Teach combine-diff to honour the path-output-order imposed by
diffcore-order, and optimize how matching paths are found in
the N-way diffs made with parents.
* ks/combine-diff:
tests: add checking that combine-diff emits only correct paths
combine-diff: simplify intersect_paths() further
combine-diff: combine_diff_path.len is not needed anymore
combine-diff: optimize combine_diff_path sets intersection
diff test: add tests for combine-diff with orderfile
diffcore-order: export generic ordering interface
"git diff --quiet -- pathspec1 pathspec2" sometimes did not return
correct status value.
* nd/diff-quiet-stat-dirty:
diff: do not quit early on stat-dirty files
diff.c: move diffcore_skip_stat_unmatch core logic out for reuse later
When QUICK is set (i.e. with --quiet) we try to do as little work as
possible, stopping after seeing the first change. stat-dirty is
considered a "change" but it may turn out not, if no actual content is
changed. The actual content test is performed too late in the process
and the shortcut may be taken prematurely, leading to incorrect return
code.
Assume we do "git diff --quiet". If we have a stat-dirty file "a" and
a really dirty file "b". We break the loop in run_diff_files() and
stop after "a" because we have got a "change". Later in
diffcore_skip_stat_unmatch() we find out "a" is actually not
changed. But there's nothing else in the diff queue, we incorrectly
declare "no change", ignoring the fact that "b" is changed.
This also happens to "git diff --quiet HEAD" when it hits
diff_can_quit_early() in oneway_diff().
This patch does the content test earlier in order to keep going if "a"
is unchanged. The test result is cached so that when
diffcore_skip_stat_unmatch() is done in the end, we spend no cycles on
re-testing "a".
Reported-by: IWAMOTO Toshihiro <iwamoto@valinux.co.jp>
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
diffcore_order() interface only accepts a queue of `struct
diff_filepair`.
In the next patches, we'll want to order `struct combine_diff_path`
by path, so let's first rework diffcore-order to also provide
generic low-level interface for ordering arbitrary objects, provided
they have path accessors.
The new interface is:
- `struct obj_order` for describing objects to ordering routine, and
- order_objects() for actually doing the ordering work.
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Bitfields need to specify their signedness explicitly or the compiler is
free to default as it sees fit. With compilers that default 'unsigned'
(SUNWspro 12 seems to do this) the tri-state nature of is_binary
vanishes and all files are treated as binary.
Signed-off-by: Richard Lowe <richlowe@richlowe.net>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The is_binary flag needs only three values: -1, 0, and 1.
However, we use a whole 32-bit int for it on most systems
(both 32- and 64- bit).
Instead, we can mark it to use only 2 bits. On 32-bit
systems, this lets it end up as part of the bitfield above
(saving 4 bytes). On 64-bit systems, we don't see any change
(because the savings end up as padding), but it does leave
room for another "free" 32-bit value to be added later.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The middle of the diff_filespec struct contains a mixture of
ints, shorts, and bit-fields, followed by a pointer. On an
x86-64 system with an LP64 or LLP64 data model (i.e., most
of them), the integers and flags end up being padded out by
41 bits to put the pointer at an 8-byte boundary.
After the pointer, we have the "int is_binary" field, which
is only 32 bits. We end up wasting another 32 bits to pad
the struct size up to a multiple of 64 bits.
We can move the is_binary field before the pointer, which
lets the compiler store it where we used to have padding.
This shrinks the top padding to only 9 bits (from the
bit-fields), and eliminates the bottom padding entirely,
dropping the struct size from 88 to 80 bytes.
On a 32-bit system, there is no benefit, but nor should
there be any harm (we only need 4-byte alignment there, so
we were already using only 9 bits of padding).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The only mention of this field in the code is by some
debugging code which prints it out (and it will always be
zero, since we never touch it otherwise). It was obsoleted
very early on by 25d5ea4 ([PATCH] Redo rename/copy detection
logic., 2005-05-24).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This struct field was obsoleted by be58e70 (diff: unify
external diff and funcname parsing code, 2008-10-05), but we
forgot to remove it.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
diff_filespec has a 2-bit "dirty_submodule" field and
defines two flags as macros. Originally these were right
next to each other, but a new field was accidentally added
in between in commit 4682d85. This patch puts the field and
its flags back together.
Using an enum like:
enum {
DIRTY_SUBMODULE_UNTRACKED = 1,
DIRTY_SUBMODULE_MODIFIED = 2
} dirty_submodule;
would be more obvious, but it bloats the structure. Limiting
the enum size like:
} dirty_submodule : 2;
might work, but it is not portable.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We do not want a link to 0{40} object stored anywhere in our objects.
