The code was not prepared to deal with pack .idx file that is
larger than 4GB.
* jk/4gb-idx:
packfile: detect overflow in .idx file size checks
block-sha1: take a size_t length parameter
fsck: correctly compute checksums on idx files larger than 4GB
use size_t to store pack .idx byte offsets
compute pack .idx byte offsets using size_t
The exchange between receive-pack and proc-receive hook did not
carefully check for errors.
* jx/t5411-flake-fix:
receive-pack: use default version 0 for proc-receive
receive-pack: gently write messages to proc-receive
t5411: new helper filter_out_user_friendly_and_stable_output
A specialization of hashmap that uses a string as key has been
introduced. Hopefully it will see wider use over time.
* en/strmap:
shortlog: use strset from strmap.h
Use new HASHMAP_INIT macro to simplify hashmap initialization
strmap: take advantage of FLEXPTR_ALLOC_STR when relevant
strmap: enable allocations to come from a mem_pool
strmap: add a strset sub-type
strmap: split create_entry() out of strmap_put()
strmap: add functions facilitating use as a string->int map
strmap: enable faster clearing and reusing of strmaps
strmap: add more utility functions
strmap: new utility functions
hashmap: provide deallocation function names
hashmap: introduce a new hashmap_partial_clear()
hashmap: allow re-use after hashmap_free()
hashmap: adjust spacing to fix argument alignment
hashmap: add usage documentation explaining hashmap_free[_entries]()
"git rev-parse" learned the "--end-of-options" to help scripts to
safely take a parameter that is supposed to be a revision, e.g.
"git rev-parse --verify -q --end-of-options $rev".
* jk/rev-parse-end-of-options:
rev-parse: handle --end-of-options
rev-parse: put all options under the "-" check
rev-parse: don't accept options after dashdash
The maximum length of output filenames "git format-patch" creates
has become configurable (used to be capped at 64).
* jc/format-patch-name-max:
format-patch: make output filename configurable
compare_tasks_by_selection() is used with QSORT and gets passed pointers
to the elements of "static struct maintenance_task tasks[]". It casts
the *addresses* of these passed pointers to element pointers, though,
and thus effectively compares some unrelated values from the stack. Fix
the casts to actually compare array elements.
Detected by USan (make SANITIZE=undefined test).
Signed-off-by: René Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git blame -L :funcname -- path" did not work well for a path for
which a userdiff driver is defined.
* pb/blame-funcname-range-userdiff:
blame: simplify 'setup_blame_bloom_data' interface
blame: simplify 'setup_scoreboard' interface
blame: enable funcname blaming with userdiff driver
line-log: mention both modes in 'blame' and 'log' short help
doc: add more pointers to gitattributes(5) for userdiff
blame-options.txt: also mention 'funcname' in '-L' description
doc: line-range: improve formatting
doc: log, gitk: move '-L' description to 'line-range-options.txt'
Preparation for a new merge strategy.
* en/merge-ort-api-null-impl:
merge,rebase,revert: select ort or recursive by config or environment
fast-rebase: demonstrate merge-ort's API via new test-tool command
merge-ort-wrappers: new convience wrappers to mimic the old merge API
merge-ort: barebones API of new merge strategy with empty implementation
Parts of "git maintenance" to ease writing crontab entries (and
other scheduling system configuration) for it.
* ds/maintenance-part-3:
maintenance: add troubleshooting guide to docs
maintenance: use 'incremental' strategy by default
maintenance: create maintenance.strategy config
maintenance: add start/stop subcommands
maintenance: add [un]register subcommands
for-each-repo: run subcommands on configured repos
maintenance: add --schedule option and config
maintenance: optionally skip --auto process
"git rebase -i" did not store ORIG_HEAD correctly.
* pw/rebase-i-orig-head:
rebase -i: simplify get_revision_ranges()
rebase -i: use struct object_id when writing state
rebase -i: use struct object_id rather than looking up commit
rebase -i: stop overwriting ORIG_HEAD buffer
"git format-patch --output=there" did not work as expected and
instead crashed. The option is now supported.
