Andrew Sayers noticed that the svn-fe | git fast-import pipeline
mishandles a subversion history that copies the root directory to a
sub-directory (e.g. doing `svn cp . trunk` to standardise your
layout). As David Barr explained, the bug arises when the following
command is sent to git fast-import:
'ls' SP ':1' SP LF
Instead of reading back what is at the root of r1, it unconditionally
reports the path as missing.
After sleeping on it, here are two patches for 'maint'. One plugs a
memory leak. The other ensures that trying to pass an empty path to
the 'ls' command results in an error message that can help the
frontend author instead of the silently broken conversion Andrew
found.
Then we can carefully add 'ls ""' support in 1.7.11.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)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=wrup
-----END PGP SIGNATURE-----
Merge "two fixes for fast-import's 'ls' command" from Jonathan
Andrew Sayers noticed that the svn-fe | git fast-import pipeline
mishandles a subversion history that copies the root directory to a
sub-directory (e.g. doing `svn cp . trunk` to standardise your
layout). As David Barr explained, the bug arises when the following
command is sent to git fast-import:
'ls' SP ':1' SP LF
Instead of reading back what is at the root of r1, it unconditionally
reports the path as missing.
After sleeping on it, here are two patches for 'maint'. One plugs a
memory leak. The other ensures that trying to pass an empty path to
the 'ls' command results in an error message that can help the
frontend author instead of the silently broken conversion Andrew
found.
Then we can carefully add 'ls ""' support in 1.7.11.
* commit 'refs/pull-request-tags/jn/maint-fast-import-empty-ls':
fast-import: don't allow 'ls' of path with empty components
fast-import: leakfix for 'ls' of dirty trees
remove_dir_recursively() has a check to avoid removing the directory it
was asked to remove without recursing into it and report success when the
directory is the top level of a working tree of a nested git repository,
to protect such a repository from "clean -f" (without double -f). If a
working tree of a nested git repository is in a subdirectory of a toplevel
project, however, this protection did not apply by mistake; we forgot to
pass the REMOVE_DIR_KEEP_NESTED_GIT down to the recursive removal
codepath.
This requires us to also teach the higher level not to remove the
directory it is asked to remove, when the recursed invocation did not
remove the directory it was asked to remove due to a nested git
repository, as it is not an error to leave the parent directories of such
a nested repository.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a manual notes merge is committed or aborted, we need to remove the
temporary worktree at .git/NOTES_MERGE_WORKTREE. However, removing the
entire directory is not good if the user ran the 'git notes merge
--commit/--abort' from within that directory. On Windows, the directory
removal would simply fail, while on POSIX systems, users would suddenly
find themselves in an invalid current directory.
Therefore, instead of deleting the entire directory, we delete everything
_within_ the directory, and leave the (empty) directory in place.
This would cause a subsequent notes merge to abort, complaining about a
previous - unfinished - notes merge (due to the presence of
.git/NOTES_MERGE_WORKTREE), so we also need to adjust this check to only
trigger when .git/NOTES_MERGE_WORKTREE is non-empty.
Finally, adjust the t3310 manual notes merge testcases to correctly handle
the existence of an empty .git/NOTES_MERGE_WORKTREE directory.
Inspired-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
notes_merge_commit() only needs to list all entries (non-recursively)
under a directory, which can be easily accomplished with
opendir/readdir and would be more lightweight than read_directory().
read_directory() is designed to list paths inside a working
directory. Using it outside of its scope may lead to undesired effects.
Apparently, one of the undesired effects of read_directory() is that it
doesn't deal with being given absolute paths. This creates problems for
notes_merge_commit() when git_path() returns an absolute path, which
happens when the current working directory is in a subdirectory of the
.git directory.
Originally-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Updated-by: Johan Herland <johan@herland.net>
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The 'git notes merge' command expected to be run from the working
tree of the project being annotated, and did not anticipate getting
run inside $GIT_DIR/.
However, because we use $GIT_DIR/NOTES_MERGE_WORKTREE as a temporary
working space for the user to work on resolving conflicts, it is not
unreasonable for a user to run "git notes merge --commit" there. But
the command fails to do so.
Found-by: David Bremner <david@tethera.net>
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When using word diff, the code sets the word_regex from various
defaults if it was not set already. The problem is that it does this
on the original diff_options, which will also be used in subsequent
diffs.
This means that when the word_regex is not given on the command line,
only the first diff for which a setting for word_regex (either from
attributes or diff.wordRegex) ever takes effect. This value then
propagates to the rest of the diff runs and in particular prevents
further attribute lookups.
Fix the problem of changing diff state once and for all, by working
with a _copy_ of the diff_options.
Noticed-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The test case applies a custom wordRegex to one file in a diff, and expects
that the default word splitting applies to the second file in the diff.
But the custom wordRegex is also incorrectly used for the second file.
Helped-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It was unclear what a test in t0204 wanted to check; it turns out
that it was only to observe an undefined behaviour of the system,
and did not anticipate one kind of reasonable error behaviour.
* jc/maint-undefined-i18n-observation-test:
t0204: clarify the "observe undefined behaviour" test
When "git config" diagnoses an error in a configuration file and
shows the line number for the offending line, it miscounted if the
error was at the end of line.
By Martin Stenberg
* ms/maint-config-error-at-eol-linecount:
config: report errors at the EOL with correct line number
Conflicts:
t/t1300-repo-config.sh
"git bundle" did not record boundary commits correctly when there
are many of them.
By Thomas Rast
* tr/maint-bundle-boundary:
bundle: keep around names passed to add_pending_object()
t5510: ensure we stay in the toplevel test dir
t5510: refactor bundle->pack conversion
"git diff-index" and its friends at the plumbing level showed the
"diff --git" header and nothing else for a path whose cached stat
info is dirty without actual difference when asked to produce a
patch. This was a longstanding bug that we could have fixed long
time ago.
By Junio C Hamano
* jc/maint-diff-patch-header:
diff -p: squelch "diff --git" header for stat-dirty paths
t4011: illustrate "diff-index -p" on stat-dirty paths
t4011: modernise style
The code to synthesize the fake ancestor tree used by 3-way merge
fallback in "git am" was not prepared to read a patch created with
a non-standard -p<num> value.
