* jc/zlib-wrap:
zlib: allow feeding more than 4GB in one go
zlib: zlib can only process 4GB at a time
zlib: wrap deflateBound() too
zlib: wrap deflate side of the API
zlib: wrap inflateInit2 used to accept only for gzip format
zlib: wrap remaining calls to direct inflate/inflateEnd
zlib wrapper: refactor error message formatter
* jk/combine-diff-binary-etc:
combine-diff: respect textconv attributes
refactor get_textconv to not require diff_filespec
combine-diff: handle binary files as binary
combine-diff: calculate mode_differs earlier
combine-diff: split header printing into its own function
The size of objects we read from the repository and data we try to put
into the repository are represented in "unsigned long", so that on larger
architectures we can handle objects that weigh more than 4GB.
But the interface defined in zlib.h to communicate with inflate/deflate
limits avail_in (how many bytes of input are we calling zlib with) and
avail_out (how many bytes of output from zlib are we ready to accept)
fields effectively to 4GB by defining their type to be uInt.
In many places in our code, we allocate a large buffer (e.g. mmap'ing a
large loose object file) and tell zlib its size by assigning the size to
avail_in field of the stream, but that will truncate the high octets of
the real size. The worst part of this story is that we often pass around
z_stream (the state object used by zlib) to keep track of the number of
used bytes in input/output buffer by inspecting these two fields, which
practically limits our callchain to the same 4GB limit.
Wrap z_stream in another structure git_zstream that can express avail_in
and avail_out in unsigned long. For now, just die() when the caller gives
a size that cannot be given to a single zlib call. In later patches in the
series, we would make git_inflate() and git_deflate() internally loop to
give callers an illusion that our "improved" version of zlib interface can
operate on a buffer larger than 4GB in one go.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Wrap deflateInit, deflate, and deflateEnd for everybody, and the sole use
of deflateInit2 in remote-curl.c to tell the library to use gzip header
and trailer in git_deflate_init_gzip().
There is only one caller that cares about the status from deflateEnd().
Introduce git_deflate_end_gently() to let that sole caller retrieve the
status and act on it (i.e. die) for now, but we would probably want to
make inflate_end/deflate_end die when they ran out of memory and get
rid of the _gently() kind.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jc/rename-degrade-cc-to-c:
diffcore-rename: fall back to -C when -C -C busts the rename limit
diffcore-rename: record filepair for rename src
diffcore-rename: refactor "too many candidates" logic
builtin/diff.c: remove duplicated call to diff_result_code()
Refactor the "do not stop feeding the backend early" logic into a small
helper function and use it in both run_diff_files() and diff_tree() that
has the stop-early optimization. We may later add other types of diffcore
transformation that require to look at the whole result like diff-filter
does, and having the logic in a single place is essential for longer term
maintainability.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This function actually does two things:
1. Load the userdiff driver for the filespec.
2. Decide whether the driver has a textconv component, and
initialize the textconv cache if applicable.
Only part (1) requires the filespec object, and some callers
may not have a filespec at all. So let's split them it into
two functions, and put part (2) with the userdiff code,
which is a better fit.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
With diff.suppress-blank-empty=true, "git diff --word-diff" would
output data that had been read from uninitialized heap memory.
The problem was that fn_out_consume did not account for the
possibility of a line with length 1, i.e., the empty context line
that diff.suppress-blank-empty=true converts from " \n" to "\n".
Since it assumed there would always be a prefix character (the space),
it decremented "len" unconditionally, thus passing len=0 to emit_line,
which would then blindly call emit_line_0 with len=-1 which would
pass that value on to fwrite as SIZE_MAX. Boom.
Signed-off-by: Jim Meyering <meyering@redhat.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jh/dirstat-lines:
Mark dirstat error messages for translation
Improve error handling when parsing dirstat parameters
New --dirstat=lines mode, doing dirstat analysis based on diffstat
Allow specifying --dirstat cut-off percentage as a floating point number
Add config variable for specifying default --dirstat behavior
Refactor --dirstat parsing; deprecate --cumulative and --dirstat-by-file
Make --dirstat=0 output directories that contribute < 0.1% of changes
Add several testcases for --dirstat and friends
* jc/fix-diff-files-unmerged:
diff-files: show unmerged entries correctly
diff: remove often unused parameters from diff_unmerge()
diff.c: return filepair from diff_unmerge()
test: use $_z40 from test-lib
* jc/fix-diff-files-unmerged:
diff-files: show unmerged entries correctly
diff: remove often unused parameters from diff_unmerge()
diff.c: return filepair from diff_unmerge()
test: use $_z40 from test-lib
* jh/dirstat:
--dirstat: In case of renames, use target filename instead of source filename
Teach --dirstat not to completely ignore rearranged lines within a file
--dirstat-by-file: Make it faster and more correct
--dirstat: Describe non-obvious differences relative to --stat or regular diff
When encountering errors or unknown tokens while parsing parameters to the
--dirstat option, it makes sense to die() with an error message informing
the user of which parameter did not make sense. However, when parsing the
diff.dirstat config variable, we cannot simply die(), but should instead
(after warning the user) ignore the erroneous or unrecognized parameter.