* jk/format-patch-output:
format-patch: support --output option
format-patch: tie file-opening logic to output_directory
format-patch: refactor output selection
"git log -L<range>:<path>" is documented to take no pathspec, but
this was not enforced by the command line option parser, which has
been corrected.
* jc/line-log-takes-no-pathspec:
log: diagnose -L used with pathspec as an error
The code to see if "git stash drop" can safely remove refs/stash
has been made more carerful.
* rs/empty-reflog-check-fix:
stash: simplify reflog emptiness check
We sometimes store the offset into a pack .idx file as an "unsigned
long", but the mmap'd size of a pack .idx file can exceed 4GB. This is
sufficient on LP64 systems like Linux, but will be too small on LLP64
systems like Windows, where "unsigned long" is still only 32 bits. Let's
use size_t, which is a better type for an offset into a memory buffer.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A pack and its matching .idx file are limited to 2^32 objects, because
the pack format contains a 32-bit field to store the number of objects.
Hence we use uint32_t in the code.
But the byte count of even a .idx file can be much larger than that,
because it stores at least a hash and an offset for each object. So
using SHA-1, a v2 .idx file will cross the 4GB boundary at 153,391,650
objects. This confuses load_idx(), which computes the minimum size like
this:
unsigned long min_size = 8 + 4*256 + nr*(hashsz + 4 + 4) + hashsz + hashsz;
Even though min_size will be big enough on most 64-bit platforms, the
actual arithmetic is done as a uint32_t, resulting in a truncation. We
actually exceed that min_size, but then we do:
unsigned long max_size = min_size;
if (nr)
max_size += (nr - 1)*8;
to account for the variable-sized table. That computation doesn't
overflow quite so low, but with the truncation for min_size, we end up
with a max_size that is much smaller than our actual size. So we
complain that the idx is invalid, and can't find any of its objects.
We can fix this case by casting "nr" to a size_t, which will do the
multiplication in 64-bits (assuming you're on a 64-bit platform; this
will never work on a 32-bit system since we couldn't map the whole .idx
anyway). Likewise, we don't have to worry about further additions,
because adding a smaller number to a size_t will convert the other side
to a size_t.
A few notes:
- obviously we could just declare "nr" as a size_t in the first place
(and likewise, packed_git.num_objects). But it's conceptually a
uint32_t because of the on-disk format, and we correctly treat it
that way in other contexts that don't need to compute byte offsets
(e.g., iterating over the set of objects should and generally does
use a uint32_t). Switching to size_t would make all of those other
cases look wrong.
- it could be argued that the proper type is off_t to represent the
file offset. But in practice the .idx file must fit within memory,
because we mmap the whole thing. And the rest of the code (including
the idx_size variable we're comparing against) uses size_t.
- we'll add the same cast to the max_size arithmetic line. Even though
we're adding to a larger type, which will convert our result, the
multiplication is still done as a 32-bit value and can itself
overflow. I didn't check this with my test case, since it would need
an even larger pack (~530M objects), but looking at compiler output
shows that it works this way. The standard should agree, but I
couldn't find anything explicit in 6.3.1.8 ("usual arithmetic
conversions").
The case in load_idx() was the most immediate one that I was able to
trigger. After fixing it, looking up actual objects (including the very
last one in sha1 order) works in a test repo with 153,725,110 objects.
That's because bsearch_hash() works with uint32_t entry indices, and the
actual byte access:
int cmp = hashcmp(table + mi * stride, sha1);
is done with "stride" as a size_t, causing the uint32_t "mi" to be
promoted to a size_t. This is the way most code will access the index
data.
However, I audited all of the other byte-wise accesses of
packed_git.index_data, and many of the others are suspect (they are
similar to the max_size one, where we are adding to a properly sized
offset or directly to a pointer, but the multiplication in the
sub-expression can overflow). I didn't trigger any of these in practice,
but I believe they're potential problems, and certainly adding in the
cast is not going to hurt anything here.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Now that hashamp has lazy initialization and a HASHMAP_INIT macro,
hashmaps allocated on the stack can be initialized without a call to
hashmap_init() and in some cases makes the code a bit shorter. Convert
some callsites over to take advantage of this.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the verison negotiation phase between "receive-pack" and
"proc-receive", "proc-receive" can send an empty flush-pkt to end the
negotiation and use default version 0. Capabilities (such as
"push-options") are not supported in version 0.
