It can sometimes be useful to write information temporarily to file,
to read back later. These functions allow a program to use the
line_buffer facilities when doing so.
It works like this:
1. find a unique filename with buffer_tmpfile_init.
2. rewind with buffer_tmpfile_rewind. This returns a stdio
handle for writing.
3. when finished writing, declare so with
buffer_tmpfile_prepare_to_read. The return value indicates
how many bytes were written.
4. read whatever portion of the file is needed.
5. if finished, remove the temporary file with buffer_deinit.
otherwise, go back to step 2,
The svn support would use this to buffer the postimage from delta
application until the length is known and fast-import can receive
the resulting blob.
Based-on-patch-by: David Barr <david.barr@cordelta.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
buffer_read_char can be used in place of buffer_read_string(1) to
avoid consuming valuable static buffer space. The delta applier will
use this to read variable-length integers one byte at a time.
Underneath, it is fgetc, wrapped so the line_buffer library can
maintain its role as gatekeeper of input.
Later it might be worth checking if fgetc_unlocked is faster ---
most line_buffer functions are not thread-safe anyway.
Helpd-by: David Barr <david.barr@cordelta.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
buffer_read_string works well for non line-oriented input except for
one problem: it does not tell the caller how many bytes were actually
written. This means that unless one is very careful about checking
for errors (and eof) the calling program cannot tell the difference
between the string "foo" followed by an early end of file and the
string "foo\0bar\0baz".
So introduce a variant that reports the length, too, a thinner wrapper
around strbuf_fread. Its result is written to a strbuf so the caller
does not need to keep track of the number of bytes read.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
POSIX makes the behavior of read(2) from a pipe fairly clear: a read
from an empty pipe will block until there is data available and any
other read will not block, prefering to return a partial result.
Likewise, fread(3) and fgets(3) are clearly specified to act as
though implemented by calling fgetc(3) in a simple loop. But the
buffering behavior of fgetc is less clear.
Luckily, no sane platform is going to implement fgetc by calling the
equivalent of read(2) more than once. fgetc has to be able to
return without filling its buffer to preserve errno when errors are
encountered anyway. So let's assume the simpler behavior (trust) but
add some tests to catch insane platforms that violate that when they
come (verify).
First check that fread can handle a 0-length read from an empty fifo.
Because open(O_RDONLY) blocks until the writing end is open, open the
writing end of the fifo in advance in a subshell.
Next try short inputs from a pipe that is not filled all the way.
Lastly (two tests) try very large inputs from a pipe that will not fit
in the relevant buffers. The first of these tests reads a little
more than 8192 bytes, which is BUFSIZ (the size of stdio's buffers)
on this Linux machine. The second reads a little over 64 KiB (the
pipe capacity on Linux) and is not run unless requested by setting
the GIT_REMOTE_SVN_TEST_BIG_FILES environment variable.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Do not expect an implicit newline after each input record.
Use a separate command to exercise buffer_skip_bytes.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Split the line_buffer test into small pieces and move it to its
own file as preparation for adding more tests.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Imitate the input format of test-obj-pool to support arbitrary
sequences of commands rather than alternating read/copy. This should
make it easier to add tests that exercise other line_buffer functions.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Collect the line_buffer state in a newly public line_buffer struct.
Callers can use multiple line_buffers to manage input from multiple
files at a time.
svn-fe's delta applier will use this to stream a delta from svnrdump
and the preimage it applies to from fast-import at the same time.
The tests don't take advantage of the new features, but I think that's
okay. It is easier to find lingering examples of nonreentrant code by
searching for "static" in line_buffer.c.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Prepare for the line_buffer lib to support input from multiple files,
by collecting global state in a struct that can be easily passed
around.
No API change yet.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
obj_pool is inherently global and does not use the standard growing
factor alloc_nr, which makes it feel out of place in the git codebase.
Plus it is overkill for this application: all that is needed is a
buffer that can grow between requests to accomodate larger strings.
Use a strbuf instead.