After all, future Git versions might add more dirstat parameters, and
using two different Git versions on the same repo should not cripple the
older Git version just because of a parameter that is only understood by
a more recent Git version.
This patch fixes the issue by refactoring the dirstat parameter parsing
so that parse_dirstat_params() keeps on parsing parameters, even if an
earlier parameter was not recognized. When parsing has finished, it returns
zero if all parameters were successfully parsed, and non-zero if one or
more parameters were not recognized (with appropriate error messages
appended to the 'errmsg' argument).
The parse_dirstat_params() callers then decide (based on the return value
from parse_dirstat_params()) whether to warn and ignore (in case of
diff.dirstat), or to warn and die (in case of --dirstat).
The patch also adds a couple of tests verifying the correct behavior of
--dirstat and diff.dirstat in the face of unknown (possibly future) dirstat
parameters.
Suggested-by: Junio C Hamano <gitster@pobox.com>
Improved-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch adds an alternative implementation of show_dirstat(), called
show_dirstat_by_line(), which uses the more expensive diffstat analysis
(as opposed to show_dirstat()'s own (relatively inexpensive) analysis)
to derive the numbers from which the --dirstat output is computed.
The alternative implementation is controlled by the new "lines" parameter
to the --dirstat option (or the diff.dirstat config variable).
For binary files, the diffstat analysis counts bytes instead of lines,
so to prevent binary files from dominating the dirstat results, the
byte counts for binary files are divided by 64 before being compared to
their textual/line-based counterparts. This is a stupid and ugly - but
very cheap - heuristic.
In linux-2.6.git, running the three different --dirstat modes:
time git diff v2.6.20..v2.6.30 --dirstat=changes > /dev/null
vs.
time git diff v2.6.20..v2.6.30 --dirstat=lines > /dev/null
vs.
time git diff v2.6.20..v2.6.30 --dirstat=files > /dev/null
yields the following average runtimes on my machine:
- "changes" (default): ~6.0 s
- "lines": ~9.6 s
- "files": ~0.1 s
So, as expected, there's a considerable performance hit (~60%) by going
through the full diffstat analysis as compared to the default "changes"
analysis (obviously, "files" is much faster than both). As such, the
"lines" mode is probably only useful if you really need the --dirstat
numbers to be consistent with the numbers returned from the other
--*stat options.
The patch also includes documentation and tests for the new dirstat mode.
Improved-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Only the first digit after the decimal point is kept, as the dirstat
calculations all happen in permille.
Selftests verifying floating-point percentage input has been added.
Improved-by: Junio C Hamano <gitster@pobox.com>
Improved-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The new diff.dirstat config variable takes the same arguments as
'--dirstat=<args>', and specifies the default arguments for --dirstat.
The config is obviously overridden by --dirstat arguments passed on the
command line.
When not specified, the --dirstat defaults are 'changes,noncumulative,3'.
The patch also adds several tests verifying the interaction between the
diff.dirstat config variable, and the --dirstat command line option.
Improved-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Instead of having multiple interconnected dirstat-related options, teach
the --dirstat option itself to accept all behavior modifiers as parameters.
- Preserve the current --dirstat=<limit> (where <limit> is an integer
specifying a cut-off percentage)
- Add --dirstat=cumulative, replacing --cumulative
- Add --dirstat=files, replacing --dirstat-by-file
- Also add --dirstat=changes and --dirstat=noncumulative for specifying the
current default behavior. These allow the user to reset other --dirstat
parameters (e.g. 'cumulative' and 'files') occuring earlier on the
command line.
The deprecated options (--cumulative and --dirstat-by-file) are still
functional, although they have been removed from the documentation.
Allow multiple parameters to be separated by commas, e.g.:
--dirstat=files,10,cumulative
Update the documentation accordingly, and add testcases verifying the
behavior of the new syntax.