Signed-off-by: Jiang Xin <zhiyou.jx@alibaba-inc.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Johannes found a flaky hang in `t5411/test-0013-bad-protocol.sh` in the
osx-clang job of the CI/PR builds, and ran into an issue when using
the `--stress` option with the following error messages:
fatal: unable to write flush packet: Broken pipe
send-pack: unexpected disconnect while reading sideband packet
fatal: the remote end hung up unexpectedly
In this test case, the "proc-receive" hook sends an error message and
dies earlier. While "receive-pack" on the other side of the pipe
should forward the error message of the "proc-receive" hook to the
client side, but it fails to do so. This is because "receive-pack"
uses `packet_write_fmt()` and `packet_flush()` to write pkt-line
message to "proc-receive" hook, and these functions die immediately
when pipe is broken. Using "gently" forms for these functions will get
more predicable output.
Add more "--die-*" options to test helper to test different stages of
the protocol between "receive-pack" and "proc-receive" hook.
Reported-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Suggested-by: Jeff King <peff@peff.net>
Signed-off-by: Jiang Xin <zhiyou.jx@alibaba-inc.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We taught rev-list a new way to separate options from revisions in
19e8789b23 (revision: allow --end-of-options to end option parsing,
2019-08-06), but rev-parse uses its own parser. It should know about
--end-of-options not only for consistency, but because it may be
presented with similarly ambiguous cases. E.g., if a caller does:
git rev-parse "$rev" -- "$path"
to parse an untrusted input, then it will get confused if $rev contains
an option-like string like "--local-env-vars". Or even "--not-real",
which we'd keep as an option to pass along to rev-list.
Or even more importantly:
git rev-parse --verify "$rev"
can be confused by options, even though its purpose is safely parsing
untrusted input. On the plus side, it will always fail the --verify
part, as it will not have parsed a revision, so the caller will
generally "fail closed" rather than continue to use the untrusted
string. But it will still trigger whatever option was in "$rev"; this
should be mostly harmless, since rev-parse options are all read-only,
but I didn't carefully audit all paths.
This patch lets callers write:
git rev-parse --end-of-options "$rev" -- "$path"
and:
git rev-parse --verify --end-of-options "$rev"
which will both treat "$rev" always as a revision parameter. The latter
is a bit clunky. It would be nicer if we had defined "--verify" to
require that its next argument be the revision. But we have not
historically done so, and:
git rev-parse --verify -q "$rev"
does currently work. I added a test here to confirm that we didn't break
that.
A few implementation notes:
- We don't document --end-of-options explicitly in commands, but rather
in gitcli(7). So I didn't give it its own section in git-rev-parse(1).
But I did call it out specifically in the --verify section, and
include it in the examples, which should show best practices.
- We don't have to re-indent the main option-parsing block, because we
can combine our "did we see end of options" check with "does it start
with a dash". The exception is the pre-setup options, which need
their own block.
- We do however have to pull the "--" parsing out of the "does it start
with dash" block, because we want to parse it even if we've seen
--end-of-options.
- We'll leave "--end-of-options" in the output. This is probably not
technically necessary, as a careful caller will do:
git rev-parse --end-of-options $revs -- $paths
and anything in $revs will be resolved to an object id. However, it
does help a slightly less careful caller like:
git rev-parse --end-of-options $revs_or_paths
where a path "--foo" will remain in the output as long as it also
exists on disk. In that case, it's helpful to retain --end-of-options
to get passed along to rev-list, s it would otherwise see just
"--foo".
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The option-parsing loop of rev-parse checks whether the first character
of an arg is "-". If so, then it enters a series of conditionals
checking for individual options. But some options are inexplicably
outside of that outer conditional.
This doesn't produce the wrong behavior; the conditional is actually
redundant with the individual option checks, and it's really only its
fallback "continue" that we care about. But we should at least be
consistent.
One obvious alternative is that we could get rid of the conditional
entirely. But we'll be using the extra block it provides in the next
patch.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Because of the order in which we check options in rev-parse, there are a
few options we accept even after a "--". This is wrong, because the
whole point of "--" is to say "everything after here is a path". Let's
move the "did we see a dashdash" check (it's called "as_is" in the code)
to the top of the parsing loop.