* jc/am-3-nonstandard-popt:
test: "am -3" can accept non-standard -p<num>
am -3: allow nonstandard -p<num> option
A section in a config file with a missing "]" reports the next line
as bad, same goes to a value with a missing end quote.
This happens because the error is not detected until the end of the
line, when line number is already increased. Fix this by decreasing
line number by one for these cases.
Signed-off-by: Martin Stenberg <martin@gnutiken.se>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When lying the author name via GIT_AUTHOR_NAME environment variable
to "git commit", the hooks run by the command saw it and could act
on the name that will be recorded in the final commit. When the user
uses the "--author" option from the command line, the command should
give the same information to the hook, and back when "git command"
was a scripted Porcelain, it did set the environment variable and
hooks can learn the author name from it.
However, when the command was reimplemented in C, the rewritten code
was not very faithful to the original, and hooks stopped getting the
authorship information given with "--author". Fix this by exporting
the necessary environment variables.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "--author" option is used to lie the authorship to "git commit"
command, hooks should learn the author name and email just like when
GIT_AUTHOR_NAME and GIT_AUTHOR_EMAIL environment variables are used
to lie the authorship. Test this.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As the fast-import manual explains:
The value of <path> must be in canonical form. That is it must
not:
. contain an empty directory component (e.g. foo//bar is invalid),
. end with a directory separator (e.g. foo/ is invalid),
. start with a directory separator (e.g. /foo is invalid),
Unfortunately the "ls" command accepts these invalid syntaxes and
responds by declaring that the indicated path is missing. This is too
subtle and causes importers to silently misbehave; better to error out
so the operator knows what's happening.
The C, R, and M commands already error out for such paths.
Reported-by: Andrew Sayers <andrew-git@pileofstuff.org>
Analysis-by: David Barr <davidbarr@google.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
This test asks for an impossible conversion to the system by
preparing an UTF-8 translation with characters that cannot be
expressed in ISO-8859-1, and then asking the message shown in
ISO-8859-1. Even though the behaviour against such a request is
undefined, it may be interesting to see what the system does, and
the purpose of this test is to see if there are platforms that
exhibit behaviour that we haven't seen.
The original recognized two known modes of behaviour:
- the key used to query the message catalog ("TEST: Old English
Runes"), saying "I cannot do that i18n".
- impossible characters replaced with ASCII "?", saying "I punt".
but they were treated totally differently. The test simply issued
an informational message "Your system punts on this one" for the
first error mode, while it diagnosed the latter as "Your system is
good; you pass!".
It turns out that Mac OS X exhibits a third mode of error behaviour,
to spew out the raw value stored in the message catalog. The test
diagnosed this behaviour as "broken", but it is merely trying to do
its best to respond to an impossible request by saying "I punt" in a
way that is slightly different from the second one.
Update the offending test to make it clear what is (and is not)
being tested, update the code structure so that newly discovered
error mode can easily be added to it later, and reword the message
that comes from a failing case to clarify that it is not the system
that is broken when it fails, but merely that the behaviour is not
something we have seen.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The 'log -3000 (baseline)' test accidentally still used -1000 from an
earlier version.
Noticed-by: Lawrence Holding <Lawrence.Holding@cubic.com>
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By Jens Lehmann (3) and Johannes Sixt (1)
* jl/maint-submodule-relative:
submodules: fix ambiguous absolute paths under Windows
submodules: refactor computation of relative gitdir path
submodules: always use a relative path from gitdir to work tree
submodules: always use a relative path to gitdir
The only bug right now is that $GIT_TEST_CMP is needed for test_cmp to
work.
However, we also export the three most important paths for tests:
TEST_DIRECTORY
TRASH_DIRECTORY
GIT_BUILD_DIR
Since they are available within test_expect_success, a future test
writer may expect them to also be defined in test_perf.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Loading it in the subshells still referred to $TEST_DIRECTORY/..,
which was only correct in preliminary versions of perf-lib.sh
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By Junio C Hamano (2) and Ramsay Jones (1)
* jc/pickaxe-ignore-case:
ctype.c: Fix a sparse warning
pickaxe: allow -i to search in patch case-insensitively
grep: use static trans-case table
By Junio C Hamano
* jc/maint-diff-patch-header:
diff -p: squelch "diff --git" header for stat-dirty paths
t4011: illustrate "diff-index -p" on stat-dirty paths
t4011: modernise style
By Thomas Rast
* tr/maint-bundle-boundary:
bundle: keep around names passed to add_pending_object()
t5510: ensure we stay in the toplevel test dir
t5510: refactor bundle->pack conversion
By Zbigniew Jędrzejewski-Szmek (8) and Junio C Hamano (1)
* zj/diff-stat-dyncol:
: This breaks tests. Perhaps it is not worth using the decimal-width stuff
: for this series, at least initially.
diff --stat: add config option to limit graph width
diff --stat: enable limiting of the graph part
diff --stat: add a test for output with COLUMNS=40
diff --stat: use a maximum of 5/8 for the filename part
merge --stat: use the full terminal width
log --stat: use the full terminal width
show --stat: use the full terminal width
diff --stat: use the full terminal width
diff --stat: tests for long filenames and big change counts
OS X's sed and grep would complain with (respectively)
sed: 1: "/^-/{p;q}": extra characters at the end of q command
grep: Regular expression too big
For sed, use an explicit ; to terminate the q command.
For grep, spell the "40 hex digits" explicitly in the regex, which
should be safe as other tests already use this and we haven't got
breakage reports on OS X about them.
Signed-off-by: Thomas Rast <trast@inf.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
8c912ee (teach --histogram to diff, 2011-07-12) claimed histogram diff
was faster than both Myers and patience.
We have since incorporated a performance testing framework, so add a
test that compares the various diff tasks performed in a real 'log -p'
workload. This does indeed show that histogram diff slightly beats
Myers, while patience is much slower than the others.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The current message is too long and at too low a level for anybody
to understand it if they don't know about the configuration format
already.
The text about setting up a remote is superfluous and doesn't help
understand or recover from the error that has happened. Show the
usage more prominently and explain how to set up the tracking
information. If there is only one remote, that name is used instead
of the generic <remote>.
Also simplify the message we print on detached HEAD to remove
unnecessary information which is better left for the documentation.
