* maint:
Update GIT 1.5.3.5 Release Notes
git-rebase--interactive.sh: Make 3-way merge strategies work for -p.
git-rebase--interactive.sh: Don't pass a strategy to git-cherry-pick.
Fix --strategy parsing in git-rebase--interactive.sh
Make merge-recursive honor diff.renamelimit
cherry-pick/revert: more compact user direction message
core-tutorial: Use new syntax for git-merge.
git-merge: document but discourage the historical syntax
Prevent send-pack from segfaulting (backport from 'master')
Documentation/git-cvsexportcommit.txt: s/mgs/msg/ in example
* cc/skip:
Bisect: add "skip" to the short usage string.
Bisect run: "skip" current commit if script exit code is 125.
Bisect: add a "bisect replay" test case.
Bisect: add "bisect skip" to the documentation.
Bisect: refactor "bisect_{bad,good,skip}" into "bisect_state".
Bisect: refactor some logging into "bisect_write".
Bisect: refactor "bisect_write_*" functions.
Bisect: implement "bisect skip" to mark untestable revisions.
Bisect: fix some white spaces and empty lines breakages.
rev-list documentation: add "--bisect-all".
rev-list: implement --bisect-all
* lt/rename:
Do the fuzzy rename detection limits with the exact renames removed
Fix ugly magic special case in exact rename detection
Do exact rename detection regardless of rename limits
Do linear-time/space rename logic for exact renames
copy vs rename detection: avoid unnecessary O(n*m) loops
Ref-count the filespecs used by diffcore
Split out "exact content match" phase of rename detection
Add 'diffcore.h' to LIB_H
* ds/gitweb:
gitweb: Use chop_and_escape_str in more places.
gitweb: Refactor abbreviation-with-title-attribute code.
gitweb: Provide title attributes for abbreviated author names.
git-rebase--interactive.sh: Make 3-way merge strategies work for -p.
git-rebase--interactive.sh used to pass all parents of a merge commit to
git-merge, which means that we have at least 3 heads to merge: HEAD,
first parent and second parent. So 3-way merge strategies like recursive
wouldn't work.
Fortunately, we have checked out the first parent right before the merge
anyway, so that is HEAD. Therefore we can drop simply it from the list
of parents, making 3-way strategies work for merge commits with only
two parents.
Signed-off-by: Björn Steinbrink <B.Steinbrink@gmx.de> Acked-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-rebase--interactive.sh: Don't pass a strategy to git-cherry-pick.
git-cherry-pick doesn't support a strategy paramter, so don't pass one.
This means that --strategy for interactive rebases is a no-op for
anything but merge commits, but that's still better than being broken. A
correct fix would probably need to port the --merge behaviour from plain
git-rebase.sh, but I have no clue how to integrate that cleanly.
Signed-off-by: Björn Steinbrink <B.Steinbrink@gmx.de> Acked-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix --strategy parsing in git-rebase--interactive.sh
For the --strategy/-s option, git-rebase--interactive.sh dropped the
parameter which it was trying to parse.
Signed-off-by: Björn Steinbrink <B.Steinbrink@gmx.de> Acked-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
It might be a sign of source code management gone bad, but when two branches
has diverged almost beyond recognition and time has come for the branches to
merge, the user is going to need all the help his tool can give him. Honoring
diff.renamelimit has great potential as a painkiller in such situations.
The painkiller effect could have been achieved by e.g. 'merge.renamelimit',
but the flexibility gained by a separate option is questionable: our user
would probably expect git to detect renames equally good when merging as
when diffing (I known I did).
Signed-off-by: Lars Hjemli <hjemli@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-merge: document but discourage the historical syntax
Historically "git merge" took its command line arguments in a
rather strange order. Document the historical syntax, and also
document clearly that it is not encouraged in new scripts.
There is no reason to deprecate the historical syntax, as the
current code can sanely tell which syntax the caller is using,
and existing scripts by people do use the historical syntax.
If we can't find a source match, and we have no destination, we
need to abort the match function early before we try to match
the destination against the remote.
