When recursing to unpack a delta base we must unuse_pack() so that
the pack window for the current object does not remain pinned in
memory while the delta base is itself being unpacked and materialized
for our use.
On a long delta chain of 50 objects we may need to access 6 different
windows from a very large (>3G) pack file in order to obtain all
of the delta base content. If the process ulimit permits us to
map/allocate only 1.5G we must release windows during this recursion
to ensure we stay within the ulimit and transition memory from pack
cache to standard malloc, or other mmap needs.
Inserting an unuse_pack() call prior to the recursion allows us to
avoid pinning the current window, making it available for garbage
collection if memory runs low.
This has been broken since at least before 1.5.1-rc1, and very
likely earlier than that. Its fixed now. :)
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-submodule add would trip if path to the submodule included a space,
or if its .git was a gitdir: link to a GIT_DIR kept elsewhere. Fix both.
Signed-off-by: Mark Levedahl <mlevedahl@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Many git commands use '--' to separate subcommands, options, and refs
from paths. However, the programmable completion for several of these
commands does not respect the '--', and offer subcommands, options, or
refs after a '--', although only paths are permitted. e.g. 'git bisect
-- <TAB>' offers subcommands, 'git log -- --<TAB>' offers options and
'git log -- git<TAB>' offers all gitgui tags.
The completion for the following commands share this wrong behaviour:
am add bisect commit diff log reset shortlog submodule gitk.
To avoid this, we check the presence of a '--' on the command line first
and let the shell do filename completion, if one is found.
Signed-off-by: SZEDER Gábor <szeder@ira.uka.de>
Acked-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Once a clone is successful we no longer need to hold onto the
.keep file created by the transport. Delete the file so we
can later repack the complete repository.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If git attempts to delete a ref, but the unlink of the ref
file fails, we print a message to stderr. This is usually a
good thing, but if the error is ENOENT, then it indicates
that the ref has _already_ been deleted. And since that's
our goal, it doesn't make sense to complain to the user.
This harmonizes the error reporting behavior for the
unpacked and packed cases; the packed case already printed
nothing on ENOENT, but the unpacked printed unconditionally.
Additionally, send-pack would, when deleting the tracking
ref corresponding to a remote delete, print "Failed to
delete" on any failure. This can be a misleading
message, since we actually _did_ delete at the remote side,
but we failed to delete locally. Rather than make the
message more precise, let's just eliminate it entirely; the
delete_ref routine already takes care of printing out a much
more specific message about what went wrong.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* 'qq/maint' (early part):
git-svn.perl: workaround assertions in svn library 1.5.0
mailinfo: feed the correct line length to decode_transfer_encoding()
git-clone: remove leftover debugging fprintf().
Fix "config_error_nonbool" used with value instead of key
clone -q: honor "quiet" option over native transports.
attribute documentation: keep EXAMPLE at end
builtin-commit.c: Use 'git_config_string' to get 'commit.template'
http.c: Use 'git_config_string' to clean up SSL config.
diff.c: Use 'git_config_string' to get 'diff.external'
convert.c: Use 'git_config_string' to get 'smudge' and 'clean'
builtin-log.c: Use 'git_config_string' to get 'format.subjectprefix' and 'format.suffix'
Documentation cvs: Clarify when a bare repository is needed
Documentation: be precise about which date --pretty uses
When GIT_TRACE is set, the user is most likely wanting to see an external
command that is about to be executed.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
With subversion 1.5.0 (C and perl libraries) the git-svn selftest
t9101-git-svn-props.sh fails at test 25 and 26. The following commands
cause assertions in the svn library
$ cd deeply
$ git-svn propget svn:ignore .
perl: /build/buildd/subversion-1.5.0dfsg1/subversion/libsvn_ra/ra_loader.c:674: svn_ra_get_dir: Assertion `*path != '/'' failed.
Aborted
$ git-svn propget svn:ignore ..
perl: /build/buildd/subversion-1.5.0dfsg1/subversion/libsvn_subr/path.c:120: svn_path_join: Assertion `is_canonical(component, clen)' failed.
With this commit, git-svn makes sure the path doesn't start with a
slash, and is not a dot, working around these assertions.
