With this, each thread get repeatedly assigned the next available chunk of
objects to process until the whole list is done. The idea is to have
reasonably small chunks so that all CPUs remain busy with a minimum
number of threads for as long as there is data to process.
Signed-off-by: Nicolas Pitre <nico@cam.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
this is still rough, hence it is disabled by default. You need to compile
with "make THREADED_DELTA_SEARCH=1 ..." at the moment.
Threading is done on different portions of the object list to be
deltified. This is currently done by spliting the list into n parts and
then a thread is spawned for each of them. A better method would consist
of spliting the list into more smaller parts and have the n threads
pick the next part available.
Signed-off-by: Nicolas Pitre <nico@cam.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is based on Martin Koegler's idea to keep the object that
was successfully used as the base of the delta when it is about
to fall off the edge of the window. Instead of doing so only
for the objects at the edge of the window, this makes the window
a lru eviction mechanism. If an entry is used as a base, it is
moved to the last of the queue to be evicted.
This is a quick-and-dirty implementation, as it keeps the original
implementation of the data structure used for the window. This
originally was done as an array, not as an array of pointers,
because it was meant to be used as a cyclic FIFO buffer and a
plain array avoids an extra pointer indirection, while its FIFOness
eant that we are not "moving" the entries like this patch does.
The runtime from three versions were comparable. It seems to
make the resulting chain even shorter, which can only be good.
git-diff: resurrect the traditional empty "diff --git" behaviour
The warning message to suggest "Consider running git-status" from
"git-diff" that we experimented with during the 1.5.3 cycle turns
out to be a bad idea. It robbed cache-dirty information from people
who valued it, while still asking users to run "update-index --refresh".
It was hoped that the new behaviour would at least have some educational
value, but not showing the cache-dirty paths like before meant that the
user would not even know easily which paths were cache-dirty, and it
made the need to refresh the index look like even more unnecessary chore.
This commit reinstates the traditional behaviour, but with a twist.
By default, the empty "diff --git" output is totally squelched out
from "git diff" output. At the end of the command, it automatically
runs "update-index --refresh" as needed, without even bothering the
user. In other words, people who do not care about the cache-dirtyness
do not even have to see the warning.
The traditional behaviour to see the stat-dirty output and to bypassing
the overhead of content comparison can be specified by setting the
configuration variable diff.autorefreshindex to false.
git-tag: Fix -l option to use better shell style globs.
This patch removes certain behaviour of "git tag -l foo", currently
listing every tag name having "foo" as a substring. The same
thing now could be achieved doing "git tag -l '*foo*'".
This feature was added recently when git-tag.sh got the -n option
for showing tag annotations, because that commit also replaced the
old "grep pattern" behaviour with a more preferable "shell pattern"
behaviour (although slightly modified as you can see).
Thus, the following builtin-tag.c implemented it in order to
ensure that tests were passing unchanged with both programs.
Since common "shell patterns" match names with a given substring
_only_ when * is inserted before and after (as in "*substring*"), and
the "plain" behaviour cannot be achieved easily with the current
implementation, this is mostly the right thing to do, in order to
make it more flexible and consistent.
Tests for "git tag" were also changed to reflect this.
Signed-off-by: Carlos Rica <jasampler@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-svn: fix dcommit clobbering upstream when committing multiple changes
Although dcommit could detect if the first commit in the series
would conflict with the HEAD revision in SVN, it could not
detect conflicts in further commits it made.
Now we rebase each uncommitted change after each revision is
committed to SVN to ensure that we are up-to-date. git-rebase
will bail out on conflict errors if our next change cannot be
applied and committed to SVN cleanly, preventing accidental
clobbering of changes on the SVN-side.
--no-rebase users will have trouble with this, and are thus
warned if they are committing more than one commit. Fixing this
for (hopefully uncommon) --no-rebase users would be more complex
and will probably happen at a later date.
Thanks to David Watson for finding this and the original test.