* jk/maint-null-in-trees:
fsck: detect null sha1 in tree entries
do not write null sha1s to on-disk index
diff: do not use null sha1 as a sentinel value
The diff code represents paths using the diff_filespec
struct. This struct has a sha1 to represent the sha1 of the
content at that path, as well as a sha1_valid member which
indicates whether its sha1 field is actually useful. If
sha1_valid is not true, then the filespec represents a
working tree file (e.g., for the no-index case, or for when
the index is not up-to-date).
The diff_filespec is only used internally, though. At the
interfaces to the diff subsystem, callers feed the sha1
directly, and we create a diff_filespec from it. It's at
that point that we look at the sha1 and decide whether it is
valid or not; callers may pass the null sha1 as a sentinel
value to indicate that it is not.
We should not typically see the null sha1 coming from any
other source (e.g., in the index itself, or from a tree).
However, a corrupt tree might have a null sha1, which would
cause "diff --patch" to accidentally diff the working tree
version of a file instead of treating it as a blob.
This patch extends the edges of the diff interface to accept
a "sha1_valid" flag whenever we accept a sha1, and to use
that flag when creating a filespec. In some cases, this
means passing the flag through several layers, making the
code change larger than would be desirable.
One alternative would be to simply die() upon seeing
corrupted trees with null sha1s. However, this fix more
directly addresses the problem (while bogus sha1s in a tree
are probably a bad thing, it is really the sentinel
confusion sending us down the wrong code path that is what
makes it devastating). And it means that git is more capable
of examining and debugging these corrupted trees. For
example, you can still "diff --raw" such a tree to find out
when the bogus entry was introduced; you just cannot do a
"--patch" diff (just as you could not with any other
corrupted tree, as we do not have any content to diff).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Due to the way "git diff --no-index" is bolted onto by touching the
low level code that is shared with the rest of the "git diff" code,
even though it has to work in a very different way, any comparison
that involves a file "-" at the root level incorrectly tried to read
from the standard input. This cleans up the no-index codepath
further to remove code that reads from the standard input from the
core side, which is never necessary when git is running its usual
diff operation.
* jc/refactor-diff-stdin:
diff-index.c: "git diff" has no need to read blob from the standard input
diff-index.c: unify handling of command line paths
diff-index.c: do not pretend paths are pathspecs
Only "diff --no-index -" does. Bolting the logic into the low-level
function diff_populate_filespec() was a layering violation from day
one. Move populate_from_stdin() function out of the generic diff.c
to its only user, diff-index.c.
Also make sure "-" from the command line stays a special token "read
from the standard input", even if we later decide to sanitize the
result from prefix_filename() function in a few obvious ways,
e.g. removing unnecessary "./" prefix, duplicated slashes "//" in
the middle, etc.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This teaches combine-diff machinery to feed a combined merge to a callback
function when DIFF_FORMAT_CALLBACK is specified.
So far, format callback functions are not used for anything but 2-way
diffs. A callback is given a diff_queue_struct, which is an array of
diff_filepair. As its name suggests, a diff_filepair is a _pair_ of
diff_filespec that represents a single preimage and a single postimage.
Since "diff -c" is to compare N parents with a single merge result and
filter out any paths whose result match one (or more) of the parent(s),
its output has to be able to represent N preimages and 1 postimage. For
this reason, a callback function that inspects a diff_filepair that
results from this new infrastructure can and is expected to view the
preimage side (i.e. pair->one) as an array of diff_filespec. Each element
in the array, except for the last one, is marked with "has_more_entries"
bit, so that the same callback function can be used for 2-way diffs and
combined diffs.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Usually, diff frontends populate the output queue with filepairs without
any rename information and call diffcore_std() to sort the renames out.
When --follow is in effect, however, diff-tree family of frontend has a
hack that looks like this:
diff-tree frontend
-> diff_tree_sha1()
. populate diff_queued_diff
. if --follow is in effect and there is only one change that
creates the target path, then
-> try_to_follow_renames()
-> diff_tree_sha1() with no pathspec but with -C
-> diffcore_std() to find renames
. if rename is found, tweak diff_queued_diff and put a
single filepair that records the found rename there
-> diffcore_std()
. tweak elements on diff_queued_diff by
- rename detection
- path ordering
- pickaxe filtering
We need to skip parts of the second call to diffcore_std() that is related
to rename detection, and do so only when try_to_follow_renames() did find
a rename. Earlier 1da6175 (Make diffcore_std only can run once before a
diff_flush, 2010-05-06) tried to deal with this issue incorrectly; it
unconditionally disabled any second call to diffcore_std().
This hopefully fixes the breakage.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
These options take an optional argument, but this optional argument was
not documented.
Original patch by Matthieu Moy, but documentation for -B mostly copied
from the explanations of Junio C Hamano.