Signed-off-by: Emil Medve <Emilian.Medve@Freescale.com> Acked-by: Pierre Habouzit <madcoder@debian.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since it is now OK to pass a null pointer to display_progress() and
stop_progress() resulting in a no-op, then we can simplify the code
and remove a bunch of lines by not making those calls conditional all
the time.
Signed-off-by: Nicolas Pitre <nico@cam.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
This allows for better management of progress "object" existence,
as well as making the progress display implementation more independent
from its callers.
Signed-off-by: Nicolas Pitre <nico@cam.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
There are some cases when one line from "raw" git-diff output (raw
format) corresponds to more than one patch in the patchset git-diff
output; we call this situation "split patch". Old code misdetected
subsequent patches (for different files) with the same pre-image and
post-image as fragments of "split patch", leading to mislabeled
from-file/to-file diff header etc.
Old code used pre-image and post-image SHA-1 identifier ('from_id' and
'to_id') to check if current patch corresponds to old raw diff format
line, to find if one difftree raw line coresponds to more than one
patch in the patch format. Now we use post-image filename for that.
This assumes that post-image filename alone can be used to identify
difftree raw line. In the case this changes (which is unlikely
considering current diff engine) we can add 'from_id' and 'to_id'
to detect "patch splitting" together with 'to_file'.
Because old code got pre-image and post-image SHA-1 identifier for the
patch from the "index" line in extended diff header, diff header had
to be buffered. New code takes post-image filename from "git diff"
header, which is first line of a patch; this allows to simplify
git_patchset_body code. A side effect of resigning diff header
buffering is that there is always "diff extended_header" div, even
if extended diff header is empty.
Alternate solution would be to check when git splits patches, and do
not check if parsed info from current patch corresponds to current or
next raw diff format output line. Git splits patches only for 'T'
(typechange) status filepair, and there always two patches
corresponding to one raw diff line. It was not used because it would
tie gitweb code to minute details of git diff output.
While at it, use newly introduced parsed_difftree_line wrapper
subroutine in git_difftree_body.
Noticed-by: Yann Dirson <ydirson@altern.org> Diagnosed-by: Petr Baudis <pasky@suse.cz> Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
parse-options: Allow abbreviated options when unambiguous
When there is an option "--amend", the option parser now recognizes
"--am" for that option, provided that there is no other option beginning
with "--am".
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
parse-options: make some arguments optional, add callbacks.
* add the possibility to use callbacks to parse some options, this can
help implementing new options kinds with great flexibility. struct option
gains a callback pointer and a `defval' where callbacks user can put
either integers or pointers. callbacks also can use the `value' pointer
for anything, preferably to the pointer to the final storage for the value
though.
* add a `flag' member to struct option to make explicit that this option may
have an optional argument. The semantics depends on the option type. For
INTEGERS, it means that if the switch is not used in its
--long-form=<value> form, and that there is no token after it or that the
token does not starts with a digit, then it's assumed that the switch has
no argument. For STRING or CALLBACK it works the same, except that the
condition is that the next atom starts with a dash. This is needed to
implement backward compatible behaviour with existing ways to parse the
command line. Its use for new options is discouraged.
Signed-off-by: Pierre Habouzit <madcoder@debian.org> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
The option parser takes argc, argv, an array of struct option
and a usage string. Each of the struct option elements in the array
describes a valid option, its type and a pointer to the location where the
value is written. The entry point is parse_options(), which scans through
the given argv, and matches each option there against the list of valid
options. During the scan, argv is rewritten to only contain the
non-option command line arguments and the number of these is returned.
Aggregation of single switches is allowed:
-rC0 is the same as -r -C 0 (supposing that -C wants an arg).
Every long option automatically support the option with the same name,
prefixed with 'no-' to unset the switch. It assumes that initial value for
strings are "NULL" and for integers is "0".
Long options are supported either with '=' or without:
--some-option=foo is the same as --some-option foo
Acked-by: Kristian Høgsberg <krh@redhat.com> Signed-off-by: Pierre Habouzit <madcoder@debian.org> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
include $PATH in generating list of commands for "help -a"
Git had previously been using the $PATH for scripts--a previous
patch moved exec'ed commands to also use the $PATH. For consistency
"help -a" should also list commands in the $PATH.