The breakage was reported by Lucas Nussbaum through
http://bugs.debian.org/489108
Signed-off-by: Gerrit Pape <pape@smarden.org>
Acked-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When handling a MIME multipart message, multi-part boundary lines are eaten
by a call to handle_boundary() function from the main loop of handle_body(),
and after that happens, we should update the line length correctly, because
handle_boundary() udpates line[] with new data.
This was caused by a thinko in 9aa2309 (mailinfo: apply the same fix not
to lose NULs in BASE64 and QP codepaths, 2008-05-25).
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The function "config_error_nonbool", that is defined in "config.c",
is used to report an error when a config key in the config file
should have a corresponding value but it hasn't.
So the parameter to this function should be the key and not the
value, because the value is undefined. And it could crash if the
value is used.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The earlier built-in conversion seems to have broken "git-clone"; this
teaches the command to honor the "-q" option again when talking to the
remote end over native transports (file://, git:// and ssh://).
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The document gives overall definition of states in DESCRIPTION, describes
various aspects of git operations that can be influenced in EFFECTS, and
finally gives examples in the EXAMPLE section. Archive creation however
was somehow documented after the EXAMPLE section, not insode EFFECTS.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
New users sometimes import a project and then immediately
try to use the imported repository as a central shared repository.
This provides pointers about setting up a bare repository for that
in the parts of the documentation dealing with CVS migration.
Signed-off-by: Matthew Ogilvie <mmogilvi_git@miniinfo.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This makes it explicit that the --pretty formats 'medium' and 'email' use the
author date (and ignore the committer date).
Signed-off-by: Junio C Hamano <gitster@pobox.com>
After master.k.org upgrade, I started seeing these warning messages:
transport.c: In function 'get_refs_via_curl':
transport.c:458: error: call to '_curl_easy_setopt_err_write_callback' declared with attribute warning: curl_easy_setopt expects a curl_write_callback argument for this option
It appears that the curl header wants to enforce the function signature
for callback function given to curl_easy_setopt() to be compatible with
that of (*curl_write_callback) or fwrite. This patch seems to work the
issue around.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If we match a lightweight (non-annotated tag) as the name to
output and --long was requested we do not have a tag, nor do
we have a tagged object to display. Instead we must use the
object we were passed as input for the long format display.
Reported-by: Mark Burton <markb@ordern.com>
Backtraced-by: Mikael Magnusson <mikachu@gmail.com>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The cmd_show loop resolves tags by showing them, then pointing the
object to the 'tagged' member. However, this object is not fully
initialized; it only contains the SHA1. (This resulted in a segfault
if there were two levels of tags.) We apply parse_object to get a
full object.
Noticed by Kalle Olavi Niemitalo on IRC.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There are basically two categories of update failures for
local refs:
1. problems outside of git, like disk full, bad
permissions, etc.
2. D/F conflicts on tracking branch ref names
In either case, there should already have been an error
message. In case '1', hopefully enough information has
already been given that the user can fix it. In the case of
'2', we can hint that the user can clean up their tracking
branch area by using 'git remote prune'.
Note that we don't actually know _which_ case we have, so
the user will receive the hint in case 1, as well. In this
case the suggestion won't do any good, but hopefully the
user is smart enough to figure out that it's just a hint.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The rev-parse manpage introduces the branch@{date} syntax,
and mentions the reflog specifically. However, new users may
not be familiar with the distinction between the reflog and
the commit date, so let's help them out with a "you may be
interested in --until" pointer.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
After initializing the config in the newly-created repository, we
need to unset GIT_CONFIG so that the global configs are read again.
Noticed by Pieter de Bie.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The original sanitization code was just taken from the
remotes2config.sh shell script in contrib.
Credit to Avery Pennarun for noticing this mistake, and Junio
for clarifying the rules for config section names:
Junio C Hamano wrote in <7vfxr23s6m.fsf@gitster.siamese.dyndns.org>:
> In
>
> [foo "bar"] baz = value
>
> foo and baz must be config.c::iskeychar() (and baz must be isalpha()), but
> "bar" can be almost anything.
Signed-off-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit ffe256f9ba ("git-svn: Speed up fetch")
introduced changes that create a temporary file for each object fetched by
svn. These files should be deleted automatically, but perl apparently
doesn't do this until the process exits (or perhaps when its garbage
collector runs).
