If autofollowing of tags is enabled, we see a new tag on the remote
that we don't have, and we already have the SHA-1 object that the
tag is peeled to, then we can fetch the tag while we are fetching
the other objects on the first connection.
This is a slight optimization for projects that have a habit of
tagging a release commit after most users have already seen and
downloaded that commit object through a prior fetch session. In
such cases the users may still find new objects in branch heads,
but the new tag will now also be part of the first pack transfer
and the subsequent connection to autofollow tags is not required.
Currently git.git does not benefit from this optimization as any
release usually gets a new commit at the same time that it gets a
new release tag, however git-gui.git and many other projects are
in the habit of tagging fairly old commits.
Users who did not already have the tagged commit still require
opening a second connection to autofollow the tag, as we are unable
to determine on the client side if $tag^{} will be sent to the
client during the first transfer or not. Such computation must be
performed on the remote side of the connection and is deferred to
another series of changes.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
To facilitate testing and verification of the requests sent by
git-fetch to the remote side we permit logging the received packet
lines to the file descriptor specified in GIT_DEBUG_SEND_PACK has
been set. Special start and end lines are included to indicate
the start and end of each connection.
$ GIT_DEBUG_SEND_PACK=3 git fetch 3>UPLOAD_LOG
$ cat UPLOAD_LOG
#S
want 8e10cf4e007ad7e003463c30c34b1050b039db78 multi_ack side-band-64k thin-pack ofs-delta
want ddfa4a33562179aca1ace2bcc662244a17d0b503
#E
#S
want 3253df4d1cf6fb138b52b1938473bcfec1483223 multi_ack side-band-64k thin-pack ofs-delta
#E
>From the above trace the first connection opened by git-fetch was to
download two refs (with values 8e and dd) and the second connection
was opened to automatically follow an annotated tag (32).
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
To support calling find_non_local_tags() more than once in a single
git-fetch process we need the existing_refs to be stack-allocated
so it resets on the second call. We also should free the path
lists to avoid unnecessary memory leaking.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By allowing the function to append onto the end of an existing list
we can do more interesting things, like join the list of tags we
want to fetch into the first fetch, rather than the second.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If we ever decided to append onto the end of this list the tail
pointer must be looking at the right memory cell at the end of
the HEAD ref_map.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We can free this ref_map as soon as the fetch is complete. It is not
used for the automatic tag following, nor is it used to disconnect the
transport. This avoids some confusion about why we are holding onto
these refs while following tags.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Apparently fetch_map is passed through, but is not actually used.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint:
Update draft release notes for 1.5.4.4
revert: actually check for a dirty index
tests: introduce test_must_fail
git-submodule: Fix typo 'url' which should be '$url'
receive-pack: Initialize PATH to include exec-dir.
Conflicts:
builtin-revert.c
The previous code mistakenly used wt_status_prepare to check whether the
index had anything commitable in it; however, that function is just an
init function, and will never report a dirty index.
The correct way with wt_status_* would be to call wt_status_print with the
output pointing to /dev/null or similar. However, that does extra work by
both examining the working tree and spewing status information to nowhere.
Instead, let's just implement the useful subset of wt_status_print as an
"is_index_dirty" function.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we expect a git command to notice and signal errors, we
carelessly wrote in our tests:
test_expect_success 'reject bogus request' '
do something &&
do something else &&
! git command
'
but a non-zero exit could come from the "git command" segfaulting.
A new helper function "tset_must_fail" is introduced and it is
meant to be used to make sure the command gracefully fails (iow,
dying and exiting with non zero status is counted as a failure
to "gracefully fail"). The above example should be written as:
test_expect_success 'reject bogus request' '
do something &&
do something else &&
test_must_fail git command
'
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix typo in 'test -z "url"' when checking whether a submodule url is
empty. "url" should be "$url".
Signed-off-by: Ping Yin <pkufranky@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
511707d (use only the $PATH for exec'ing git commands) made it a
requirement to call setup_path() to include the git exec-dir in PATH
before spawning any other git commands. git-receive-pack was not yet
adapted to do this and therefore fails to spawn git-unpack-objects if that
is not in the standard PATH.
Signed-off-by: Björn Steinbrink <B.Steinbrink@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git fast-import expects to be run from an existing (possibly
empty) repository. It was dying with a suboptimal message if that
wasn't the case.
Signed-off-by: Jean-Luc Herren <jlh@gmx.ch>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
* np/verify-pack:
add storage size output to 'git verify-pack -v'
fix unimplemented packed_object_info_detail() features
make verify_one_pack() a bit less wrong wrt packed_git structure
factorize revindex code out of builtin-pack-objects.c
Conflicts:
Makefile
Earlier, overly-long onelines would not be wrapped at all, and indented
with 6 spaces.