The main commands are still listed from the git_exec_path(), but
the $PATH is walked and other git commands (probably extensions) are
listed.
Signed-off-by: Scott R Parish <srp@srparish.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
We need to correctly set up $PATH for non-c based git commands.
Since we already do this, we can just use that $PATH and execvp,
instead of looping over the paths with execve.
This patch adds a setup_path() function to exec_cmd.c, which sets
the $PATH order correctly for our search order. execv_git_cmd() is
stripped down to setting up argv and calling execvp(). git.c's
main() only only needs to call setup_path().
Signed-off-by: Scott R Parish <srp@srparish.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
remove unused/unneeded "pattern" argument of list_commands
list_commands() currently accepts and ignores a "pattern" argument,
and then hard codes a prefix as well as some magic numbers. This
hardcodes the prefix inside of the function and removes the magic
numbers.
Signed-off-by: Scott R Parish <srp@srparish.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Correct handling of upload-pack in builtin-fetch-pack
The field in the args was being ignored in favor of a static constant
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org> Thanked-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Try to avoid a lot of work scanning for excluded files,
by caching some more information when setting up the exclusion
data structure.
Speeds up 'git runstatus' on a repository containing the Qt sources by 30% and
reduces the amount of instructions executed (as measured by valgrind) by a
factor of 2.
* maint:
RelNotes-1.5.3.5: describe recent fixes
merge-recursive.c: mrtree in merge() is not used before set
sha1_file.c: avoid gcc signed overflow warnings
Fix a small memory leak in builtin-add
honor the http.sslVerify option in shell scripts
merge-recursive.c: mrtree in merge() is not used before set
The called function merge_trees() sets its *result, to which the
address of the variable mrtree in merge() function is passed,
only when index_only is set. But that is Ok as the function
uses the value in the variable only under index_only iteration.
However, recent gcc does not realize this. Work it around by
adding a fake initializer.
sha1_file.c: In check_packed_git_:
sha1_file.c:527: warning: assuming signed overflow does not
occur when assuming that (X + c) < X is always false
sha1_file.c:527: warning: assuming signed overflow does not
occur when assuming that (X + c) < X is always false
for a piece of code that tries to make sure that off_t is large
enough to hold more than 2^32 offset. The test tried to make
sure these do not wrap-around:
/* make sure we can deal with large pack offsets */
off_t x = 0x7fffffffUL, y = 0xffffffffUL;
if (x > (x + 1) || y > (y + 1)) {
but gcc assumes it can do whatever optimization it wants for a
signed overflow (undefined behaviour) and warns about this
construct.
Follow Linus's suggestion to check sizeof(off_t) instead to work
around the problem.
No longer talk about Cogito since it's deprecated. Some scripts (such as
git-reset or git-branch) have undergone builtinification so adjust the text
to reflect this.
Fix a typo in the description of git-show-branch (merges are indicated by a
`-', not by a `.').
git-pull/git-push do not seem to use the dumb git-ssh-fetch/git-ssh-upload
(the text was probably missing a word).
Adjust a link that wasn't rendered properly because it was wrapped.
Signed-off-by: Benoit Sigoure <tsuna@lrde.epita.fr> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-fetch: do not fail when remote branch disappears
When the branch named with branch.$name.merge is not covered by
the fetch configuration for the remote repository named with
branch.$name.remote, we automatically add that branch to the set
of branches to be fetched. However, if the remote repository
does not have that branch (e.g. it used to exist, but got
removed), this is not a reason to fail the git-fetch itself.
The situation however will be noticed if git-fetch was called by
git-pull, as the resulting FETCH_HEAD would not have any entry
that is marked for merging.
