from user input). Both of
these options don't seem very obvious/useful.
Reported-by: Viresh Kumar <viresh.kumar@st.com>
Signed-off-by: Stephen Boyd <bebarino@gmail.com>
Tested-by: Viresh Kumar <viresh.kumar@st.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git stash branch <branch> <stash>" started discarding the stash
when the branch creation fails. It should have kept the stash
intact when aborting.
Signed-off-by: Jon Seymour <jon.seymour@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This bug was disovered by someone on IRC when he tried to
$ git stash branch <branch> <stash>
while <branch> already existed. In that case the stash is dropped even
though it isn't applied on any branch, so the stash is effectively lost.
Signed-off-by: Tomas Carnecky <tom@dbservice.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Formalize our dependency on perl 5.8, bumped from 5.6.[12]. We already
used the three-arg form of open() which was introduced in 5.6.1, but
t/t9700/test.pl explicitly depended on 5.6.2.
However git-add--interactive.pl has been failing on the 5.6 line since
it was introduced in v1.5.0-rc0~12^2~2 back in 2006 due to this open
syntax:
sub run_cmd_pipe {
my $fh = undef;
open($fh, '-|', @_) or die;
return <$fh>;
}
Which when executed dies on "Can't use an undefined value as
filehandle reference". Several of our tests also fail on 5.6 (even
more when compiled with NO_PERL_MAKEMAKER=1):
t2016-checkout-patch.sh
t3904-stash-patch.sh
t3701-add-interactive.sh
t7105-reset-patch.sh
t7501-commit.sh
t9700-perl-git.sh
Our code is bitrotting on 5.6 with no-one interested in fixing it, and
pinning us to such an ancient release of Perl is keeping us from using
useful features introduced in the 5.8 release.
The 5.6 series is now over 10 years old, and the 5.6.2 maintenance
release almost 7. 5.8 on the other hand is more than 8 years old.
All the modern Unix-like operating systems have now upgraded to it or
a later version, and 5.8 packages are available for old IRIX, AIX
Solaris and Tru64 systems.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Acked-by: Tor Arntsen <tor@spacetec.no>
Acked-by: Randal L. Schwartz <merlyn@stonehenge.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add the ability to use a command line --to-cmd=cmd
to create the list of "To:" addresses.
Used a shared routine for --cc-cmd and --to-cmd.
Did not use IPC::Open2, leaving that for Ævar if
ever he decides to fix the other bugs he might find.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The test was missing some "&&" at the end of some lines and it
was wrong because, as the replacement refs were not fetched,
the commits from the parallel branch should not show up. This
was found by Elijah Newren.
This is fixed by checking that after the branch from HASH6 is
fetched, the commits from the parallel branch don't show up,
and then by fetching the replacement refs and checking that
they do show up afterwards.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
For a long time (29508e1 "Isolate shared HTTP request functionality", Fri
Nov 18 11:02:58 2005), we've followed HTTP redirects with
CURLOPT_FOLLOWLOCATION.
However, when the remote HTTP server returns a redirect the default
libcurl action is to change a POST request into a GET request while
following the redirect, but the remote http backend does not expect
that.
Fix this by telling libcurl to always keep the request as type POST with
CURLOPT_POSTREDIR.
For users of libcurl older than 7.19.1, use CURLOPT_POST301 instead,
which only follows 301s instead of both 301s and 302s.
Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: Tay Ray Chuan <rctay89@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Even though "-L" is POSIX, the former is more portable, and
we tend to prefer it already.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
On systems which have dash as /bin/sh, such as Ubuntu, the final
test (master@{n} for various n) fails with a syntax error while
processing an arithmetic expansion. The syntax error is caused by
using a bare name ('N') as a variable reference in the expression.
In order to avoid the syntax error, we spell the variable reference
as '$N' rather than simply 'N'.
Signed-off-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
cherry-pick will segfault when transplanting a root commit if the --ff
option is used. This happens because the "parent" pointer is set to NULL
when the commit being cherry-picked has no parents. Later, when "parent"
is dereferenced, the cherry-pick segfaults.
Fix this by checking whether "parent" is NULL before dereferencing it and
add a test for this case of cherry-picking a root commit with --ff.