Signed-off-by: Carlos Martín Nieto <cmn@elego.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint:
Update draft release notes to 1.7.9.3 for the last time
http.proxy: also mention https_proxy and all_proxy
t0300: work around bug in dash 0.5.6
t5512 (ls-remote): modernize style
tests: fix spurious error when run directly with Solaris /usr/xpg4/bin/sh
* jk/maint-avoid-streaming-filtered-contents:
do not stream large files to pack when filters are in use
teach dry-run convert_to_git not to require a src buffer
teach convert_to_git a "dry run" mode
* tr/maint-bundle-long-subject:
t5704: match tests to modern style
strbuf: improve strbuf_get*line documentation
bundle: use a strbuf to scan the log for boundary commits
bundle: put strbuf_readline_fd in strbuf.c with adjustments
In module_clone() the rel_gitdir variable was computed differently when
"git rev-parse --git-dir" returned a relative path than when it returned
an absolute path. This is not optimal, as different code paths are used
depending on the return value of that command.
Fix that by reusing the differing path components computed for setting the
core.worktree config setting, which leaves a single code path for setting
both instead of having three and makes the code much shorter.
This also fixes the bug that in the computation of how many directories
have to be traversed up to hit the root directory of the submodule the
name of the submodule was used where the path should have been used. This
lead to problems after renaming submodules into another directory level.
Even though the "(cd $somewhere && pwd)" approach breaks the flexibility
of symlinks, that is no issue here as we have to have one relative path
pointing from the work tree to the gitdir and another pointing back, which
will never work anyway when a symlink along one of those paths is changed
because the directory it points to was moved.
Also add a test moving a submodule into a deeper directory to catch any
future breakage here and to document what has to be done when a submodule
needs to be moved until git mv learns to do that. Simply moving it to the
new location doesn't work, as the core.worktree and possibly the gitfile
setting too will be wrong. So it has to be removed from filesystem and
index, then the new location has to be added into the index and the
.gitmodules file has to be updated. After that a git submodule update will
check out the submodule at the new location.
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since recently a submodule with name <name> has its git directory in the
.git/modules/<name> directory of the superproject while the work tree
contains a gitfile pointing there. To make that work the git directory has
the core.worktree configuration set in its config file to point back to
the work tree.
That core.worktree is an absolute path set by the initial clone of the
submodule. A relative path is preferable here because it allows the
superproject to be moved around without invalidating that setting, so
compute and set that relative path after cloning or reactivating the
submodule.
This also fixes a bug when moving a submodule around inside the
superproject, as the current code forgot to update the setting to the new
submodule work tree location.
Enhance t7400 to ensure that future versions won't re-add absolute paths
by accident and that moving a superproject won't break submodules.
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since recently a submodule with name <name> has its git directory in the
.git/modules/<name> directory of the superproject while the work tree
contains a gitfile pointing there. When the submodule git directory needs
to be cloned because it is not found in .git/modules/<name> the clone
command will write an absolute path into the gitfile. When no clone is
necessary the git directory will be reactivated by the git-submodule.sh
script by writing a relative path into the gitfile.
This is inconsistent, as the behavior depends on the submodule having been
cloned before into the .git/modules of the superproject. A relative path
is preferable here because it allows the superproject to be moved around
without invalidating the gitfile. We do that by always writing the
relative path into the gitfile, which overwrites the absolute path the
clone command may have written there.
This is only the first step to make superprojects movable again like they
were before the separate-git-dir approach was introduced. The second step
is to use a relative path in core.worktree too.
Enhance t7400 to ensure that future versions won't re-add absolute paths
by accident.
While at it also replace an if/else construct evaluating the presence
of the 'reference' option with a single line of bash code.
Reported-by: Antony Male <antony.male@gmail.com>
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The user specifies a long option but forgets to type the second
leading dash, we currently detect and report that fact if its first
letter is a valid short option. This is done for safety, to avoid
ambiguity between short options (and their arguments) and a long
option with a missing dash.
This diagnostic message is also helpful for long options whose first
letter is not a valid short option, however. Print it in that case,
too, as a courtesy.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The construct 'while IFS== read' makes dash 0.5.6 execute
read without changing IFS, which results in test breakages
all over the place in t0300. Neither dash 0.5.5.1 and older
nor dash 0.5.7 and newer are affected: The problem was
introduded resp. fixed by the commits
55c46b7 ([BUILTIN] Honor tab as IFS whitespace when
splitting fields in readcmd, 2009-08-11)
1d806ac ([VAR] Do not poplocalvars prematurely on regular
utilities, 2010-05-27)
in http://git.kernel.org/?p=utils/dash/dash.git
Putting 'IFS==' before that line makes all versions of dash
work.
This looks like a dash bug, not a misinterpretation of the
standard. However, it's worth working around for two
reasons. One, this version of dash was released in Fedora
14-16, so the bug is found in the wild. And two, at least
one other shell, Solaris /bin/sh, choked on this by
persisting IFS after the read invocation. That is not a
shell we usually care about, and I think this use of IFS is
acceptable by POSIX (which allows other behavior near
"special builtins", but "read" is not one of those). But it
seems that this may be a subtle, not-well-tested case for
some shells. Given that the workaround is so simple, it's
worth just being defensive.
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Prepare expected output inside test_expect_success that uses it.
Also remove excess blank lines.
Signed-off-by: Tom Grennan <tmgrennan@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Match the style to more modern test scripts, namely:
- Prefer tabs for indentation.
- The first line of each test has prereq, title and opening sq for the
script body.
- Move cleanup or initialization of data used by a test inside the test
itself.
- Put a newline before the closing sq for each test.
- Don't conclude the test descriptions with a full stop.
- Prefer 'test_line_count = COUNT FILE' over 'test $(wc -l <FILE) = COUNT'
- Prefer 'test_line_count = 0 FILE' over 'cmp -s /dev/null FILE'
- Use '<<-EOF' style for here documents, so that they can be indented
as well. Bot don't do that in case the resulting lines would be too
long. Also when there is no $variable_substitution in the body of a
here document, quote \EOF.
- Don't redirect the output of commands to /dev/null unconditionally,
the git testing framework should already take care of handling test
verbosity transparently and uniformly.