Note there is one subtlety here. The options are ordered so that some
are checked before we even see if we're in a repository (they continue
the loop, and if we get past a certain point, then we do the repository
setup). By moving the as_is check higher, it's also in that "before
setup" section, even though it might look at the repository via
verify_filename(). However, this works out: we'd never set as_is until
we parse "--", and we don't parse that until after doing the setup.
An alternative here to avoid the subtlety is to put the as_is check at
the top of the post-setup options. But then every pre-setup option would
have to remember to check "if (!as_is && !strcmp(...))". So while this
is a bit magical, it's harder for future code to get wrong.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
For the past 15 years, we've used the hardcoded 64 as the length
limit of the filename of the output from the "git format-patch"
command. Since the value is shorter than the 80-column terminal, it
could grow without line wrapping a bit. At the same time, since the
value is longer than half of the 80-column terminal, we could fit
two or more of them in "ls" output on such a terminal if we allowed
to lower it.
Introduce a new command line option --filename-max-length=<n> and a
new configuration variable format.filenameMaxLength to override the
hardcoded default.
While we are at it, remove a check that the name of output directory
does not exceed PATH_MAX---this check is pointless in that by the
time control reaches the function, the caller would already have
done an equivalent of "mkdir -p", so if the system does not like an
overly long directory name, the control wouldn't have reached here,
and otherwise, we know that the system allowed the output directory
to exist. In the worst case, we will get an error when we try to
open the output file and handle the error correctly anyway.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Exit codes from "git remote add" etc. were not usable by scripted
callers.
* ab/git-remote-exit-code:
remote: add meaningful exit code on missing/existing
"git checkout-index" did not consistently signal an error with its
exit status.
* jk/checkout-index-errors:
checkout-index: propagate errors to exit code
checkout-index: drop error message from empty --stage=all
Now that all the external users of head_hash have been converted to
use a opts->orig_head instead we can stop returning head_hash from
get_revision_ranges().
Because we want to pass the full object names back to the caller in
`revisions` the find_unique_abbrev_r() call that was used to initialize
`head_hash` is replaced with oid_to_hex().
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Rather than passing a string around pass the struct object_id that the
string was created from call oid_hex() when we write the file.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We already have a struct object_id containing the oid that we want to
set ORIG_HEAD to so use that rather than converting it to a string and
then calling get_oid() on that string.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
After rebasing, ORIG_HEAD is supposed to point to the old HEAD of the
rebased branch. The code used find_unique_abbrev() to obtain the
object name of the old HEAD and wrote to both
.git/rebase-merge/orig-head (used by `rebase --abort` to go back to
the previous state) and to ORIG_HEAD. The buffer find_unique_abbrev()
gives back is volatile, unfortunately, and was overwritten after the
former file is written but before ORIG_FILE is written, leaving an
incorrect object name in it.
Avoid relying on the volatile buffer of find_unique_abbrev(), and
instead supply our own buffer to keep the object name.
I think that all of the users of head_hash should actually be using
opts->orig_head instead as passing a string rather than a struct
object_id around is a hang over from the scripted implementation. This
patch just fixes the immediate bug and adds a regression test based on
Caspar's reproduction example[1]. The users will be converted to use
struct object_id and head_hash removed in the next few commits.
[1] https://lore.kernel.org/git/CAFzd1+7PDg2PZgKw7U0kdepdYuoML9wSN4kofmB_-8NHrbbrHg@mail.gmail.com
Reported-by: Caspar Duregger <herr.kaste@gmail.com>
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We've never intended to support diff's --output option in format-patch.
And until baa4adc66a (parse-options: disable option abbreviation with
PARSE_OPT_KEEP_UNKNOWN, 2019-01-27), it was impossible to trigger. We
first parse the format-patch options before handing the remainder off to
setup_revisions(). Before that commit, we'd accept "--output=foo" as an
abbreviation for "--output-directory=foo". But afterwards, we don't
check abbreviations, and --output gets passed to the diff code.