Reported-by: Zbyszek Szmek <zbyszek@in.waw.pl>
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Currently git-stash uses `git rev-parse --no-revs -- "$@"` to set its
FLAGS variable. This is the same as `FLAGS="-- $@"`. It should use
`git rev-parse --no-revs --flags "$@"`, but that eats any "-q" or
"--quiet" argument. So move the check for quiet before rev-parse.
Signed-off-by: Brian Gernhardt <brian@gernhardtsoftware.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Recently, the 'stash show' functionality was broken for the case when a
stash-like argument was supplied. Since, commit 9bf09e, 'stash show' when
supplied a stash-like argument prints nothing and still exists with a zero
status. Unfortunately, the flaw slipped through the test suite cracks
since the output of 'stash show' was not verified to be correct.
Improve and expand on the existing tests so that this flaws is detected.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The merge-base between @$parents and $merge_tip may have been reached
through a merge commit. This means that some commits that are ancestors
of @$parents will not be ancestors of $merge_base. The mergeinfo
property will not list commits that are ancestors of @$parents, so we
need to explicitly exclude them.
[ew: squashed and cleaned up test case from Steven]
Signed-off-by: Steven Walter <stevenrwalter@gmail.com>
Acked-by: Eric Wong <normalperson@yhbt.net>
When we're diffing symlinks, we consider the contents to be
the pathname that the symlink points to. When a user sets up
a userdiff driver like "*.pdf diff=pdf", their "diff.pdf.*"
config generally tells us what to do with the content of
pdf files.
With the current code, we will actually process a symlink
like "link.pdf" using a configured pdf driver, meaning we
are using contents which consist of a pathname with
configuration that is expecting contents that consist of an
actual pdf file.
The most noticeable example of this would have been
textconv; however, it was already protected in its own
textconv-specific code path. We can still see the breakage
with something like "diff.*.binary", though. You could
also see it with diff.*.funcname, though it is a bit harder
to trigger accidentally there.
This patch adds a check for S_ISREG lower in the callstack
than the textconv-specific check, which should block use of
any userdiff config for non-regular files. We can drop the
check in the textconv code, which is now redundant.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The ancient touch on Solaris 7 thinks that a decimal number supplied as
the first argument specifies a date_time to give to the files specified by
the remaining arguments. In this case, it fails to parse '1' as a proper
date_time and exits with a failure status. Workaround this flaw by
rearranging the arguments supplied to touch so that a non-digit appears
first and touch will not be confused.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jl/fix-test:
t1020: Get rid of 'cd "$HERE"' at the start of each test
t2016 (checkout -p): add missing &&
t1302 (core.repositoryversion): style tweaks
t2105 (gitfile): add missing &&
t1450 (fsck): remove dangling objects
tests: subshell indentation stylefix
Several tests: cd inside subshell instead of around
t7003-filter-branch.sh had a make_commit() function that was identical
to test_commit() in test-lib.sh except that it used tr to create a
lowercase file name from the uppercase branch name instead of
appending ".t".
Not only is this unneeded code duplication, it also was something
simply waiting to fail on case-insensitive file systems. So replace
all uses of make_commit with test_commit.
While we're editing the setup, chain it together with && so that
failures early in the sequence don't get lost and add a commit graph.
Signed-off-by: Brian Gernhardt <brian@gernhardtsoftware.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the olden days,
log --author=me --committer=him --grep=this --grep=that
used to be turned into:
(OR (HEADER-AUTHOR me)
(HEADER-COMMITTER him)
(PATTERN this)
(PATTERN that))
showing my patches that do not have any "this" nor "that", which was
totally useless.
80235ba ("log --author=me --grep=it" should find intersection, not union,
2010-01-17) improved it greatly to turn the same into:
(ALL-MATCH
(HEADER-AUTHOR me)
(HEADER-COMMITTER him)
(OR (PATTERN this) (PATTERN that)))
That is, "show only patches by me and committed by him, that have either
this or that", which is a lot more natural thing to ask.
We however need to be a bit more clever when the user asks more than one
"author" (or "committer"); because a commit has only one author (and one
committer), they ought to be interpreted as asking for union to be useful.