Signed-off-by: Stefano Lattarini <stefano.lattarini@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If any test script is run directly with Solaris 10 /usr/xpg4/bin/sh or
/bin/ksh, it fails spuriously with a message like:
t0000-basic.sh[31]: unset: bad argument count
This happens because those shells bail out when encountering a call to
"unset" with no arguments, and such unset call could take place in
'test-lib.sh'. Fix that issue, and add a proper comment to ensure we
don't regress in this respect.
Signed-off-by: Stefano Lattarini <stefano.lattarini@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The 'name' field passed to add_pending_object() is used to later
deduplicate in object_array_remove_duplicates().
git-bundle had a bug in this area since 18449ab (git-bundle: avoid
packing objects which are in the prerequisites, 2007-03-08): it passed
the name of each boundary object in a static buffer. In other words,
all that object_array_remove_duplicates() saw was the name of the
*last* added boundary object.
The recent switch to a strbuf in bc2fed4 (bundle: use a strbuf to scan
the log for boundary commits, 2012-02-22) made this slightly worse: we
now free the buffer at the end, so it is not even guaranteed that it
still points into addressable memory by the time object_array_remove_
duplicates looks at it. On the plus side however, it was now
detectable by valgrind.
The fix is easy: pass a copy of the string to add_pending_object.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The last test descended into a subdir without ever re-emerging, which
is not so nice to the next test writer.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It's not so much a conversion as a "strip everything up to and
including the first blank line", but it will come in handy again.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The plumbing "diff" commands look at the working tree files without
refreshing the index themselves for performance reasons (the calling
script is expected to do that upfront just once, before calling one or
more of them). In the early days of git, they showed the "diff --git"
header before they actually ask the xdiff machinery to produce patches,
and ended up showing only these headers if the real contents are the same
and the difference they noticed was only because the stat info cached in
the index did not match that of the working tree. It was too late for the
implementation to take the header that it already emitted back.
But 3e97c7c (No diff -b/-w output for all-whitespace changes, 2009-11-19)
introduced necessary logic to keep the meta-information headers in a
strbuf and delay their output until the xdiff machinery noticed actual
changes. This was primarily in order to generate patches that ignore
whitespaces. When operating under "-w" mode, we wouldn't know if the
header is needed until we actually look at the resulting patch, so it was
a sensible thing to do, but we did not realize that the same reasoning
applies to stat-dirty paths.
Later, 296c6bb (diff: fix "git show -C -C" output when renaming a binary
file, 2010-05-26) generalized this machinery and added must_show_header
toggle. This is turned on when the header must be shown even when there
is no patch to be produced, e.g. only the mode was changed, or the path
was renamed, without changing the contents. However, when it did so, it
still kept the special case for the "-w" mode, which meant that the
plumbing would keep showing these phantom changes.
This corrects this historical inconsistency by allowing the plumbing to
omit paths that are only stat-dirty from its output in the same way as it
handles whitespace only changes under "-w" option.
The change in the behaviour can be seen in the updated test.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The plumbing that looks at the working tree, i.e. "diff-index" and
"diff-files", always emit the "diff --git a/path b/path" header lines
without anything else for paths that are only stat-dirty (i.e. different
only because the cached stat information in the index no longer matches
that of the working tree, but the real contents are the same), when
these commands are run with "-p" option to produce patches.
Illustrate this current behaviour. Also demonstrate that with the "-w"
option, we (correctly) hold off showing a "diff --git" header until actual
differences have been found. This also suppresses the header for merely
stat-dirty files, which is inconsistent.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Match the style to more modern test scripts, namely:
- The first line of each test has prereq, title and opening sq for the
script body. This makes the test shorter while reducing the need for
backslashes.
- Be prepared for the case in which the previous test may have failed.
If a test wants to start from not having "frotz" that the previous test
may have created, write "rm -f frotz", not "rm frotz".
- Prepare the expected output inside your own test.
- The order of comparison to check the result is "diff expected actual",
so that the output will show how the output from the git you just broke
is different from what is expected.
- Write no SP between redirection '>' (or '<' for that matter) and the
filename.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Config option diff.statGraphWidth=<width> is equivalent to
--stat-graph-width=<width>, except that the config option is ignored
by format-patch.
For the graph-width limiting to be usable, it should happen
'automatically' once configured, hence the config option.
Nevertheless, graph width limiting only makes sense when used on a
wide terminal, so it should not influence the output of format-patch,
which adheres to the 80-column standard.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A new option --stat-graph-width=<width> can be used to limit the width
of the graph part even is more space is available. Up to <width>
columns will be used for the graph.
If commits changing a lot of lines are displayed in a wide terminal
window (200 or more columns), and the +- graph uses the full width,
the output can be hard to comfortably scan with a horizontal movement
of human eyes. Messages wrapped to about 80 columns would be
interspersed with very long +- lines. It makes sense to limit the
width of the graph part to a fixed value (e.g. 70 columns), even if
more columns are available.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In preparation for the introduction on the limit of the width of the
graph part, a new test with COLUMNS=40 is added to check that the
environment variable influences diff, show, log, but not format-patch.
A new test is added because limiting the graph part makes COLUMNS=200
stop influencing diff --stat behaviour, which isn't wide enough now.
The old test with COLUMNS=200 is retained to check for regressions.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The way that available columns are divided between the filename part
and the graph part is modified to use as many columns as necessary for
the filenames and the rest for the graph.
If there isn't enough columns to print both the filename and the
graph, at least 5/8 of available space is devoted to filenames. On a
standard 80 column terminal, or if not connected to a terminal and
using the default of 80 columns, this gives the same partition as
before.
The effect of this change is visible in the patch to the test vector
in t4052; with a small change with long filename, it stops truncating
the name part too short, and also allocates a bit more columns to the
graph for larger changes.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make merge --stat behave like diff --stat and use the full terminal
width.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make log --stat behave like diff --stat and use the full terminal
width.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make show --stat behave like diff --stat and use the full terminal
width.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Default to the real terminal width for diff --stat output, instead
of the hard-coded 80 columns.
Some projects (especially in Java), have long filename paths, with
nested directories or long individual filenames. When files are
renamed, the filename part in stat output can be almost useless. If
the middle part between { and } is long (because the file was moved to
a completely different directory), then most of the path would be
truncated.