Improved-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The expected output from --dirstat=0, is to include any directory with
changes, even if those changes contribute a minuscule portion of the total
changes. However, currently, directories that contribute less than 0.1% are
not included, since their 'permille' value is 0, and there is an
'if (permille)' check in gather_dirstat() that causes them to be ignored.
This test is obviously intended to exclude directories that contribute no
changes whatsoever, but in this case, it hits too broadly. The correct
check is against 'this_dir' from which the permille is calculated. Only if
this value is 0 does the directory truly contribute no changes, and should
be skipped from the output.
This patches fixes this issue, and updates corresponding testcases to
expect the new behvaior.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jc/rename-degrade-cc-to-c:
diffcore-rename: fall back to -C when -C -C busts the rename limit
diffcore-rename: record filepair for rename src
diffcore-rename: refactor "too many candidates" logic
builtin/diff.c: remove duplicated call to diff_result_code()
* jh/dirstat:
--dirstat: In case of renames, use target filename instead of source filename
Teach --dirstat not to completely ignore rearranged lines within a file
--dirstat-by-file: Make it faster and more correct
--dirstat: Describe non-obvious differences relative to --stat or regular diff
e9c8409 (diff-index --cached --raw: show tree entry on the LHS for
unmerged entries., 2007-01-05) added a <mode, object name> pair as
parameters to this function, to store them in the pre-image side of an
unmerged file pair. Now the function is fixed to return the filepair it
queued, we can make the caller on the special case codepath to do so.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The underlying diff_queue() returns diff_filepair so that the caller can
further add information to it, and the helper function diff_unmerge()
utilizes the feature itself, but does not expose it to its callers, which
was kind of selfish.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This changes --dirstat analysis to count "damage" toward the target filename,
rather than the source filename. For renames within a directory, this won't
matter to the final output, but when moving files between diretories, the
output now lists the target directory rather than the source directory.
Signed-off-by: Johan Herland <johan@herland.net>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Currently, the --dirstat analysis ignores when lines within a file are
rearranged, because the "damage" calculated by show_dirstat() is 0.
However, if the object name has changed, we already know that there is
some damage, and it is unintuitive to claim there is _no_ damage.
Teach show_dirstat() to assign a minimum amount of damage (== 1) to
entries for which the analysis otherwise yields zero damage, to still
represent that these files are changed, instead of saying that there
is no change.
Also, skip --dirstat analysis when the object names are the same (e.g. for
a pure file rename).
Signed-off-by: Johan Herland <johan@herland.net>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Currently, when using --dirstat-by-file, it first does the full --dirstat
analysis (using diffcore_count_changes()), and then resets 'damage' to 1,
if any damage was found by diffcore_count_changes().
But --dirstat-by-file is not interested in the file damage per se. It only
cares if the file changed at all. In that sense it only cares if the blob
object for a file has changed. We therefore only need to compare the
object names of each file pair in the diff queue and we can skip the
entire --dirstat analysis and simply set 'damage' to 1 for each entry
where the object name has changed.
This makes --dirstat-by-file faster, and also bypasses --dirstat's practice
of ignoring rearranged lines within a file.
The patch also contains an added testcase verifying that --dirstat-by-file
now detects changes that only rearrange lines within a file.
Signed-off-by: Johan Herland <johan@herland.net>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When reviewing a patch while concentrating primarily on the text after
then change, wading through pages of deleted text involves a cognitive
burden.
Introduce the -D option that omits the preimage text from the patch output
for deleted files. When used with -B (represent total rewrite as a single
wholesale deletion followed by a single wholesale addition), the preimage
text is also omitted.
To prevent such a patch from being applied by mistake, the output is
designed not to be usable by "git apply" (or GNU "patch"); it is strictly
for human consumption.
It of course is possible to "apply" such a patch by hand, as a human can
read the intention out of such a patch. It however is impossible to apply
such a patch even manually in reverse, as the whole point of this option
is to omit the information necessary to do so from the output.
Initial request by Mart Sõmermaa, documentation and tests helped by
Michael J Gruber.
Signed-off-by: Michael J Gruber <git@drmicha.warpmail.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When there are too many paths in the project, the number of rename source
candidates "git diff -C -C" finds will exceed the rename detection limit,
and no inexact rename detection is performed. We however could fall back
to "git diff -C" if the number of modified paths is sufficiently small.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix warnings from 'make check'.