Instead, we now wrap around at 72 characters, with a first-line indent
of 2 spaces, and the rest with 4 spaces.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Earlier, when you called "git format-patch --cover-letter -M", the
diffstat in the cover letter would not inherit the "-M". Now it does.
While at it, add a few "|| break" statements in the test's loops;
otherwise, breakages inside the loops would not be caught.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Due to greediness of a pattern, gitweb used to mark (show) last match
in line, if there are more than one match in line. Now it shows first.
Showing all matches in a line would require further work.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* pb/cvsimport:
cvsimport: document that -M can be used multiple times
cvsimport: allow for multiple -M options
cvsimport: have default merge regex allow for dashes in the branch name
* mk/maint-parse-careful:
receive-pack: use strict mode for unpacking objects
index-pack: introduce checking mode
unpack-objects: prevent writing of inconsistent objects
unpack-object: cache for non written objects
add common fsck error printing function
builtin-fsck: move common object checking code to fsck.c
builtin-fsck: reports missing parent commits
Remove unused object-ref code
builtin-fsck: move away from object-refs to fsck_walk
add generic, type aware object chain walker
Conflicts:
Makefile
builtin-fsck.c
The --max-age=<timestamp> and --min-age=<timestamp> are now shown only
in the git-rev-list manpage (plumbing).
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When passing "xyz" to make_absolute_path(), make_absolute_path()
erroneously tried to chdir("xyz"), and then append "/xyz". Instead,
skip the chdir() completely when no slash was found.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Ping Yin noticed that "git diff-index --raw" shows 0{40} when work tree
has submodule difference, but "git diff --raw" didn't correctly do so.
There was a mistake in the diffcore_skip_stat_unmatch() that was meant to
clean up the stat-only difference for running diff between the index and
work tree and diff between the tree and the work tree, to cause it re-read
from the submodule repository HEAD. When ce_stat_match() says work tree
is different, we should always say 0{40} on the work tree side.
This patch fixes the issue, and adds tests.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The internal implementation of diff-index codepath used to use non const
pointer to pass sha1 around, but it did not have to. With this, we can
also lose the private no_sha1[] array, as we can use the public null_sha1[]
array that exists exactly for the same purpose.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Now find_unique_abbrev() never returns NULL, there is no need for callers
to prepare for seeing NULL and fall back to giving the full 40-hexdigits.
While we are at it, drop "..." in the "git reset" output that reports the
location of the new HEAD, between the abbreviated commit object name and
the one line commit summary. Because we are always showing the HEAD
(which cannot be missing!), we never had a case where we show the full 40
hexdigits that is not followed by three dots, and these three dots were
stealing 3 columns from the precious horizontal screen real estate out of
80 that can better be used for the one line commit summary.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The function returned NULL when no object that matches the name
was found, but that made the callers more complicated, as nobody
used that NULL return as an indication that no object with such
a name exists. They (at least the careful ones) instead took
the full 40-hexdigit and used in such a case, and the careless
ones segfaulted.
With this "git rev-parse --short 5555555555555555555555555555555555555555"
would stop segfaulting.
This is based on Jeff King's rewrite to my RFC patch, but "missing"
logic swapped to "exists". The final logic reads:
For existing objects, make sure the abbreviated string uniquely
identifies it. Otherwise, make sure the abbreviated string is
long enough so that it would not name any existing object.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously, --abort would end by git resetting to ORIG_HEAD, but some
commands, such as git reset --hard (which happened in git rebase --skip,
but could just as well be typed by the user), would have already modified
ORIG_HEAD.
Just use the orig-head we store in $dotest instead.
Signed-off-by: Mike Hommey <mh@glandium.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Our scripts try to stick to fairly limited subset of POSIX BRE for
portability. It is unclear from manual page from GNU grep which is GNU
extension and which is portable, so let's spell it out to help new people
to keep their contributions from hurting porters.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This can possibly break external scripts that depend on the previous
output, but those script can't possibly be critical to Git usage, and
fixing them should be trivial.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since commit eb32d236df, there was a TODO
comment in packed_object_info_detail() about the SHA1 of base object to
OBJ_OFS_DELTA objects. So here it is at last.
While at it, providing the actual storage size information as well is now
trivial.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Simply freeing it is wrong. There are many things attached to this
structure that are not cleaned up. In practice this doesn't matter much
since this happens just before the program exits, but it is still
a bit more "correct" to leak it implicitly rather than explicitly.
And therefore it is also a good idea to register it with
install_packed_git(). Not only might it have better chance of being
properly cleaned up if such functionality is implemented for the general
case, but some functions like init_revindex() expect all packed_git
instances to be globally accessible.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
No functional change. This is needed to fix verify-pack in a later patch.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We just move to the top of the tree and proceed. This
shouldn't break any existing callers, since the behavior was
previously disallowed.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We expect git rebase --abort to come back to the original (pre-rebase)
head, independently from when it's run during a rebase.
Signed-off-by: Mike Hommey <mh@glandium.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