As a side effect, this improves the error handling: allocation
failures will result in a clean exit instead of segfaults. It would
be nice to add a test case (using ulimit or failmalloc) but that can
wait for another day.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
The data stored in byte_buffer[] is always either discarded or
written to stdout immediately. No need for it to persist between
function calls.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Lazy fast-import frontend authors that want to rely on the backend to
keep track of the content of the imported trees _almost_ have what
they need in the 'cat-blob' command (v1.7.4-rc0~30^2~3, 2010-11-28).
But it is not quite enough, since
(1) cat-blob can be used to retrieve the content of files, but
not their mode, and
(2) using cat-blob requires the frontend to keep track of a name
(mark number or object id) for each blob to be retrieved
Introduce an 'ls' command to complement cat-blob and take care of the
remaining needs. The 'ls' command finds what is at a given path
within a given tree-ish (tag, commit, or tree):
'ls' SP <dataref> SP <path> LF
or in fast-import's active commit:
'ls' SP <path> LF
The response is a single line sent through the cat-blob channel,
imitating ls-tree output. So for example:
FE> ls :1 Documentation
gfi> 040000 tree 9e6c2b599341d28a2a375f8207507e0a2a627fe9 Documentation
FE> ls 9e6c2b599341d28a2a375f8207507e0a2a627fe9 git-fast-import.txt
gfi> 100644 blob 4f92954396e3f0f97e75b6838a5635b583708870 git-fast-import.txt
FE> ls :1 RelNotes
gfi> 120000 blob b942e49944 RelNotes
FE> cat-blob b942e49944
gfi> b942e49944 blob 32
gfi> Documentation/RelNotes/1.7.4.txt
The most interesting parts of the reply are the first word, which is
a 6-digit octal mode (regular file, executable, symlink, directory,
or submodule), and the part from the second space to the tab, which is
a <dataref> that can be used in later cat-blob, ls, and filemodify (M)
commands to refer to the content (blob, tree, or commit) at that path.
If there is nothing there, the response is "missing some/path".
The intent is for this command to be used to read files from the
active commit, so a frontend can apply patches to them, and to copy
files and directories from previous revisions.
For example, proposed updates to svn-fe use this command in place of
its internal representation of the repository directory structure.
This simplifies the frontend a great deal and means support for
resuming an import in a separate fast-import run (i.e., incremental
import) is basically free.
Signed-off-by: David Barr <david.barr@cordelta.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Improved-by: Junio C Hamano <gitster@pobox.com>
Improved-by: Sverre Rabbelier <srabbelier@gmail.com>
In a variable-args function, the code for writing into a strbuf is
non-trivial. We ended up cutting and pasting it in several places
because there was no vprintf-style function for strbufs (which in turn
was held up by a lack of va_copy).
Now that we have a fallback va_copy, we can add strbuf_vaddf, the
strbuf equivalent of vsprintf. And we can clean up the cut and paste
mess.
Signed-off-by: Jeff King <peff@peff.net>
Improved-by: Christian Couder <christian.couder@gmail.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
va_copy is C99. We have avoided using va_copy many times in the past,
which has led to a bunch of cut-and-paste. From everything I found
searching the web, implementations have historically either provided
va_copy or just let your code assume that simple assignment of worked.
So my guess is that this will be sufficient, though we won't really
know for sure until somebody reports a problem.
Signed-off-by: Jeff King <peff@peff.net>
Improved-by: Erik Faye-Lund <kusmabite@gmail.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In an attempt to overwrite the 'Description:' section of the p4 change
log to include the git commit messages, it also overwrote the 'Jobs:'
section. Â This fix restores the 'Job:' section.
Signed-off-by: Michael Horowitz <michael.horowitz@ieee.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When git-rebase.sh recently started sourcing
git-rebase--interactive.sh instead of executing it, executable bit of
the latter file should have been turned off and it should have been
moved from SCRIPT_SH to SCRIPT_LIB in the Makefile. Its two new
siblings, git-rebase--am.sh and git-rebase--merge.sh (whose executable
bits are already off) should also be moved to SCRIPT_LIB in the
Makefile.