While we're there, fix a typo in a comment in diffcore.h.
Signed-off-by: Matthieu Moy <Matthieu.Moy@imag.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When file renames/copies detection is turned on, the
second diffcore_std will degrade a 'C' pair to a 'R' pair.
And this may happen when we run 'git log --follow' with
hard copies finding. That is, the try_to_follow_renames()
will run diffcore_std to find the copies, and then
'git log' will issue another diffcore_std, which will reduce
'src->rename_used' and recognize this copy as a rename.
This is not what we want.
So, I think we really don't need to run diffcore_std more
than one time.
Signed-off-by: Bo Yang <struggleyb.nku@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refactor the diff_queue_struct code, this macro help
to reset the structure.
Signed-off-by: Bo Yang <struggleyb.nku@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When encountering a dirty submodule while doing "git diff --submodule"
print an extra line for new untracked content and another for modified
but already tracked content. And if the HEAD of the submodule is equal
to the ref diffed against in the superproject, drop the output which
would just show the same SHA1s and no commit message headlines.
To achieve that, the dirty_submodule bitfield is expanded to two bits.
The output of "git status" inside the submodule is parsed to set the
according bits.
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the worst case is_submodule_modified() got called three times for
each submodule. The information we got from scanning the whole
submodule tree the first time can be reused instead.
New parameters have been added to diff_change() and diff_addremove(),
the information is stored in a new member of struct diff_filespec. Its
value is then reused instead of calling is_submodule_modified() again.
When no explicit "-dirty" is needed in the output the call to
is_submodule_modified() is not necessary when the submodules HEAD
already disagrees with the ref of the superproject, as this alone
marks it as modified. To achieve that, get_stat_data() got an extra
argument.
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Also fix error in diff_filepair::status documentation, and point to
the in-code reference as well as the doc.
Signed-off-by: Yann Dirson <ydirson@altern.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The "diff" gitattribute is somewhat overloaded right now. It
can say one of three things:
1. this file is definitely binary, or definitely not
(i.e., diff or !diff)
2. this file should use an external diff engine (i.e.,
diff=foo, diff.foo.command = custom-script)
3. this file should use particular funcname patterns
(i.e., diff=foo, diff.foo.(x?)funcname = some-regex)
Most of the time, there is no conflict between these uses,
since using one implies that the other is irrelevant (e.g.,
an external diff engine will decide for itself whether the
file is binary).
However, there is at least one conflicting situation: there
is no way to say "use the regular rules to determine whether
this file is binary, but if we do diff it textually, use
this funcname pattern." That is, currently setting diff=foo
indicates that the file is definitely text.
This patch introduces a "binary" config option for a diff
driver, so that one can explicitly set diff.foo.binary. We
default this value to "don't know". That is, setting a diff
attribute to "foo" and using "diff.foo.funcname" will have
no effect on the binaryness of a file. To get the current
behavior, one can set diff.foo.binary to true.
This patch also has one additional advantage: it cleans up
the interface to the userdiff code a bit. Before, calling
code had to know more about whether attributes were false,
true, or unset to determine binaryness. Now that binaryness
is a property of a driver, we can represent these situations
just by passing back a driver struct.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
This concept was retired by 77882f6 (Retire diffcore-pathspec.,
2006-04-10), more than 2 years ago.
Signed-off-by: Yann Dirson <ydirson@altern.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
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>
We find rename candidates by computing a fingerprint hash of
each file, and then comparing those fingerprints. There are
inherently O(n^2) comparisons, so it pays in CPU time to
hoist the (rather expensive) computation of the fingerprint
out of that loop (or to cache it once we have computed it once).
Previously, we didn't keep the filespec information around
because then we had the potential to consume a great deal of
memory. However, instead of keeping all of the filespec
data, we can instead just keep the fingerprint.
This patch implements and uses diff_free_filespec_data_large
to accomplish that goal. We also have to change
estimate_similarity not to needlessly repopulate the
filespec data when we already have the hash.
Practical tests showed 4.5x speedup for a 10% memory usage
increase.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This updates the hunk header customization syntax. The special
case 'funcname' attribute is gone.
You assign the name of the type of contents to path's "diff"
attribute as a string value in .gitattributes like this:
*.java diff=java
*.perl diff=perl
*.doc diff=doc
If you supply "diff.<name>.funcname" variable via the
configuration mechanism (e.g. in $HOME/.gitconfig), the value is
used as the regexp set to find the line to use for the hunk
header (the variable is called "funcname" because such a line
typically is the one that has the name of the function in
programming language source text).
If there is no such configuration, built-in default is used, if
any. Currently there are two default patterns: default and java.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This makes"diff -p" hunk headers customizable via gitattributes mechanism.
It is based on Johannes's earlier patch that allowed to define a single
regexp to be used for everything.