It makes sense to detect and use the full terminal width and display
full filenames if possible.
The are commands like diff, show, and log, which can adapt the output
to the terminal width. There are also commands like format-patch,
whose output should be independent of the terminal width. Since it is
safer to use the 80-column default, the real terminal width is only
used if requested by the calling code by setting diffopts.stat_width=-1.
Normally this value is 0, and can be set by the user only to a
non-negative value, so -1 is safe to use internally.
This patch only changes the diff builtin to use the full terminal width.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In preparation for updates to the "diff --stat" that updates the logic
to split the allotted columns into the name part and the graph part to
make the output more readable, add a handful of tests to document the
corner case behaviour in which long filenames and big changes are shown.
When a pathname is so long that it cannot fit on the column, the current
code truncates it to make sure that the graph part has enough room to show
a meaningful graph. If the actual change is small (e.g. only one line
changed), this results in the final output that is shorter than the width
we aim for.
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git log -S<string>" is a useful way to find the last commit in the
codebase that touched the <string>. As it was designed to be used by a
porcelain script to dig the history starting from a block of text that
appear in the starting commit, it never had to look for anything but an
exact match.
When used by an end user who wants to look for the last commit that
removed a string (e.g. name of a variable) that he vaguely remembers,
however, it is useful to support case insensitive match.
When given the "--regexp-ignore-case" (or "-i") option, which originally
was designed to affect case sensitivity of the search done in the commit
log part, e.g. "log --grep", the matches made with -S/-G pickaxe search is
done case insensitively now.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The default output from "fsck" is often overwhelmed by informational
message on dangling objects, especially if you do not repack often, and a
real error can easily be buried.
Add "--no-dangling" option to omit them, and update the user manual to
demonstrate its use.
Based on a patch by Clemens Buchacher, but reverted the part to change
the default to --no-dangling, which is unsuitable for the first patch.
The usual three-step procedure to break the backward compatibility over
time needs to happen on top of this, if we were to go in that direction.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When using regexp search ('sr' parameter / $search_use_regexp variable
is true), check first that regexp is valid.
Without this patch we would get an error from Perl during search (if
searching is performed by gitweb), or highlighting matches substring
(if applicable), if user provided invalid regexp... which means broken
HTML, with error page (including HTTP headers) generated after gitweb
already produced some output.
Add test that illustrates such error: for example for regexp "*\.git"
we would get the following error:
Quantifier follows nothing in regex; marked by <-- HERE in m/* <-- HERE \.git/
at /var/www/cgi-bin/gitweb.cgi line 3084.
Reported-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Without the magic line, prove shows lots and lots of errors:
% prove ./t9804-git-p4-label.sh
./t9804-git-p4-label.sh .. syntax error at ./t9804-git-p4-label.sh line 3, near ". ."
...
When #!/bin/sh is added, tests are skipped (I have no p4d).
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Earlier, fba4f1 (grep -P: Fix matching ^ and $) fixed an ancient bug. Add
some tests to protect the change from future breakages; a slightly broken
version of this was a part of the originally submitted patch.
Signed-off-by: Michał Kiedrowicz <michal.kiedrowicz@gmail.com>
Signed-off-by: Zbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
print_ref_list looks up the merge_filter_ref and assumes that a valid
pointer is returned. When the object doesn't exist, it tries to
dereference a NULL pointer. This can be the case when git branch
--merged is given an argument that isn't a valid commit name.
Check whether the lookup returns a NULL pointer and die with an error
if it does. Add a test, while we're at it.
Signed-off-by: Carlos Martín Nieto <cmn@elego.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* tr/maint-bundle-long-subject:
t5704: match tests to modern style
strbuf: improve strbuf_get*line documentation
bundle: use a strbuf to scan the log for boundary commits
bundle: put strbuf_readline_fd in strbuf.c with adjustments
* fc/push-prune:
push: add '--prune' option
remote: refactor code into alloc_delete_ref()
remote: reorganize check_pattern_match()
remote: use a local variable in match_push_refs()
Conflicts:
builtin/push.c
* jk/maint-avoid-streaming-filtered-contents:
do not stream large files to pack when filters are in use
teach dry-run convert_to_git not to require a src buffer
teach convert_to_git a "dry run" mode
This adds a test for the previous one to make sure that "am -3 -p0" can
read patches created with the --no-prefix option.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint:
Document accumulated fixes since 1.7.9.2
Git 1.7.8.5
grep -P: Fix matching ^ and $
am: don't infloop for an empty input file
rebase -m: only call "notes copy" when rewritten exists and is non-empty
git-p4: remove bash-ism in t9800
git-p4: remove bash-ism in t9809
git-p4: fix submit regression with clientSpec and subdir clone
git-p4: set useClientSpec variable on initial clone
Makefile: add thread-utils.h to LIB_H
Conflicts:
RelNotes
t/t9809-git-p4-client-view.sh
git-am.sh's check_patch_format function would attempt to preview
the patch to guess its format, but would go into an infinite loop
when the patch file happened to be empty. The solution: exit the
loop when "read" fails, not when the line var, "$l1" becomes empty.
Signed-off-by: Jim Meyering <meyering@redhat.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This works in both bash and dash:
$ bash -c 'VAR=1 env' | grep VAR
VAR=1
$ dash -c 'VAR=1 env' | grep VAR
VAR=1
But environment variables assigned this way are not necessarily propagated
through a function in POSIX compliant shells:
$ bash -c 'f() { "$@"
}; VAR=1 f "env"' | grep VAR
VAR=1
$ dash -c 'f() { "$@"
}; VAR=1 f "env"' | grep VAR
Fix constructs like this, in particular, setting variables through
test_must_fail.
Based-on-patch-by: Vitor Antunes <vitor.hda@gmail.com>
Signed-off-by: Pete Wyckoff <pw@padd.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Plain old $# works to count the number of arguments in
either bash or dash, even if the arguments have spaces.
Based-on-patch-by: Vitor Antunes <vitor.hda@gmail.com>
Signed-off-by: Pete Wyckoff <pw@padd.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the --use-client-spec is given to clone, and the clone
path is a subset of the full tree as specified in the client,
future submits will go to the wrong place.