- These files don't include 'builtin.h' causing sparse to complain that
cmd_* isn't declared:
builtin/clone.c:364, builtin/fetch-pack.c:797,
builtin/fmt-merge-msg.c:34, builtin/hash-object.c:78,
builtin/merge-index.c:69, builtin/merge-recursive.c:22
builtin/merge-tree.c:341, builtin/mktag.c:156, builtin/notes.c:426
builtin/notes.c:822, builtin/pack-redundant.c:596,
builtin/pack-refs.c:10, builtin/patch-id.c:60, builtin/patch-id.c:149,
builtin/remote.c:1512, builtin/remote-ext.c:240,
builtin/remote-fd.c:53, builtin/reset.c:236, builtin/send-pack.c:384,
builtin/unpack-file.c:25, builtin/var.c:75
- These files have symbols which should be marked static since they're
only file scope:
submodule.c:12, diff.c:631, replace_object.c:92, submodule.c:13,
submodule.c:14, trace.c:78, transport.c:195, transport-helper.c:79,
unpack-trees.c:19, url.c:3, url.c:18, url.c:104, url.c:117, url.c:123,
url.c:129, url.c:136, thread-utils.c:21, thread-utils.c:48
- These files redeclare symbols to be different types:
builtin/index-pack.c:210, parse-options.c:564, parse-options.c:571,
usage.c:49, usage.c:58, usage.c:63, usage.c:72
- These files use a literal integer 0 when they really should use a NULL
pointer:
daemon.c:663, fast-import.c:2942, imap-send.c:1072, notes-merge.c:362
While we're in the area, clean up some unused #includes in builtin files
(mostly exec_cmd.h).
Signed-off-by: Stephen Boyd <bebarino@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a struct definitions, unlike functions, the prevailing style is for
the opening brace to go on the same line as the struct name, like so:
struct foo {
int bar;
char *baz;
};
Indeed, grepping for 'struct [a-z_]* {$' yields about 5 times as many
matches as 'struct [a-z_]*$'.
Linus sayeth:
Heretic people all over the world have claimed that this inconsistency
is ... well ... inconsistent, but all right-thinking people know that
(a) K&R are _right_ and (b) K&R are right.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We only need the size, which is much cheaper to get,
especially if it is a big binary file.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The logic in builtin_diffstat assumes that a
complete_rewrite pair should have its lines counted. This is
nonsensical for binary files and leads to confusing things
like:
$ git diff --stat --summary HEAD^ HEAD
foo.rand | Bin 4096 -> 4096 bytes
1 files changed, 0 insertions(+), 0 deletions(-)
$ git diff --stat --summary -B HEAD^ HEAD
foo.rand | 34 +++++++++++++++-------------------
1 files changed, 15 insertions(+), 19 deletions(-)
rewrite foo.rand (100%)
So let's reorder the function to handle binary files first
(which from diffstat's perspective look like complete
rewrites anyway), then rewrites, then actual diffstats.
There are two bonus prizes to this reorder:
1. It gets rid of a now-superfluous goto.
2. The binary case is at the top, which means we can
further optimize it in the next patch.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We did this once before in 5070591 (bump rename limit
defaults, 2008-04-30). Back then, we were shooting for about
1 second for a diff/log calculation, and 5 seconds for a
merge.
There are a few new things to consider, though:
1. Average processors are faster now.
2. We've seen on the mailing list some ugly merges where
not using inexact rename detection leads to many more
conflicts. Merges of this size take a long time
anyway, so users are probably happy to spend a little
bit of time computing the renames.
Let's bump the diff/merge default limits from 200/500 to
400/1000. Those are 2 seconds and 10 seconds respectively on
my modern hardware.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* ks/blame-worktree-textconv-cached:
fill_textconv(): Don't get/put cache if sha1 is not valid
t/t8006: Demonstrate blame is broken when cachetextconv is on
When blaming files in the working tree, the filespec is marked with
!sha1_valid, as we have not given the contents an object name yet. The
function to cache textconv results (keyed on the object name), however,
didn't check this condition, and ended up on storing the cached result
under a random object name.
Cc: Axel Bonnet <axel.bonnet@ensimag.imag.fr>
Cc: Clément Poulain <clement.poulain@ensimag.imag.fr>
Cc: Diane Gasselin <diane.gasselin@ensimag.imag.fr>
Cc: Jeff King <peff@peff.net>
Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* kb/diff-C-M-synonym:
diff: use "find" instead of "detect" as prefix for long forms of -M and -C
diff: add --detect-copies-harder as a synonym for --find-copies-harder
It is more consistent with existing --find-copies-harder; luckily "detect"
variant has not appeared in any officially released version of git.
Signed-off-by: Yann Dirson <ydirson@altern.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