Reported-by: Johannes Sixt <j6t@kdbg.org>
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Martin von Zweigbergk <martin.von.zweigbergk@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "git commit" was rewritten in C (v1.5.4-rc0~78^2~30,
2007-11-08), a subtle bug in --template was introduced. If the
file named by a --template parameter is missing, previously git
would error out with a message:
Commit template file does not exist.
but in the C version the --template parameter gets ignored and
the default template is used.
t7500 has two tests for this case which would have caught it, except
that with the default $EDITOR, the commit message template is left
unmodified, causing 'git commit' to error out and the test to
succeed.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When [g]vimdiff is called for files which are opened already, the editor
complains about the existing swap file. But we do not want to write
anything when called from difftool. So, make difftool use "-R" for the
vim family. This
- prevents the use of a swap file and
- marks the buffers readonly.
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
You can detach the HEAD at an arbitrary commit in order to browse the
files in various points in the history or build older versions of the
software, without recording any new commit, and come back to an existing
branch. When used in this "sightseer" mode, detached HEAD is a perfectly
safe mechanism. It also is a useful state to experiment with throw-away
commits.
When coming back to an existing branch with "git checkout master",
however, the commits that were created on the detached HEAD will become
unreachable from anywhere but the reflog of the HEAD. Check if the commit
we are about to leave is connected to some ref, and give a final warning
otherwise to remind the user for safety.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The user might have got used to the order the remotes appeared previously.
Lets add the all entry last so the all entry does not confuse previous
users.
Signed-off-by: Heiko Voigt <hvoigt@hvoigt.net>
Tested-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Pat Thoyts <patthoyts@users.sourceforge.net>
In case there is only one remote a fetch/prune all entry
is redundant.
Signed-off-by: Heiko Voigt <hvoigt@hvoigt.net>
Tested-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Pat Thoyts <patthoyts@users.sourceforge.net>
These should generally never happen, as we already
concatenate multiples in subjects into a single line. But
let's be defensive, since not encoding them means we will
output malformed headers.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Subject and identity headers may be arbitrarily long. In the
past, we just assumed that single-line headers would be
reasonably short. For multi-line subjects that we squish
into a single line, we just "pre-folded" the data in
pp_title_line by adding a newline and indentation.
There were two problems. One is that, although rare,
single-line messages can actually be longer than the
recommended line-length limits. The second is that the
pre-folding interacted badly with rfc2047 encoding, leading
to malformed headers.
Instead, let's stop pre-folding the subject lines, and just
fold everything based on length in add_rfc2047, whether
it is encoded or not.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The function strbuf_add_wrapped_text takes a NUL-terminated
string. This makes it annoying to wrap strings we have as a
pointer and a length.
Refactoring strbuf_add_wrapped_text and all of its
sub-functions to handle fixed-length strings turned out to
be really ugly. So this implementation is lame; it just
strdups the text and operates on the NUL-terminated version.
This should be fine as the strings we are wrapping are
generally pretty short. If it becomes a problem, we can
optimize later.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It is possible to break your repository config by creating an invalid key. The
config parser in turn chokes on it:
$ git init
Initialized empty Git repository in /tmp/gittest/.git/
$ git config .foo false
$ git config core.bare
fatal: bad config file line 6 in .git/config
This patch makes git-config reject keys which start or end with a dot and adds
tests for these cases.
Signed-off-by: Libor Pechacek <lpechacek@suse.cz>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Sanity-check config variable names when adding and retrieving them. As a side
effect code duplication between git_config_set_multivar and get_value (in
builtin/config.c) was removed and the common functionality was placed in
git_config_parse_key.
This breaks a test in t1300 which used invalid section-less keys in the tests
for "git -c". However, allowing such names there was useless, since there was
no way to set them via config file, and no part of git actually tried to use
section-less keys. This patch updates the test to use more realistic examples
as well as adding its own test.