This results in nonsense behavior and bugs. The diff code will have
opened a filehandle at rev.diffopt.file, but we'll overwrite that with
our own handles that we open for each individual patch file. So the
--output file will always just be empty. But worse, the diff code also
sets rev.diffopt.close_file, so log_tree_commit() will close the
filehandle itself. And then the main loop in cmd_format_patch() will try
to close it again, resulting in a double-free.
The simplest solution would be to just disallow --output with
format-patch, as nobody ever intended it to work. However, we have
accidentally documented it (because format-patch includes diff-options).
And it does work with "git log", which writes the whole output to the
specified file. It's easy enough to make that work for format-patch,
too: it's really the same as --stdout, but pointed at a specific file.
We can detect the use of the --output option by the "close_file" flag
(note that we can't use rev.diffopt.file, since the diff setup will
otherwise set it to stdout). So we just need to unset that flag, but
don't have to do anything else. Our situation is otherwise exactly like
--stdout (note that we don't fclose() the file, but nor does the stdout
case; exiting the program takes care of that for us).
Reported-by: Johannes Postler <johannes.postler@txture.io>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In format-patch we're either outputting to stdout or to individual files
in an output directory (which may be just "./"). Our logic for whether
to open a new file for each patch is checked with "!use_stdout", but it
is equally correct to check for a non-NULL output_directory.
The distinction will matter when we add a new single-stream output in a
future patch, when only one of the three methods will want individual
files. Let's swap the logic here in preparation.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The --stdout and --output-directory options are mutually exclusive, but
it's hard to tell from reading the code. We have three separate
conditionals that check for use_stdout, and it's only after we've set up
the output_directory fully that we check whether the user also specified
--stdout.
Instead, let's check the exclusion explicitly first, then have a single
conditional that handles stdout versus an output directory. This is
slightly easier to follow now, and also will keep things sane when we
add another output mode in a future patch.
We'll add a few tests as well, covering the mutual exclusion and the
fact that we are not confused by a configured output directory.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The -L option is documented to accept no pathspec, but the
command line option parser has allowed the combination without
checking so far. Ensure that there is no pathspec when the -L
option is in effect to fix this.
Incidentally, this change fixes another bug in the command line
option parser, which has allowed the -L option used together
with the --follow option. Because the latter requires exactly
one path given, but the former takes no pathspec, they become
mutually incompatible automatically. Because the -L option
follows renames on its own, there is no reason to give --follow
at the same time.
The new tests say they may fail with "-L and --follow being
incompatible" instead of "-L and pathspec being incompatible".
Currently the expected failure can come only from the latter, but
this is to futureproof them, in case we decide to add code to
explicititly die on -L and --follow used together.
Heled-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Allow the testsuite to run where it treats requests for "recursive" or
the default merge algorithm via consulting the environment variable
GIT_TEST_MERGE_ALGORITHM which is expected to either be "recursive" (the
old traditional algorithm) or "ort" (the new algorithm).
Also, allow folks to pick the new algorithm via config setting. It
turns out builtin/merge.c already had a way to allow users to specify a
different default merge algorithm: pull.twohead. Rather odd
configuration name (especially to be in the 'pull' namespace rather than
'merge') but it's there. Add that same configuration to rebase,
cherry-pick, and revert.
This required updating the various callsites that called merge_trees()
or merge_recursive() to conditionally call the new API, so this serves
as another demonstration of what the new API looks and feels like.
There are almost certainly some callsites that have not yet been
modified to work with the new merge algorithm, but this represents the
ones that I have been testing with thus far.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git diff" family of commands learned the "-I<regex>" option to
ignore hunks whose changed lines all match the given pattern.
* mk/diff-ignore-regex:
diff: add -I<regex> that ignores matching changes
merge-base, xdiff: zero out xpparam_t structures
"git credential' didn't honor the core.askPass configuration
variable (among other things), which has been corrected.
* tk/credential-config:
credential: load default config
Document that the meaning of a Signed-off-by trailer can vary from
project to project in the end-user documentation, and clarify what
it means to this project.