If the configuration of the user has "diff.color = true", the
output from "log" we invoke internally added color codes, which
broke the parser.
Signed-off-by: Junio C Hamano <gitster@pobox.com> Tested-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Acked-by: Eric Wong <normalperson@yhbt.net>
filter-branch: introduce convenience function "skip_commit"
With this function, a commit filter can leave out unwanted commits
(such as temporary commits). It does _not_ undo the changeset
corresponding to that commit, but it _skips_ the revision. IOW
no tree object is changed by this.
If you like to commit early and often, but want to filter out all
intermediate commits, marked by "@@@" in the commit message, you can
now do this with
filter-branch: fix remnants of old syntax in documentation
Some time ago, filter-branch's syntax changed so that more than one
ref can be rewritten at the same time. This involved the removal of
the ref name for the result; instead, the refs are rewritten in-place.
This updates the last leftovers in the documentation to reflect the
new behavior.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Lately I have been doing a lot of calls to `git tag -d` and also to
`git tag -v`. In both such cases being able to complete the names
of existing tags saves the fingers some typing effort. We now look
for the -d or -v option to git-tag in the bash completion support
and offer up existing tag names as possible choices for these.
When creating a new tag we now also offer bash completion support
for the second argument to git-tag (the object to be tagged) as this
can often be a specific existing branch name and is not necessarily
the current HEAD.
If the -f option is being used to recreate an existing tag we now
also offer completion support on the existing tag names for the
first argument of git-tag, helping to the user to reselect the
prior tag name that they are trying to replace.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Make "git-log --" without paths behave the same as "git-log" without --
"git log" family of commands, even when run from a subdirectory,
do not limit the revision range with the current directory as
the path limiter, but with double-dash without any paths after
it, i.e. "git log --" do so. It was a mistake to have a
difference between "git log --" and "git log" introduced in
commit ae563542bf10fa8c33abd2a354e4b28aca4264d7 (First cut at
libifying revlist generation).
When appending the "git-archimport-id:" line to the end of log entries,
git-archimport would use two blank lines as a separator when there was no
body in the arch log (only a Summary: line), and zero blank lines when there
was a body (making it hard to see the break between the actual log message
and the git-archimport-id: line).
This patch makes git-archimport generate one blank line as a separator in all
cases.
The code favoring shallower deltas when size is equal was triggered
only when previous delta was also cached. There should be no relation
between cached deltas and same sized deltas.
Signed-off-by: Nicolas Pitre <nico@cam.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
gitk: Fix bug causing undefined variable error when cherry-picking
When "Show nearby tags" is turned off and the user did a cherry-pick,
we were trying to access variables relating to the descendent/ancestor
tag & head computations in addnewchild though they hadn't been set.
This makes sure we don't do that. Reported by Johannes Sixt.
git-add: Make the "tried to add ignored file" error message less confusing
Currently the error message seems to imply (at least to me) that only
the listed files were withheld and the rest of the files was added to the
index, even though that's obviously not the case.
Signed-off-by: Petr Baudis <pasky@suse.cz> Signed-off-by: Junio C Hamano <gitster@pobox.com>
gitweb: Fix escaping HTML of project owner in 'projects_list' and
'summary' views
This for example allows to put email address in the project owner
field in the projects index file (when $projects_list points to
a file, and not to a directory), in the form of:
Noticed-by: Jon Smirl <jonsmirl@gmail.com> Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Petr Baudis <pasky@suse.cz> Signed-off-by: Junio C Hamano <gitster@pobox.com>
"format-patch --root rev" is the way to show everything.
We used to trigger the special case "things not in origin"
semantics only when one and only one positive ref is given, and
no number (e.g. "git format-patch -4 origin") was specified, and
used the general revision range semantics for everything else.
This narrows the special case a bit more, by making:
git format-patch --root this_version
to show everything that leads to the named commit.
More importantly, document the two different semantics better.