Signed-off-by: Libor Pechacek <lpechacek@suse.cz>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We only need the size, which is much cheaper to get,
especially if it is a big binary file.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The logic in builtin_diffstat assumes that a
complete_rewrite pair should have its lines counted. This is
nonsensical for binary files and leads to confusing things
like:
$ git diff --stat --summary HEAD^ HEAD
foo.rand | Bin 4096 -> 4096 bytes
1 files changed, 0 insertions(+), 0 deletions(-)
$ git diff --stat --summary -B HEAD^ HEAD
foo.rand | 34 +++++++++++++++-------------------
1 files changed, 15 insertions(+), 19 deletions(-)
rewrite foo.rand (100%)
So let's reorder the function to handle binary files first
(which from diffstat's perspective look like complete
rewrites anyway), then rewrites, then actual diffstats.
There are two bonus prizes to this reorder:
1. It gets rid of a now-superfluous goto.
2. The binary case is at the top, which means we can
further optimize it in the next patch.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously the user was advised to use commit -c CHERRY_PICK_HEAD after
a conflicting cherry-pick. While this would preserve the original
commit's authorship, it would sadly discard cherry-pick's carefully
crafted MERGE_MSG (which contains the list of conflicts as well as the
original commit-id in the case of cherry-pick -x).
On the other hand, if a bare 'commit' were performed, it would preserve
the MERGE_MSG while resetting the authorship.
In other words, there was no way to simultaneously take the authorship
from CHERRY_PICK_HEAD and the commit message from MERGE_MSG.
This change fixes that situation. A bare 'commit' will now take the
authorship from CHERRY_PICK_HEAD and the commit message from MERGE_MSG.
If the user wishes to reset authorship, that must now be done explicitly
via --reset-author.
A side-benefit of passing commit authorship along this way is that we
can eliminate redundant authorship parsing code from revert.c.
(Also removed an unused include from revert.c)
Signed-off-by: Jay Soffian <jaysoffian@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make the git prompt (when enabled) show a CHERRY-PICKING indicator
when we are in the middle of a conflicted cherry-pick, analogous
to the existing MERGING and BISECTING flags.
Signed-off-by: Jay Soffian <jaysoffian@gmail.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a cherry-pick conflicts git advises:
$ git commit -c <original commit id>
to preserve the original commit message and authorship. Instead, let's
record the original commit id in CHERRY_PICK_HEAD and advise:
$ git commit -c CHERRY_PICK_HEAD
A later patch teaches git to handle the '-c CHERRY_PICK_HEAD' part.
Note that we record CHERRY_PICK_HEAD even in the case where there
are no conflicts so that we may use it to communicate authorship to
commit; this will then allow us to remove set_author_ident_env from
revert.c. However, we do not record CHERRY_PICK_HEAD when --no-commit
is used, as presumably the user intends to further edit the commit
and possibly even cherry-pick additional commits on top.
Tests and documentation contributed by Jonathan Nieder.
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Jay Soffian <jaysoffian@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
All the tests in t3507 (cherry-pick with conflicts) begin with the
same checkout + read-tree + clean incantation to ensure a predictable
starting point. Factor out a function for that so the interesting
part of the tests is easier to read.
The "update-index --refresh" and "diff-index --exit-code HEAD" are not
necessary as the point of this testsuite is not about testing
"read-tree --reset".
Improved-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Git traditionally overwrites untracked symlinks silently. This will
generally not cause massive data loss, but it is inconsistent with
the behavior for regular files, which are not silently overwritten.
With this change, git refuses to overwrite untracked symlinks by
default. If the user really wants to overwrite the untracked
symlink, he has git-clean and git-checkout -f at his disposal.
Signed-off-by: Clemens Buchacher <drizzd@aon.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The detached HEAD state is a source of much confusion for users
new to git. Here we try to document it better.
Reworked from http://article.gmane.org/gmane.comp.version-control.git/138440
Requested-by: Nicolas Pitre <nico@fluxnic.net>
Signed-off-by: Jay Soffian <jaysoffian@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint:
t/t7500-commit.sh: use test_cmp instead of test
t/gitweb-lib.sh: Ensure that errors are shown for --debug --immediate
gitweb/gitweb.perl: don't call S_ISREG() with undef
gitweb/gitweb.perl: remove use of qw(...) as parentheses
Test 5 wants to test --cherry-pick but limits by pathspec in such a way
that there are no commits on the left side of the range.