Acked-By: Daniel Barkalow <barkalow@iabervon.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* 'master' of git://git.kernel.org/pub/scm/gitk/gitk: (34 commits)
gitk: Use the UI font for the diff/old version/new version radio buttons
gitk: Simplify the code for finding commits
gitk: Fix a couple more bugs in the path limiting
gitk: Fix some bugs with path limiting in the diff display
gitk: Use the status window for other functions
gitk: Integrate the reset progress bar in the main frame
gitk: Ensure tabstop setting gets restored by Cancel button
gitk: Limit diff display to listed paths by default
gitk: Fix Tcl error: can't unset findcurline
gitk: Get rid of the diffopts variable
gitk: Fix bug where the last few commits would sometimes not be visible
gitk: Add a font chooser
gitk: Keep track of font attributes ourselves instead of using font actual
gitk: Use named fonts instead of the font specification
gitk: Fix bug causing Tcl error when changing find match type
gitk: Fix the tab setting in the diff display window
gitk: Add progress bars for reading in stuff and for finding
gitk: Fix a couple of bugs
gitk: Simplify highlighting interface and combine with Find function
gitk: Fix bug in generating patches
...
gitk: Use the UI font for the diff/old version/new version radio buttons
This makes the radio buttons for selecting whether to see the full diff,
the old version or the new version use the same font as the other user
interface elements.
This unifies findmore and findmorerev, and adds the ability to do
a search with or without wrap around from the end of the list of
commits to the beginning (or vice versa for reverse searches).
findnext and findprev are gone, and the buttons and keys for searching
all call dofind now. dofind doesn't unmark the matches to start with.
Shift-up and shift-down are back by popular request, and the searches
they do don't wrap around. The other keys that do searches (/, ?,
return, M-f) do wrapping searches except for M-g.
Usually you cannot revert a merge because you do not know which
side of the merge should be considered the mainline (iow, what
change to reverse).
With this patch, cherry-pick and revert learn -m (--mainline)
option that lets you specify the parent number (starting from 1)
of the mainline, so that you can:
git revert -m 1 $merge
to reverse the changes introduced by the $merge commit relative
to its first parent, and:
git cherry-pick -m 2 $merge
to replay the changes introduced by the $merge commit relative
to its second parent.
Bisect run: "skip" current commit if script exit code is 125.
This is incompatible with previous versions because an exit code
of 125 used to mark current commit as "bad". But hopefully this exit
code is not much used by test scripts or other programs. (126 and 127
are used by POSIX compliant shells to mean "found but not
executable" and "command not found", respectively.)
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Also fix "bisect bad" and "bisect good" short usage description.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Bisect: refactor "bisect_{bad,good,skip}" into "bisect_state".
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Bisect: refactor some logging into "bisect_write".
Also use "die" instead of "echo >&2 something ; exit 1".
And simplify "bisect_replay".
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Bisect: implement "bisect skip" to mark untestable revisions.
When there are some "skip"ped revisions, we add the '--bisect-all'
option to "git rev-list --bisect-vars". Then we filter out the
"skip"ped revisions from the result of the rev-list command, and we
modify the "bisect_rev" var accordingly.
We don't always use "--bisect-all" because it is slower
than "--bisect-vars" or "--bisect".
When we cannot find for sure the first bad commit because of
"skip"ped commits, we print the hash of each possible first bad
commit and then we exit with code 2.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Do the fuzzy rename detection limits with the exact renames removed
When we do the fuzzy rename detection, we don't care about the
destinations that we already handled with the exact rename detector.
And, in fact, the code already knew that - but the rename limiter, which
used to run *before* exact renames were detected, did not.
This fixes it so that the rename detection limiter now bases its
decisions on the *remaining* rename counts, rather than the original
ones.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix ugly magic special case in exact rename detection
For historical reasons, the exact rename detection had populated the
filespecs for the entries it compared, and the rest of the similarity
analysis depended on that. I hadn't even bothered to debug why that was
the case when I re-did the rename detection, I just made the new one
have the same broken behaviour, with a note about this special case.
This fixes that fixme. The reason the exact rename detector needed to
fill in the file sizes of the files it checked was that the _inexact_
rename detector was broken, and started comparing file sizes before it
filled them in.
Fixing that allows the exact phase to do the sane thing of never even
caring (since all *it* cares about is really just the SHA1 itself, not
the size nor the contents).