* bk/sob-dco:
Documentation: stylistically normalize references to Signed-off-by:
SubmittingPatches: clarify DCO is our --signoff rule
Documentation: clarify and expand description of --signoff
doc: preparatory clean-up of description on the sign-off option
Test-coverage enhancement of running commit-graph task "git
maintenance" as needed led to discovery and fix of a bug.
* ds/maintenance-commit-graph-auto-fix:
maintenance: core.commitGraph=false prevents writes
maintenance: test commit-graph auto condition
"git fast-import" wasted a lot of memory when many marks were in use.
* jk/fast-import-marks-alloc-fix:
fast-import: fix over-allocation of marks storage
hashmap_free(), hashmap_free_entries(), and hashmap_free_() have existed
for a while, but aren't necessarily the clearest names, especially with
hashmap_partial_clear() being added to the mix and lazy-initialization
now being supported. Peff suggested we adopt the following names[1]:
- hashmap_clear() - remove all entries and de-allocate any
hashmap-specific data, but be ready for reuse
- hashmap_clear_and_free() - ditto, but free the entries themselves
- hashmap_partial_clear() - remove all entries but don't deallocate
table
- hashmap_partial_clear_and_free() - ditto, but free the entries
This patch provides the new names and converts all existing callers over
to the new naming scheme.
[1] https://lore.kernel.org/git/20201030125059.GA3277724@coredump.intra.peff.net/
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The penultimate commit moved the initialization of 'sb.path' in
'builtin/blame.c::cmd_blame' before the call to
'blame.c::setup_blame_bloom_data'. Since 'cmd_blame' is the only caller
of 'setup_blame_bloom_data', it is now unnecessary for
'setup_blame_bloom_data' to receive 'path' as a separate argument, as
'sb.path' is already initialized.
Remove this argument from setup_blame_bloom_data's interface and use the
'path' field of the 'sb' 'struct blame_scoreboard' instead.
Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The previous commit moved the initialization of 'sb.path' in
'builtin/blame.c::cmd_blame' before the call to
'blame.c::setup_scoreboard'. Since 'cmd_blame' is the only caller of
'setup_scoreboard', it is now unnecessary for 'setup_scoreboard' to
receive 'path' as a separate argument, as 'sb.path' is already
initialized.
Remove this argument from setup_scoreboard's interface and use the
'path' field of the 'sb' 'struct blame_scoreboard' instead.
Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In blame.c::cmd_blame, we send the 'path' field of the 'sb' 'struct
blame_scoreboard' as the 'path' argument to
'line-range.c::parse_range_arg', but 'sb.path' is not set yet; it's set
to the local variable 'path' a few lines later at line 1137.
This 'path' argument is only used in 'parse_range_arg' if we are blaming
a funcname, i.e. `git blame -L :<funcname> <path>`, and in that case it
is sent to 'parse_range_funcname', where it is used to determine if a
userdiff driver should be used for said <path> to match the given
funcname.
Since 'path' is yet unset, the userdiff driver is never used, so we fall
back to the default funcname regex, which is usually not appropriate for
paths that are set to use a specific userdiff driver, and thus either we
match some unrelated lines, or we die with
fatal: -L parameter '<funcname>' starting at line 1: no match
This has been the case ever since `git blame` learned to blame a
funcname in 13b8f68c1f (log -L: :pattern:file syntax to find by
funcname, 2013-03-28).
Enable funcname blaming for paths using specific userdiff drivers by
initializing 'sb.path' earlier in 'cmd_blame', when some of its other
fields are initialized, so that it is set when passed to
'parse_range_arg'.
Add a regression test in 'annotate-tests.sh', which is sourced in
t8001-annotate.sh and t8002-blame.sh, leveraging an existing file used
to test the userdiff patterns in t4018-diff-funcname.
Also, use 'sb.path' instead of 'path' when constructing the error
message at line 1114, for consistency.
Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
'git blame -h' and 'git log -h' both show '-L <n,m>' and describe this
option as "Process only line range n,m, counting from 1". No hint is
given that a function name regex can also be used.
Use <range> instead, and expand the description of the option to mention
both modes. Remove "counting from 1" as it's uneeded; it's uncommon to
refer to the first line of a file as "line 0".
Also, for 'git log', improve the wording to better reflect the long help.
Signed-off-by: Philippe Blain <levraiphilippeblain@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