Factor out getClientSpec() so both clone/sync and submit can
use it. Introduce getClientRoot() that is needed for the client
spec case, and use it instead of p4Where().
Test the five possible submit behaviors (add, modify, rename,
copy, delete).
Reported-by: Laurent Charrière <lcharriere@promptu.com>
Signed-off-by: Pete Wyckoff <pw@padd.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If --use-client-spec was given, set the matching configuration
variable. This is necessary to ensure that future submits
work properly.
The alternatives of requiring the user to set it, or providing
a command-line option on every submit, are error prone.
Signed-off-by: Pete Wyckoff <pw@padd.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Long options can be negated by adding no- right after the leading
two dashes. This is useful e.g. to override options set by aliases.
For options that are defined to start with no- already, this looks
a bit funny. Allow such options to also be negated by removing the
prefix.
The following thirteen options are affected:
apply --no-add
bisect--helper --no-checkout
checkout-index --no-create
clone --no-checkout --no-hardlinks
commit --no-verify --no-post-rewrite
format-patch --no-binary
hash-object --no-filters
read-tree --no-sparse-checkout
revert --no-commit
show-branch --no-name
update-ref --no-deref
The following five are NOT affected because they are defined with
PARSE_OPT_NONEG or the non-negated version is defined as well:
branch --no-merged
format-patch --no-stat --no-numbered
update-index --no-assume-unchanged --no-skip-worktree
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Introduce OPT_BOOL() to test-parse-options and add some tests for
these "true" boolean options. Rename OPT_BOOLEAN to OPT_COUNTUP and
OPTION_BOOLEAN to OPTION_COUNTUP as well.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Because git's object format requires us to specify the
number of bytes in the object in its header, we must know
the size before streaming a blob into the object database.
This is not a problem when adding a regular file, as we can
get the size from stat(). However, when filters are in use
(such as autocrlf, or the ident, filter, or eol
gitattributes), we have no idea what the ultimate size will
be.
The current code just punts on the whole issue and ignores
filter configuration entirely for files larger than
core.bigfilethreshold. This can generate confusing results
if you use filters for large binary files, as the filter
will suddenly stop working as the file goes over a certain
size. Rather than try to handle unknown input sizes with
streaming, this patch just turns off the streaming
optimization when filters are in use.
This has a slight performance regression in a very specific
case: if you have autocrlf on, but no gitattributes, a large
binary file will avoid the streaming code path because we
don't know beforehand whether it will need conversion or
not. But if you are handling large binary files, you should
be marking them as such via attributes (or at least not
using autocrlf, and instead marking your text files as
such). And the flip side is that if you have a large
_non_-binary file, there is a correctness improvement;
before we did not apply the conversion at all.
The first half of the new t1051 script covers these failures
on input. The second half tests the matching output code
paths. These already work correctly, and do not need any
adjustment.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The test did not adhere to the current style on several counts:
. empty lines around the test blocks, but within the test string
. ': > file' or even just '> file' with an extra space
. inconsistent indentation
. hand-rolled commits instead of using test_commit
Fix all of them.
There's a catch to the last point: test_commit creates a tag, which the
original test did not create. We still change it to test_commit, and
explicitly delete the tags, so as to highlight that the test relies on not
having them.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* ld/git-p4-expanded-keywords:
: Teach git-p4 to unexpand $RCS$-like keywords that are embedded in
: tracked contents in order to reduce unnecessary merge conflicts.
git-p4: add initial support for RCS keywords
* jk/config-include:
: An assignment to the include.path pseudo-variable causes the named file
: to be included in-place when Git looks up configuration variables.
config: add include directive
config: eliminate config_exclusive_filename
config: stop using config_exclusive_filename
config: provide a version of git_config with more options
config: teach git_config_rename_section a file argument
config: teach git_config_set_multivar_in_file a default path
config: copy the return value of prefix_filename
t1300: add missing &&-chaining
docs/api-config: minor clarifications
docs: add a basic description of the config API
* tr/perftest:
Add a performance test for git-grep
Introduce a performance testing framework
Move the user-facing test library to test-lib-functions.sh
RCS keywords cause problems for git-p4 as perforce always
expands them (if +k is set) and so when applying the patch,
git reports that the files have been modified by both sides,
when in fact they haven't.
This change means that when git-p4 detects a problem applying
a patch, it will check to see if keyword expansion could be
the culprit. If it is, it strips the keywords in the p4
repository so that they match what git is expecting. It then
has another go at applying the patch.
This behaviour is enabled with a new git-p4 configuration
option and is off by default.
Acked-by: Pete Wyckoff <pw@padd.com>
Signed-off-by: Luke Diamand <luke@diamand.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The first part of the bundle header contains the boundary commits, and
could be approximated by
# v2 git bundle
$(git rev-list --pretty=oneline --boundary <ARGS> | grep ^-)
git-bundle actually spawns exactly this rev-list invocation, and does
the grepping internally.
There was a subtle bug in the latter step: it used fgets() with a
1024-byte buffer. If the user has sufficiently long subjects (e.g.,
by not adhering to the git oneline-subject convention in the first
place), the 'oneline' format can easily overflow the buffer. fgets()
then returns the rest of the line in the next call(s). If one of
these remaining parts started with '-', git-bundle would mistakenly
insert it into the bundle thinking it was a boundary commit.
Fix it by using strbuf_getwholeline() instead, which handles arbitrary
line lengths correctly.
Note that on the receiving side in parse_bundle_header() we were
already using strbuf_getwholeline_fd(), so that part is safe.
Reported-by: Jannis Pohlmann <jannis.pohlmann@codethink.co.uk>
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When pushing groups of refs to a remote, there is no simple way to remove
old refs that still exist at the remote that is no longer updated from us.
This will allow us to remove such refs from the remote.
With this change, running this command
$ git push --prune remote refs/heads/*:refs/remotes/laptop/*
removes refs/remotes/laptop/foo from the remote if we do not have branch
"foo" locally anymore.
Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When cherry-pick fails it offers a helpful hint about how to
proceed. The hint tells the user how to do the cleanup
needed by the conflicted cherry-pick and finish the job after
conflict resolution. In case of cherry-pick --no-commit, the
hint goes too far. It tells the user to finish up with
'git commit'. That is not what this git-cherry-pick was
trying to do in the first place.