The current implementation simply added another author/committer pattern
at the same top-level for ALL-MATCH to insist on matching all, finding
nothing.
Turn
log --author=me --author=her \
--committer=him --committer=you \
--grep=this --grep=that
into
(ALL-MATCH
(OR (HEADER-AUTHOR me) (HEADER-AUTHOR her))
(OR (HEADER-COMMITTER him) (HEADER-COMMITTER you))
(OR (PATTERN this) (PATTERN that)))
instead.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Use "$GIT_BUILD_DIR" instead of "$TEST_DIRECTORY/.." (both defined in
t/test-lib.sh) in t/gitweb-lib.sh. It better describes the intent.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When applying two pathspecs, one of which is named as a prefix to the
other, we mistakenly recursed into the shorter one.
Noticed and fixed by David Reis.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This adds fortran xfuncname and wordRegex patterns to the list of builtin
patterns. The intention is for the patterns to be appropriate for all
versions of fortran including 77, 90, 95. The patterns can be enabled by
adding the diff=fortran attribute to the .gitattributes file for the
desired file glob.
This also adds a new macro named IPATTERN which is just like the PATTERNS
macro except it sets the REG_ICASE flag so that case will be ignored.
The test code in t4018 and the docs were updated as appropriate.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously (e3bf5e43), a test was added to test whether the builtin
xfuncname regular expressions could be compiled without error by regcomp.
Let's do the same for the word_regex patterns. This should help catch any
cross-platform incompatibilities that exist between the pattern creator's
system and the various platforms that the test suite is commonly run on.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit 6106ce46 introduced a test to demonstrate fetch's failure to
retrieve any objects or update FETCH_HEAD when it was supplied a repository
URL and the current branch had a configured merge spec. This commit
expands the original test based on comments from Junio Hamano. In addition
to actually verifying that the fetch updates FETCH_HEAD correctly, and does
not update the current branch, two more tests are added to ensure that the
merge configuration is ignored even when the supplied URL matches the URL
of the remote configured for the branch.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
On IRIX 6.5, the printf utility in /usr/bin does not appear to handle the
\ddd notation according to POSIX. This printf appears to halt processing
of the string argument and ignore any additional characters in the string.
Work around this flaw by replacing the \000's with 'Q' and using the
q_to_nul helper function provided by test-lib.sh
This problem with printf is not apparent when using the Bash shell since
Bash implements a POSIX compatible printf function internally.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In merge-recursive.c, whenever there was a rename where a file name on one
side of the rename matches a directory name on the other side of the merge,
then the very first check that
string_list_has_string(&o->current_directory_set, ren1_dst)
would trigger forcing it into marking it as a rename/directory conflict.
However, if the path is only renamed on one side and a simple three-way
merge between the separate files resolves cleanly, then we don't need to
mark it as a rename/directory conflict. So, we can simply move the check
for rename/directory conflicts after we've verified that there isn't a
rename/rename conflict and that a threeway content merge doesn't work.
This changes the particular error message one gets in the case where the
directory name that a file on one side of the rename matches is not also
part of the rename pair. For example, with commits containing the files:
COMMON -> (HEAD, MERGE )
--------- --------------- -------
sub/file1 -> (sub/file1, newsub)
<NULL> -> (newsub/newfile, <NULL>)
then previously when one tried to merge MERGE into HEAD, one would get
CONFLICT (rename/directory): Rename sub/file1->newsub in HEAD directory newsub added in merge
Renaming sub/file1 to newsub~HEAD instead
Adding newsub/newfile
Automatic merge failed; fix conflicts and then commit the result.
After this patch, the error message will instead become:
Removing newsub
Adding newsub/newfile
CONFLICT (file/directory): There is a directory with name newsub in merge. Adding newsub as newsub~HEAD
Automatic merge failed; fix conflicts and then commit the result.
That makes more sense to me, because git can't know that there's a conflict
until after it's tried resolving paths involving newsub/newfile to see if
they are still in the way at the end (and if newsub/newfile is not in the
way at the end, there should be no conflict at all, which did not hold with
git previously).
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When one side of a file rename matches a directory name on the other side,
the recursive merge strategy will fail. This is true even if the merge is
trivially resolvable.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