Add a test without "--cherry-pick" which displays this, and add two
more commits and another test which tests what we're after. This also
shortens the last test.
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The existing "--cherry-pick" does not work with unsymmetric ranges
(A..B) for obvious reasons.
Introduce "--left-only" and "--right-only" which limit the output to
commits on the respective sides of a symmetric range (i.e. only "<"
resp. ">" commits as per "--left-right").
This is especially useful for things like
git log --cherry-pick --right-only @{u}...
which is much more flexible (and descriptive) than
git cherry @{u} | sed -ne 's/^+ //p'
and potentially more useful than
git log --cherry-pick @{u}...
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add new config options:
git-p4.detectCopies - Enable copy detection.
git-p4.detectCopiesHarder - Find copies harder.
The detectCopies option should be set to a true/false value.
The detectCopiesHarder option should be set to true/false value.
P4Submit can now process diff-tree C status and integrate files accordingly.
Signed-off-by: Vitor Antunes <vitor.hda@gmail.com>
Acked-by: Pete Wyckoff <pw@padd.com>
Acked-by: Tor Arvid Lund <torarvid@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Only open files for edit after integrating if the SHA1 of source and destination
differ from each other.
Add git config option detectRenames to allow permanent rename detection. This
options should be set to a true/false value.
Rename "detectRename" variable to "detectRenames" to make it more coherent with
the description in git man pages, which always use plural.
Signed-off-by: Vitor Antunes <vitor.hda@gmail.com>
Acked-by: Pete Wyckoff <pw@padd.com>
Acked-by: Tor Arvid Lund <torarvid@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change commit_msg_is() in t/t7500-commit.sh to use test_cmp instead of
the shell's test function. Now if a test fails we'll get test_cmp
output showing us what failed.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Because '--immediate' stops test suite after first error, therefore in
this mode
test_debug 'cat gitweb.log'
was never ran, thus in effect negating effect of '--debug' option.
This made finidng the cause of errors in gitweb test sute difficult.
Modify the gitweb_run test subroutine to run test_debug itself in the
case of errors (and also remove "test_debug 'cat gitweb.log'" from
gitweb tests).
This makes it possible to run *gitweb tests* with --immediate ---debug
combination of options; also it makes gitweb tests to not output
spurious debug data that is not considered error.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Jakub Narębski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change S_ISREG($to_mode_oct) to S_ISREG($from_mode_oct) in the branch
that handles from modes, not to modes. This logic appears to have been
caused by copy/paste programming by Jakub Narebski in e8e41a93. It
would be better to rewrite this code not to be duplicated, but I
haven't done so.
This issue caused a failing test on perl 5.13.9, which has a warning
that turned this up:
gitweb.perl: Use of uninitialized value in subroutine entry at /home/avar/g/git/t/../gitweb/gitweb.perl line 4415.
Which caused the Git test suite to fail on this test:
./t9500-gitweb-standalone-no-errors.sh (Wstat: 256 Tests: 90 Failed: 84)
Failed tests: 1-8, 10-36, 38-45, 47-48, 50-88
Non-zero exit status: 1
Reported-by: perl 5.13.9
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Acked-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Using the qw(...) construct as implicit parentheses was deprecated in
perl 5.13.5. Change the relevant code in gitweb to not use the
deprecated construct. The offending code was introduced in 3562198b by
Jakub Narebski.
The issue is that perl will now warn about this:
$ perl -wE 'for my $i qw(a b) { say $i }'
Use of qw(...) as parentheses is deprecated at -e line 1.
a
b
This caused gitweb.perl to warn on perl 5.13.5 and above, and these
tests to fail on those perl versions:
./t9501-gitweb-standalone-http-status.sh (Wstat: 256 Tests: 11 Failed: 10)
Failed tests: 2-11
Non-zero exit status: 1
./t9502-gitweb-standalone-parse-output.sh (Wstat: 256 Tests: 10 Failed: 9)
Failed tests: 2-10
Non-zero exit status: 1
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Acked-by: Jakub Narębski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