Restrict the hint in case of --no-commit to avoid giving this
extra advice.
Also, add a test verifying the reduced hint for the --no-commit
version of cherry-pick.
Signed-off-by: Phil Hord <hordp@cisco.com>
Acked-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* 'master' of git://bogomips.org/git-svn:
git-svn.perl: fix a false-positive in the "already exists" test
git-svn.perl: perform deletions before anything else
git-svn: Fix time zone in --localtime
git-svn: un-break "git svn rebase" when log.abbrevCommit=true
git-svn: remove redundant porcelain option to rev-list
completion: add --interactive option to git svn dcommit
* jk/grep-binary-attribute:
grep: pre-load userdiff drivers when threaded
grep: load file data after checking binary-ness
grep: respect diff attributes for binary-ness
grep: cache userdiff_driver in grep_source
grep: drop grep_buffer's "name" parameter
convert git-grep to use grep_source interface
grep: refactor the concept of "grep source" into an object
grep: move sha1-reading mutex into low-level code
grep: make locking flag global
open_or_add_dir checks to see if the directory already exists or not.
If it already exists and is not a directory, then we fail. However,
open_or_add_dir did not previously account for the possibility that the
path did exist as a file, but is deleted in the current commit.
In order to prevent this legitimate case from failing, open_or_add_dir
needs to know what files are deleted in the current commit.
Unfortunately that information has to be plumbed through a couple of
layers.
Signed-off-by: Steven Walter <stevenrwalter@gmail.com>
Acked-by: Eric Wong <normalperson@yhbt.net>
Commit ff7f218 (gitweb: Fix file links in "grep" search, 2012-01-05),
added $file_href variable, to reduce duplication and have the fix
applied in single place.
Unfortunately it made variable defined inside the loop, not taking into
account the fact that $file_href was set only if file changed.
Therefore for files with multiple matches $file_href was undefined for
second and subsequent matches.
Fix this bug by moving $file_href declaration outside loop.
Adds tests for almost all forms of sarch in gitweb, which were missing
from testuite. Note that it only tests if there are no warnings, and
it doesn't check that gitweb finds what it should find.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When running "git add --refresh <pathspec>", we incorrectly showed the
path that is unmerged even if it is outside the specified pathspec, even
though we did honor pathspec and refreshed only the paths that matched.
Note that this cange does not affect "git update-index --refresh"; for
hysterical raisins, it does not take a pathspec (it takes real paths) and
more importantly itss command line options are parsed and executed one by
one as they are encountered, so "git update-index --refresh foo" means
"first refresh the index, and then update the entry 'foo' by hashing the
contents in file 'foo'", not "refresh only entry 'foo'".
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a repository whose HEAD points to an unborn branch with no commits,
"heads" view and "summary" view (which shows what is shown in "heads"
view) compared the object names of commits at the tip of branches with the
output from "git rev-parse HEAD", which caused comparison of a string with
undef and resulted in a warning in the server log.
This can happen if non-bare repository (with default 'master' branch)
is updated not via committing but by other means like push to it, or
Gerrit. It can happen also just after running "git checkout --orphan
<new branch>" but before creating any new commit on this branch.
Rewrite the comparison so that it also works when $head points at nothing;
in such a case, no branch can be "the current branch", add a test for it.
While at it, rename local variable $head to $head_at, as it points to
current commit rather than current branch name (HEAD contents).
The code still incorrectly shows all branches that point at the same
commit as what HEAD points as "the current branch", even when HEAD is
detached. Fixing this bug is outside the scope of this patch.
Reported-by: Rajesh Boyapati
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The only catch is that we don't really know what our repo contains, so
we have to ignore any possible "not found" status from git-grep.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This introduces a performance testing framework under t/perf/. It
tries to be as close to the test-lib.sh infrastructure as possible,
and thus should be easy to get used to for git developers.
The following points were considered for the implementation:
1. You usually want to compare arbitrary revisions/build trees against
each other. They may not have the performance test under
consideration, or even the perf-lib.sh infrastructure.
To cope with this, the 'run' script lets you specify arbitrary
build dirs and revisions. It even automatically builds the revisions
if it doesn't have them at hand yet.
2. Usually you would not want to run all tests. It would take too
long anyway. The 'run' script lets you specify which tests to run;
or you can also do it manually. There is a Makefile for
discoverability and 'make clean', but it is not meant for
real-world use.
3. Creating test repos from scratch in every test is extremely
time-consuming, and shipping or downloading such large/weird repos
is out of the question.
We leave this decision to the user. Two different sizes of test
repos can be configured, and the scripts just copy one or more of
those (using hardlinks for the object store). By default it tries
to use the build tree's git.git repository.
This is fairly fast and versatile. Using a copy instead of a clone
preserves many properties that the user may want to test for, such
as lots of loose objects, unpacked refs, etc.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This just moves all the user-facing functions to a separate file and
sources that instead.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It can be useful to split your ~/.gitconfig across multiple
files. For example, you might have a "main" file which is
used on many machines, but a small set of per-machine
tweaks. Or you may want to make some of your config public
(e.g., clever aliases) while keeping other data back (e.g.,
your name or other identifying information). Or you may want
to include a number of config options in some subset of your
repos without copying and pasting (e.g., you want to
reference them from the .git/config of participating repos).
This patch introduces an include directive for config files.
It looks like:
[include]
path = /path/to/file
This is syntactically backwards-compatible with existing git
config parsers (i.e., they will see it as another config
entry and ignore it unless you are looking up include.path).
The implementation provides a "git_config_include" callback
which wraps regular config callbacks. Callers can pass it to
git_config_from_file, and it will transparently follow any
include directives, passing all of the discovered options to
the real callback.
Include directives are turned on automatically for "regular"
git config parsing. This includes calls to git_config, as
well as calls to the "git config" program that do not
specify a single file (e.g., using "-f", "--global", etc).
They are not turned on in other cases, including:
1. Parsing of other config-like files, like .gitmodules.
There isn't a real need, and I'd rather be conservative
and avoid unnecessary incompatibility or confusion.
2. Reading single files via "git config". This is for two
reasons:
a. backwards compatibility with scripts looking at
config-like files.
b. inspection of a specific file probably means you
care about just what's in that file, not a general
lookup for "do we have this value anywhere at
all". If that is not the case, the caller can
always specify "--includes".