This means that on a large fetch, especially with lots of branches, we
sometimes fill up /tmp completely, which prevents the next temp file from
being written completely. This is aggravated by the fact that a new temp
file is created for each updated file, even if that update produces a file
identical to one already in git. Thus, it can happen even if there's lots
of disk space to store the finished repository.
We weren't adequately checking for write errors, so this would result in an
invalid file getting committed, which caused git-svn to fail later with an
invalid checksum.
This patch adds a check to syswrite() so similar problems don't lead to
corruption in the future. It also unlink()'s each temp file explicitly
when we're done with it, so the disk doesn't need to fill up.
Signed-off-by: Avery Pennarun <apenwarr@gmail.com>
Tested-by: Björn Steinbrink <B.Steinbrink@gmx.de>
Acked-by: Eric Wong <normalperson@yhbt.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When run in batch mode, git cat-file never frees the memory for the blob
contents it is printing. This quickly adds up and causes git-svn to be
hardly usable for imports of large svn repos, because it uses cat-file in
batch mode and cat-file's memory usage easily reaches several hundred MB
without any good reason.
Signed-off-by: Björn Steinbrink <B.Steinbrink@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-config expects a space, not '=' between option and value.
Also, quote the value since it contains globs, which some shells will not
pass through unchanged, or will abort if the glob doesn't expand.
Signed-off-by: Joey Hess <joey@kitenet.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make sure that buf has enough space to store the trailing \0 of
the command line argument, too.
Signed-off-by: Jochen Voss <voss@seehuhn.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously, if there was an error while storing a local
tracking ref, the low-level functions would report an error,
but fetch's status output wouldn't indicate any problem.
E.g., imagine you have an old "refs/remotes/origin/foo/bar" but
upstream has deleted "foo/bar" in favor of a new branch
"foo". You would get output like this:
error: there are still refs under 'refs/remotes/origin/foo'
From $url_of_repo
* [new branch] foo -> origin/foo
With this patch, the output takes into account the status of
updating the local ref:
error: there are still refs under 'refs/remotes/origin/foo'
From $url_of_repo
! [new branch] foo -> origin/foo (unable to update local ref)
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we call "git clone" with a url that has a rewrite rule in either
$HOME/.gitconfig or /etc/gitconfig, the URL can be different from
what the command line expects it to be.
So, let's use the URL as the remote structure has it, not the literal
string from the command line.
Noticed by Pieter de Bie.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Acked-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is a backport of 0a47dc110e
to 'maint' to be included in 1.5.6.2 so that older server side
can accept dashless form of request when clients are updated.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git diff --check" should return non-zero when there was any whitespace
error but the code only paid attention to the error status of the last
new line in the patch.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Resetting a selected set of index entries is done with
"git reset -- paths" syntax, but we did not allow -- to be omitted
even when the command is unambiguous.
This updates the command to follow the general rule:
* When -- appears, revs come before it, and paths come after it;
* When there is no --, earlier ones are revs and the rest are paths, and
we need to guess. When lack of -- marker forces us to guess, we
protect from user errors and typoes by making sure what we treat as
revs do not appear as filenames in the work tree, and what we treat as
paths do appear as filenames in the work tree, and by erroring out if
that is not the case. We tell the user to disambiguate by using -- in
such a case.
which is employed elsewhere in the system.
When this rule is applied to "reset", because we can have only zero or one
rev to the command, the check can be slightly simpler than other programs.
We have to check only the first one or two tokens after the command name
and options, and when they are:
-- A:
no explicit rev given; "A" and whatever follows it are paths.
A --:
explicit rev "A" given and whatever follows the "--" are paths.
A B:
"A" could be rev or path and we need to guess. "B" could
be missing but if exists that (and everything that follows) would
be paths.
So we apply the guess only in the last case and only to "A" (not "B" and
what comes after it).
* As long as "A" is unambiguously a path, index entries for "A", "B" (and
everything that follows) are reset to the HEAD revision.
* If "A" is unambiguously a rev, on the other hand, the index entries for
"B" (and everything that follows) are reset to the "A" revision.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