The generic revision range semantics came later and bolted on
without being clearly documented.
git-merge: do up-to-date check also for all strategies
This clarifies the logic to omit fast-forward check and omit
trivial merge before running the specified strategy.
The "index_merge" variable started out as a flag to say "do not
do anything clever", but when recursive was changed to skip the
trivial merge, the semantics were changed and the variable alone
does not make sense anymore.
This splits the variable into two, allow_fast_forward (which is
almost always true, and avoids making a merge commit when the
other commit is a descendant of our branch, but is set to false
for ours and subtree) and allow_trivial_merge (which is false
for ours, recursive and subtree).
Unlike the earlier implementation, the "ours" strategy allows an
up-to-date condition. When we are up-to-date, the result will
be our commit, and by definition, we will have our tree as the
result.
git --bare cmd: do not unconditionally nuke GIT_DIR
"GIT_DIR=some.where git --bare cmd" and worse yet
"git --git-dir=some.where --bare cmd" were very confusing. They
both ignored git-dir specified, and instead made $cwd as GIT_DIR.
This changes --bare not to override existing GIT_DIR.
This has been like this for a long time. Let's hope nobody sane
relied on this insane behaviour.
Here is my attempt to fix this with a minimally intrusive patch.
* As "git --bare init" cannot tell if it was called with --bare or
just "GIT_DIR=. git init", I added an explicit assignment of
is_bare_repository_cfg on the codepath for "git --bare".
* GIT_WORK_TREE alone without GIT_DIR does not make any sense,
nor GIT_WORK_TREE with an explicit "git --bare". Catch that
mistake. It might make sense to move this check to "git.c"
side as well, but I tried to shoot for the minimum change for
now.
* Some scripts, especially from the olden days, rely on
traditional GIT_DIR behaviour in "git init". Namely, these
are some notable patterns:
(create a bare repository)
- mkdir some.git && cd some.git && GIT_DIR=. git init
- mkdir some.git && cd some.git && git --bare init
This comes with a new test script and also passes the existing
test suite, but there may be cases that are still broken with
the current tip of master and this patch does not yet fix. I'd
appreciate help in straightening this mess out.
Uwe Kleine-König noticed that under certain circumstances, name-rev
picked a non-optimal tag. Jeff King analyzed that name-rev only
takes into account the number of merge traversals, and then the
_last_ number in the description.
As an easy way to fix it, use a weighting factor for merge traversals:
A merge traversal is now made 65535 times more expensive than a
first-parent traversal.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Acked-by: Uwe Kleine-König <ukleinek@informatik.uni-freiburg.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* 'master' of git://repo.or.cz/git-gui:
git-gui: Correct 'git gui blame' in a subdirectory
git-gui: Do not offer to stage three-way diff hunks into the index
git-gui: Refactor diff pane popup support for future improvements
git-gui: Fix "unoptimized loading" to not cause git-gui to crash
git-gui: Paper bag fix "Stage Hunk For Commit" in diff context menu
git-gui: Allow git-merge to use branch names in conflict markers
git-gui: Fix window manager problems on ion3
When nothing to git-commit, honor the git-status color setting.
Instead of disabling color all of the time during a git-commit, allow
the user's config preference in the situation where there is nothing
to commit. In this situation, the status is printed to the terminal
and not sent to COMMIT_EDITMSG, so honoring the status color setting
is expected.
Signed-off-by: Brian Hetro <whee@smaertness.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Currently, searchbox is CSS'd to have position: absolute, which has the
unfortunate consequence that if the viewport is too small and can't fit
into the page width together with the navbar, it gets overlapped and part
of the navbar gets obscured. This makes searchbox float: right instead,
thus the navbar simply gets wrapped.
Discovered and fix pointed out by Michael Olson <mwolson@gnu.org>.