3. Writing files via "git config"; we want to treat
include.* variables as literal items to be copied (or
modified), and not expand them. So "git config
--unset-all foo.bar" would operate _only_ on
.git/config, not any of its included files (just as it
also does not operate on ~/.gitconfig).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The git-config command sometimes operates on the default set
of config files (either reading from all, or writing to repo
config), and sometimes operates on a specific file. In the
latter case, we set the magic global config_exclusive_filename,
and the code in config.c does the right thing.
Instead, let's have git-config use the "advanced" variants
of config.c's functions which let it specify an individual
filename (or NULL for the default). This makes the code a
lot more obvious, and fixes two small bugs:
1. A relative path specified by GIT_CONFIG=foo will look
in the wrong directory if we have to chdir as part of
repository setup. We already handle this properly for
"git config -f foo", but the GIT_CONFIG lookup used
config_exclusive_filename directly. By dropping to a
single magic variable, the GIT_CONFIG case now just
works.
2. Calling "git config -f foo --edit" would not respect
core.editor. This is because just before editing, we
called git_config, which would respect the
config_exclusive_filename setting, even though this
particular git_config call was not about looking in the
user's specified file, but rather about loading actual
git config, just as any other git program would.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By default, a missing filter driver or a failure from the filter driver is
not an error, but merely makes the filter operation a no-op pass through.
This is useful to massage the content into a shape that is more convenient
for the platform, filesystem, and the user to use, and the content filter
mechanism is not used to turn something unusable into usable.
However, we could also use of the content filtering mechanism and store
the content that cannot be directly used in the repository (e.g. a UUID
that refers to the true content stored outside git, or an encrypted
content) and turn it into a usable form upon checkout (e.g. download the
external content, or decrypt the encrypted content). For such a use case,
the content cannot be used when filter driver fails, and we need a way to
tell Git to abort the whole operation for such a failing or missing filter
driver.
Add a new "filter.<driver>.required" configuration variable to mark the
second use case. When it is set, git will abort the operation when the
filter driver does not exist or exits with a non-zero status code.
Signed-off-by: Jehan Bing <jehan@orb.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jk/grep-binary-attribute:
grep: pre-load userdiff drivers when threaded
grep: load file data after checking binary-ness
grep: respect diff attributes for binary-ness
grep: cache userdiff_driver in grep_source
grep: drop grep_buffer's "name" parameter
convert git-grep to use grep_source interface
grep: refactor the concept of "grep source" into an object
grep: move sha1-reading mutex into low-level code
grep: make locking flag global
* nd/pack-objects-parseopt:
pack-objects: convert to use parse_options()
pack-objects: remove bogus comment
pack-objects: do not accept "--index-version=version,"
* mh/war-on-extra-refs:
refs: remove the extra_refs API
clone: do not add alternate references to extra_refs
everything_local(): mark alternate refs as complete
fetch-pack.c: inline insert_alternate_refs()
fetch-pack.c: rename some parameters from "path" to "refname"
clone.c: move more code into the "if (refs)" conditional
t5700: document a failure of alternates to affect fetch
* jk/maint-tag-show-fixes:
tag: do not show non-tag contents with "-n"
tag: die when listing missing or corrupt objects
tag: fix output of "tag -n" when errors occur
Conflicts:
t/t7004-tag.sh
Receive runs rev-list --verify-objects in order to detect missing
objects. However, such errors are ignored and overridden later.
Instead, consequently ignore all update commands for which an error has
already been detected.
Some tests in t5504 are obsoleted by this change, because invalid
objects are detected even if fsck is not enabled. Instead, they now test
for different error messages depending on whether or not fsck is turned
on. A better fix would be to force a corruption that will be detected by
fsck but not by rev-list.
Signed-off-by: Clemens Buchacher <drizzd@aon.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Otherwise the test cannot be run with custom port set to LIB_HTTPD_PORT.
Signed-off-by: Clemens Buchacher <drizzd@aon.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By default, progress output is disabled if stderr is not a terminal.
The --progress option can be used to force progress output anyways.
Conversely, --no-progress does not force progress output. In particular,
if stderr is a terminal, progress output is enabled.
This is unintuitive. Change --no-progress to force output off.
Signed-off-by: Clemens Buchacher <drizzd@aon.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git rev-list passes rev_list_info, not rev_list objects. Without this
fix, rev-list enables or disables the --verify-objects option depending
on a read from an undefined memory location.
Signed-off-by: Clemens Buchacher <drizzd@aon.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change several tests to use the sane_unset function introduced in
v1.7.3.1-35-g00648ba instead of the built-in unset function.
This fixes a failure I was having on t9130-git-svn-authors-file.sh on
Solaris, and prevents several other issues from occurring.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
I'm no longer running the Git smoke testing service at
smoke.git.nix.is due to Smolder being a fragile piece of software not
having time to follow through on making it easy for third parties to
run and submit their own smoke tests.
So remove the support in Git for sending smoke tests to
smoke.git.nix.is, it's still easy to modify the test suite to submit
smokes somewhere else.
This reverts the following commits:
Revert "t/README: Add SMOKE_{COMMENT,TAGS}= to smoke_report target" -- e38efac87d
Revert "t/README: Document the Smoke testing" -- d15e9ebc5c
Revert "t/Makefile: Create test-results dir for smoke target" -- 617344d77b
Revert "tests: Infrastructure for Git smoke testing" -- b6b84d1b74
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jk/maint-tag-show-fixes:
tag: do not show non-tag contents with "-n"
tag: die when listing missing or corrupt objects
tag: fix output of "tag -n" when errors occur
Conflicts:
t/t7004-tag.sh
Objects in an alternate object database are already available to the
local repository and therefore don't need to be fetched. So mark them
as complete in everything_local().
This fixes a test in t5700.
Signed-off-by: Michael Haggerty <mhagger@alum.mit.edu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If an alternate supplies some, but not all, of the objects needed for
a fetch, fetch-pack nevertheless generates "want" lines for the
alternate objects that are present. Demonstrate this problem via a
failing test.
Signed-off-by: Michael Haggerty <mhagger@alum.mit.edu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
