|
0903cac1
|
2021-08-11T01:30:38
|
|
openssl: dynamically load libssl and symbols (optionally)
Provide an interface around OpenSSL to dynamically load the libraries
and symbols, so that users can distribute a libgit2 library that is not
linked directly against OpenSSL. This enables users to target multiple
distributions with a single binary.
This mechanism is optional and disabled by default. Configure cmake
with -DUSE_HTTPS=OpenSSL-Dynamic to use it.
|
|
150eddd9
|
2021-08-10T21:41:05
|
|
openssl: separate legacy api
Refactor the OpenSSL stream implementation so that the legacy code is better
abstracted. This will enable future development.
|
|
1903cfef
|
2021-08-19T16:49:41
|
|
openssl: don't fail when we can't customize allocators
During valgrind runs, we try to swap out the OpenSSL allocators for our
own. This allows us to avoid some unnecessary warnings about usage.
Unfortunately, many builds of OpenSSL do not allow you to swap
allocators; for example FIPS builds and the builds running in CentOS.
Try to swap the allocators, but do not fail when they cannot be
customized.
|
|
727d903a
|
2021-08-18T21:01:36
|
|
sha1dc: remove conditional for <sys/types.h>
|
|
2dbe413e
|
2021-08-17T11:34:11
|
|
Fix LIBGIT2_FILENAME not being passed to the resource compiler
|
|
c1c59843
|
2021-08-14T11:15:51
|
|
Merge pull request #5972 from libgit2/ethomson/http_empty_password
http: don't require a password
|
|
5fd44239
|
2021-08-10T10:06:43
|
|
Merge pull request #5854 from alexjg/respect-force-flag-in-remote-fetch
Respect the force flag on refspecs in git_remote_fetch
|
|
28841241
|
2021-08-05T08:12:28
|
|
http: don't require a password
Attempt authentication when a username is presented but a password is
not; this can happen in particular when users are doing token
authentication and specifying the token in the URL itself. For example,
`https://token@host/` is a valid URI and should be treated as a username
of `token` with an empty password.
|
|
a5696702
|
2021-07-08T11:58:28
|
|
skip descendant check if remote ref not found
|
|
3062a633
|
2021-07-30T12:03:35
|
|
cmake: extended futimens checking on macOS
|
|
e5975f36
|
2021-07-30T11:37:12
|
|
tests: reset odb backend priority
|
|
cd460522
|
2020-04-20T22:16:52
|
|
odb: Implement option for overriding of default odb backend priority
Introduce GIT_OPT_SET_ODB_LOOSE_PRIORITY and GIT_OPT_SET_ODB_PACKED_PRIORITY
to allow overriding the default priority values for the default ODB
backends. Libgit2 has historically assumed that most objects for long-
running operations will be packed, therefore GIT_LOOSE_PRIORITY is
set to 1 by default, and GIT_PACKED_PRIORITY to 2.
When a client allows libgit2 to set the default backends, they can
specify an override for the two priority values in order to change
the order in which each ODB backend is accessed.
|
|
cc68c19a
|
2021-07-30T08:56:51
|
|
Merge branch 'pr/5861'
|
|
708d9336
|
2021-07-30T08:34:28
|
|
Merge pull request #5953 from boretrk/c99-inline
common.h: use inline when compiling for C99 and later
|
|
003a1df6
|
2021-07-29T08:48:17
|
|
Merge pull request #5952 from libgit2/ethomson/attrs_from_commit
Support reading attributes from a specific commit
|
|
8d453f16
|
2021-07-27T13:00:31
|
|
Swap the order of the `git_graph_reachable_from_any` params
len, array -> array, len
|
|
ce5400cd
|
2021-01-06T06:26:09
|
|
graph: Create `git_graph_reachable_from_any()`
This change introduces a new API function
`git_graph_reachable_from_any()`, that answers the question whether a
commit is reachable from any of the provided commits through following
parent edges.
This function can take advantage of optimizations provided by the
existence of a `commit-graph` file, since it makes it faster to know
whether, given two commits X and Y, X cannot possibly be an reachable
from Y.
Part of: #5757
|
|
6f544140
|
2021-01-05T19:45:23
|
|
commit-graph: Introduce `git_commit_list_generation_cmp`
This change makes calculations of merge-bases a bit faster when there
are complex graphs and the commit times cause visiting nodes multiple
times. This is done by visiting the nodes in the graph in reverse
generation order when the generation number is available instead of
commit timestamp. If the generation number is missing in any pair of
commits, it can safely fall back to the old heuristic with no negative
side-effects.
Part of: #5757
|
|
88a3b8e1
|
2021-07-26T23:12:12
|
|
common.h: use inline when compiling for c99 and later
|
|
2370e491
|
2021-07-26T16:27:54
|
|
Merge pull request #5765 from lhchavez/cgraph-revwalks
commit-graph: Use the commit-graph in revwalks
|
|
43b5075d
|
2021-07-22T17:07:56
|
|
Merge pull request #5890 from lolgear/git_submodule_dup
[Submodule] Git submodule dup
|
|
c87e4760
|
2021-07-22T16:51:11
|
|
Merge branch 'pr/5948' into main
|
|
fe40de40
|
2021-07-22T16:50:26
|
|
trace: drop unnecessary write indirection
|
|
1439b9ff
|
2021-07-22T15:29:54
|
|
filter: introduce GIT_BLOB_FILTER_ATTRIBUTES_FROM_COMMIT
Provide a mechanism to filter using attribute data from a specific
commit (making use of `GIT_ATTR_CHECK_INCLUDE_COMMIT`).