The mechanism to arrive at the regexp that is used to define hunk header
is the same as other use of gitattributes. You assign an attribute, funcname
(because "diff -p" typically uses the name of the function the patch is about
as the hunk header), a simple string value. This can be one of the names of
built-in pattern (currently, "java" is defined) or a custom pattern name, to
be looked up from the configuration file.
(in .gitattributes)
*.java funcname=java
*.perl funcname=perl
(in .git/config)
[funcname]
java = ... # ugly and complicated regexp to override the built-in one.
perl = ... # another ugly and complicated regexp to define a new one.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This replaces an explicit initialization of filespec->is_binary
field used for rename/break followed by direct access to that
field with a wrapper function that lazily iniaitlizes and
accesses the field. We would add more attribute accesses for
the use of diff routines, and it would be better to make this
abstraction earlier.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
diffcore-break and diffcore-rename would want to behave slightly
differently depending on the binary-ness of the data, so add one
bit to the filespec, as the structure is now passed down to
diffcore_count_changes() function.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We may want to use richer information on the data we are dealing
with in this function, so instead of passing a buffer address
and length, just pass the diffcore_filespec structure. Existing
callers always call this function with parameters taken from a
filespec anyway, so there is no functionality changes.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This updates the way diffcore represents an unmerged pair
somewhat. It used to be that entries with mode=0 on both sides
were used to represent an unmerged pair, but now it has an
explicit flag. This is to allow diff-index --cached to report
the entry from the tree when the path is unmerged in the index.
This is used in updating "git reset <tree> -- <path>" to restore
absense of the path in the index from the tree.
Signed-off-by: Junio C Hamano <junkio@cox.net>
The final output from diff used to compare pathnames between
preimage and postimage to tell if the filepair is a rename/copy.
By explicitly marking the filepair created by diffcore_rename(),
the output routine, resolve_rename_copy(), does not have to do
so anymore. This helps feeding a filepair that has different
pathnames in one and two elements to the diff machinery (most
notably, comparing two blobs).
Signed-off-by: Junio C Hamano <junkio@cox.net>
This changes diffcore-rename to reuse statistics information
gathered during similarity estimation, and updates the hashtable
implementation used to keep track of the statistics to be
denser. This seems to give better performance.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This is a companion patch to the previous fix to diffcore-rename.
The merging-back process should use a logic similar to what is used
there.
Signed-off-by: Junio C Hamano <junkio@cox.net>
This is to rework diffcore break/rename/copy detection code
so that it does not affected when deltifier code gets improved.
Signed-off-by: Junio C Hamano <junkio@cox.net>
While reviewing the end user tutorial rewrite by J. Bruce
Fields, I noticed that "git-diff-tree -B -C" did not correctly
break the total rewrite of Documentation/tutorial.txt. It turns
out that we had integer overflow during the break score
computations.
Cop out by using floating point. This is not a kernel.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When many paths are modified, rename detection takes a lot of time.
The new option -l<num> can be used to disable rename detection when
more than <num> paths are possibly created as renames.
Signed-off-by: Junio C Hamano <junkio@cox.net>
When I run git-diff-tree on big change, it seems the command eats so
much memory. so I just put git under valgrind to see what's going on.
diff_free_filespec_data() doesn't free diff_filespec itself.
[jc: I ended up doing things slightly differently from Yasushi's
patch. The original idea was to use free_filespec_data() only to
free the data portion and keep useing the filespec itself, but
no existing code seems to do things that way, so I just yanked
that part out.]
Signed-off-by: Yasushi SHOJI <yashi@atmark-techno.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>
When an unmerged path was fed via diff_unmerged() into diffcore,
it eventually called run_diff() with "one" and "two" parameters
with NULL, but run_diff() was not written carefully enough to
notice this situation.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This fixes three bugs in the -B heuristics.
- Although it was advertised that the initial break criteria
used was the same as what diffcore-rename uses, it was using
something different. Instead of using smaller of src and dst
size to compare with "edit" size, (insertion and deletion),
it was using larger of src and dst, unlike the rename/copy
detection logic. This caused the parameter to -B to mean
something different from the one to -M and -C. To compensate
for this change, the default break score is also changed to
match that of the default for rename/copy.
- The code would have crashed with division by zero when trying
to break an originally empty file.
- Contrary to what the comment said, the algorithm was breaking
small files, only to later merge them together.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Junio C Hamano
Nguyễn Thái Ngọc Duy
Kirill Smelkov
Richard Lowe
Jeff King
Jonathan Nieder
Matthieu Moy
Bo Yang
Jens Lehmann
Yann Dirson
Linus Torvalds
Junio C Hamano
Yasushi SHOJI
Linus Torvalds