Signed-off-by: Petr Baudis <pasky@suse.cz> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Merge branch 'master' of git://linux-nfs.org/~bfields/git
* 'master' of git://linux-nfs.org/~bfields/git:
Documentation/user-manual.txt: fix a few omissions of gitlink commands.
user-manual: fix incorrect header level
user-manual: use pithier example commit
user-manual: introduce the word "commit" earlier
user-manual: minor editing for conciseness
user-manual: edit "ignoring files" for conciseness
Documentation/user-manual.txt: fix a few omissions of gitlink commands.
user-manual: edit "ignoring files" for conciseness
The immediate motivation for writing this section was to explain the
various places ignore patterns could be used. However, I still think
.gitignore is the case most people will want to learn about first. It
also makes it a bit more concrete to introduce ignore patterns in the
context of .gitignore first. And the existance of gitignore(5) relieves
the pressure to explain it all here.
So, stick to the .gitignore example, with only a brief mention of the
others, explain the syntax only by example, and leave the rest to
gitignore(5).
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu> Cc: Johan Herland <johan@herland.net>
The usage string for the executable was missing --refresh. In
addition, the documentation referred to "file", but the usage string
referred to "filepattern". Updated the documentation to
"filepattern", as git-add does handle patterns.
Signed-off-by: Brian Hetro <whee@smaertness.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Don't segfault if we failed to inflate a packed delta
Under some types of packfile corruption the zlib stream holding the
data for a delta within a packfile may fail to inflate, due to say
a CRC failure within the compressed data itself. When this occurs
the unpack_compressed_entry function will return NULL as a signal to
the caller that the data is not available. Unfortunately we then
tried to use that NULL as though it referenced a memory location
where a delta was stored and tried to apply it to the delta base.
Loading a byte from the NULL address typically causes a SIGSEGV.
cate on #git noticed this failure in `git fsck --full` where the
call to verify_pack() first noticed that the packfile was corrupt
by finding that the packfile's SHA-1 did not match the raw data of
the file. After finding this fsck went ahead and tried to verify
every object within the packfile, even though the packfile was
already known to be bad. If we are going to shovel bad data at
the delta unpacking code, we better handle it correctly.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-gui: Correct 'git gui blame' in a subdirectory
David Kastrup pointed out that the following sequence was not
working as we had intended:
$ cd lib
$ git gui blame console.tcl
fatal: cannot stat path lib/console.tcl: No such file or directory
The problem here was we disabled the chdir to the root of the
working tree when we are running with a "bare allowed" feature
such as blame or browser, but we still kept the prefix we found via
`git rev-parse --show-prefix`. This caused us to try and look for
the file "console.tcl" within the subdirectory but also include
the subdirectory's own path from the root of the working tree.
This is unlikely to succeed, unless the user just happened to have
a "lib/lib/console.tcl" file in the repository, in which case we
would produce the wrong result.
In the case of a bare repository we shouldn't get back a value from
`rev-parse --show-prefix`, so really $_prefix should only be set
to the non-empty string if we are in a working tree and we are in a
subdirectory of that working tree. If this is true we really want
to always be at the top level of the working tree, as all paths are
accessed as though they were relative to the top of the working tree.
Converting $_prefix to a ../ sequence is a fairly simple approach
to moving up the requisite levels.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
git-p4: Make 'git-p4 branches' work after an initial clone with git clone from an origin-updated repository.
After a clone with "git clone" of a repository the p4 branches are only in remotes/origin/p4/* and not in remotes/p4/*.
Separate the code for detection and creation out of the P4Sync command class into standalone methods and use them
from the P4Branches command.
Signed-off-by: Simon Hausmann <simon@lst.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
After git-write-tree finishes computing the tree, it updates the
index so that later operations can take advantage of fully
populated cache tree.
However, anybody writing the index file has to mark the entries
that are racily clean. For each entry whose cached lstat(3)
data in the index exactly matches what is obtained from the
filesystem, if the timestamp on the index file was the same or
older than the modification timestamp of the file, the blob
contents and the work tree file, after convert_to_git(), need to
be compared, and if they are different, its index entry needs to
be marked not to match the lstat(3) data from the filesystem.