It turns out that this also indirectly fixes a bug: trying to populate
all the filespecs will run out of virtual memory if there is tons and
tons of possible rename options. The fuzzy similarity analysis does the
right thing in this regard, and free's the blob info after it has
generated the hash tables, so the special case code caused more trouble
than just some extra illogical code.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Do exact rename detection regardless of rename limits
Now that the exact rename detection is linear-time (with a very small
constant factor to boot), there is no longer any reason to limit it by
the number of files involved.
In some trivial testing, I created a repository with a directory that
had a hundred thousand files in it (all with different contents), and
then moved that directory to show the effects of renaming 100,000 files.
With the new code, that resulted in
[torvalds@woody big-rename]$ time ~/git/git show -C | wc -l
400006
real 0m2.071s
user 0m1.520s
sys 0m0.576s
ie the code can correctly detect the hundred thousand renames in about 2
seconds (the number "400006" comes from four lines for each rename:
diff --git a/really-big-dir/file-1-1-1-1-1 b/moved-big-dir/file-1-1-1-1-1
similarity index 100%
rename from really-big-dir/file-1-1-1-1-1
rename to moved-big-dir/file-1-1-1-1-1
and the extra six lines is from a one-liner commit message and all the
commit information and spacing).
Most of those two seconds weren't even really the rename detection, it's
really all the other stuff needed to get there.
With the old code, this wouldn't have been practically possible. Doing
a pairwise check of the ten billion possible pairs would have been
prohibitively expensive. In fact, even with the rename limiter in
place, the old code would waste a lot of time just on the diff_filespec
checks, and despite not even trying to find renames, it used to look
like:
[torvalds@woody big-rename]$ time git show -C | wc -l 1400006
real 0m12.337s
user 0m12.285s
sys 0m0.192s
ie we used to take 12 seconds for this load and not even do any rename
detection! (The number 1400006 comes from fourteen lines per file moved:
seven lines each for the delete and the create of a one-liner file, and
the same extra six lines of commit information).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Do linear-time/space rename logic for exact renames
This implements a smarter rename detector for exact renames, which
rather than doing a pairwise comparison (time O(m*n)) will just hash the
files into a hash-table (size O(n+m)), and only do pairwise comparisons
to renames that have the same hash (time O(n+m) except for unrealistic
hash collissions, which we just cull aggressively).
Admittedly the exact rename case is not nearly as interesting as the
generic case, but it's an important case none-the-less. A similar general
approach should work for the generic case too, but even then you do need
to handle the exact renames/copies separately (to avoid the inevitable
added cost factor that comes from the _size_ of the file), so this is
worth doing.
In the expectation that we will indeed do the same hashing trick for the
general rename case, this code uses a generic hash-table implementation
that can be used for other things too. In fact, we might be able to
consolidate some of our existing hash tables with the new generic code
in hash.[ch].
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
copy vs rename detection: avoid unnecessary O(n*m) loops
The core rename detection had some rather stupid code to check if a
pathname was used by a later modification or rename, which basically
walked the whole pathname space for all renames for each rename, in
order to tell whether it was a pure rename (no remaining users) or
should be considered a copy (other users of the source file remaining).
That's really silly, since we can just keep a count of users around, and
replace all those complex and expensive loops with just testing that
simple counter (but this all depends on the previous commit that shared
the diff_filespec data structure by using a separate reference count).
Note that the reference count is not the same as the rename count: they
behave otherwise rather similarly, but the reference count is tied to
the allocation (and decremented at de-allocation, so that when it turns
zero we can get rid of the memory), while the rename count is tied to
the renames and is decremented when we find a rename (so that when it
turns zero we know that it was a rename, not a copy).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Rather than copy the filespecs when introducing new versions of them
(for rename or copy detection), use a refcount and increment the count
when reusing the diff_filespec.
This avoids unnecessary allocations, but the real reason behind this is
a future enhancement: we will want to track shared data across the
copy/rename detection. In order to efficiently notice when a filespec
is used by a rename, the rename machinery wants to keep track of a
rename usage count which is shared across all different users of the
filespec.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Split out "exact content match" phase of rename detection
This makes the exact content match a separate function of its own.
Partly to cut down a bit on the size of the diffcore_rename() function
(which is too complex as it is), and partly because there are smarter
ways to do this than an O(m*n) loop over it all, and that function
should be rewritten to take that into account.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>