In order for this to work, convert_to_git() needs to work
correctly, which in turn means you need to read the config file
to get the settings of core.crlf and friends.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When squashing, rebase -i did not prevent fast forwards. This could
happen when picking some other commit than the first one, and then
squashing the first commit. So do not allow fast forwards when
squashing.
Noticed by Johannes Sixt.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Install man3 manpages to $PREFIX/share/man/man3 even for site installs
MakeMaker supports three installation modes: perl, site, and vendor. The first
and third install manpages to $PREFIX/share/man, only site installs to
$PREFIX/man. For consistency with the rest of git, which does not make the
distinction and writes all manpages to $PREFIX/share/man, this change makes
sure that perl does too, even when it's installed in site mode.
Signed-off-by: martin f. krafft <madduck@madduck.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-svn init/clone --stdlayout option to default-init trunk/tags/branches
The --stdlayout option to git-svn init/clone initialises the default
Subversion values of trunk,tags,branches: -T trunk -b branches -t tags.
If any of the -T/-t/-b options are given in addition, they are given
preference.
[ew: fixed whitespace and added "-s" shortcut]
Signed-off-by: martin f. krafft <madduck@madduck.net> Signed-off-by: Eric Wong <normalperson@yhbt.net>
A particularly bad case was HASH_LIMIT <= hash_count[i] < 2*HASH_LIMIT:
in that case, only a single hash survived. For larger cases,
2*HASH_LIMIT was the actual limiting value after pruning.
Signed-off-by: David Kastrup <dak@gnu.org> Acked-by: Nicolas Pitre <nico@cam.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-gui: Do not offer to stage three-way diff hunks into the index
git-apply does not accept a patch that was generated as a three-way
combined diff format such as we see during merge conflicts. If we
get such a diff in our diff viewer and try to send it to git-apply
it just errors out and the user is left confused wondering why they
cannot stage that hunk.
Instead of feeding a known to be unacceptable hunk to git-apply we
now just disable the stage/unstage context menu option if the hunk
came from a three way diff. The user may still be confused about
why they cannot work with a combined diff, but at least they are
only confused as to why git-gui is not offering them the action.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Merge branch 'master' of git://repo.or.cz/git/fastimport
* 'master' of git://repo.or.cz/git/fastimport:
Teach bash about git-submodule and its subcommands
Teach bash to complete ref arguments to git-describe
Update bash completion with new 1.5.3 command line options
git-gui: Refactor diff pane popup support for future improvements
The current popup_diff_menu procedure is somewhat messy as it has a
few duplications of the same logic in each of the different legs of
the routine. We can simplify these by setting a few state variables
in the different legs.
No functional change, just a cleanup to make it easier to implement
future functional changes within this block.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Teach bash about git-submodule and its subcommands
The git-submodule command is new in 1.5.3 and contains a number
of useful subcommands for working on submodules. We usually try
to offer the subcommands of a git command in the bash completion,
so here they are for git-submodule.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Teach bash to complete ref arguments to git-describe
I'm often finding that I need to run git-describe on very long
remote tracking branch names, to find out what tagged revision
the remote tracking branch is now at (or not at). Typing out
the ref names is painful, so bash completion on them is a very
useful feature.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Update bash completion with new 1.5.3 command line options
A number of commands have learned new tricks as part of git 1.5.3.
If these are long options (--foo) we tend to support them in the
bash completion, as it makes the user's task of using the option
slightly easier.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Reset terminal attributes when terminating git send-email
If you break out of the prompts presented to you by git send-email
your terminal can be left in an inconsistent state. Here we trap
the interrupt signal and reset the terminal before exiting.
Signed-off-by: Sean Estabrooks <seanlkml@sympatico.ca> Signed-off-by: Junio C Hamano <gitster@pobox.com>