|
|
0bd547a8
|
2021-07-22T15:29:46
|
|
attr: introduce GIT_ATTR_CHECK_INCLUDE_COMMIT
Introduce `GIT_ATTR_CHECK_INCLUDE_COMMIT`, which like 4fd5748 allows
attribute information to be read from files in the repository. 4fd5748
always reads the information from HEAD, while
`GIT_ATTR_CHECK_INCLUDE_COMMIT` allows users to provide the commit to
read the attributes from.
|
|
093d579f
|
2021-07-22T15:07:35
|
|
attr: cache nonexistent attr files from commits
When looking up an attribute file in a commit, we can cache a
nonexistent attribute file indefinitely (since a commit could not
somehow later contain an attribute file). Cache an empty buffer when an
attribute file does not exist in a given commit.
|
|
d7e8b934
|
2021-06-16T09:08:38
|
|
filter: add git_filter_options
Allow filter users to provide an options structure instead of simply
flags. This allows for future growth for filter options.
|
|
1db5b219
|
2021-06-16T09:06:26
|
|
filter: filter options are now "filter sessions"
Filters use a short-lived structure to keep state during an operation to
allow for caching and avoid unnecessary reallocations. This was
previously called the "filter options", despite the fact that they
contain no configurable options. Rename them to a "filter session" in
keeping with an "attribute session", which more accurately describes
their use (and allows us to create "filter options" in the future).
|
|
3779a047
|
2021-05-27T18:47:22
|
|
attr: introduce `git_attr_options` for extended queries
Allow more advanced attribute queries using a `git_attr_options`, and
extended functions to use it. Presently there is no additional
configuration in a `git_attr_options` beyond the flags, but this is for
future growth.
|
|
1cd863fd
|
2021-05-24T13:44:45
|
|
attr: include the filename in the attr source
The attribute source object is now the type and the path.
|
|
96dc1ffd
|
2021-05-22T20:14:47
|
|
attr: the attr source is now a struct
We may want to extend the attribute source; use a structure instead of
an enum.
|
|
5ee50488
|
2021-05-22T18:47:03
|
|
attr: rename internal attr file source enum
The enum `git_attr_file_source` is better suffixed with a `_t` since
it's a type-of source. Similarly, its members should have a matching
name.
|
|
b0980dcf
|
2021-07-20T08:25:59
|
|
Update src/allocators/stdalloc.c
Co-authored-by: lhchavez <lhchavez@lhchavez.com>
|
|
688bf4ce
|
2021-07-19T19:25:58
|
|
Remove macro magic
|
|
31e84edb
|
2021-07-19T16:02:30
|
|
merge: don't try to malloc(0)
|
|
419ffdde
|
2021-07-19T15:51:53
|
|
packbuilder: don't try to malloc(0)
|
|
48e6b02b
|
2021-07-19T15:41:44
|
|
alloc: add GIT_DEBUG_STRICT_ALLOC
Add `GIT_DEBUG_STRICT_ALLOC` to help identify problematic callers of
allocation code that pass a `0` size to the allocators and then expect a
non-`NULL` return.
When given a 0-size allocation, `malloc` _may_ return either a `NULL`
_or_ a pointer that is not writeable. Most systems return a non-`NULL`
pointer; AIX is an outlier. We should be able to cope with this AIXy
behavior, so this adds an option to emulate it.
|
|
950a7f76
|
2021-07-17T16:21:05
|
|
Variadic arguments aren't in C89
This refactors this to a wrapper function, as well as changing the
existing API have a va_list and ... version.
|
|
6a7f0403
|
2021-07-16T08:47:37
|
|
Merge pull request #5941 from NattyNarwhal/stdintification
stdintification: use int64_t and INT64_C instead of long long
|
|
589654d8
|
2021-07-15T20:34:14
|
|
khash 64-bit type is unsigned, use int32_t too
For 32-bit int: There's no real reason for that ifdef if we explicitly
specify the stdint.h type.
For 64-bit int: ope, I made it signed accidentally
|
|
e4e173e8
|
2021-07-15T21:00:02
|
|
Allow compilation on systems without CLOCK_MONOTONIC
Makes usage of CLOCK_MONOTONIC conditional and makes functions that uses
git__timer handle clock resynchronization.
Call gettimeofday with tzp set to NULL as required by
https://pubs.opengroup.org/onlinepubs/9699919799/functions/gettimeofday.html
|
|
84ce9746
|
2021-07-14T08:39:24
|
|
Merge pull request #5824 from palmin/fix-ignore-negate
fix check for ignoring of negate rules
|
|
e5649e10
|
2021-07-14T08:30:16
|
|
Merge pull request #5938 from NattyNarwhal/time-type-mismatch
Fix wrong time_t used in function
|
|
97cd165e
|
2021-07-14T08:30:00
|
|
Merge pull request #5926 from Batchyx/batchyx/config-refresh-twice
config: fix included configs not refreshed more than once
|
|
95a2966f
|
2021-07-13T17:32:02
|
|
Merge pull request #5908 from punkymaniac/patch-mem-leak
Fix memory leak in git_smart__connect
|
|
fcfaea4d
|
2021-07-07T23:29:28
|
|
Remove sus unsigned long constant
|
|
52505ab5
|
2021-07-07T19:12:02
|
|
Convert long long constant specifiers to stdint macros
|
|
c1aca3fe
|
2021-07-06T23:25:13
|
|
Initial pass at using int64_t instead of long long
Even on systems without C99 where long long and stdint are both
missing, we can shim stdint and point it to any compiler-specific
type (i.e long long, _int64, etc.).
Also next is constant suffixes and determining what needs to
include stdint.
|
|
be67f512
|
2021-07-06T22:27:14
|
|
Fix wrong time_t used in function
This function doesn't interoperate with any system functions that
use the system time_t, but rather only works with the git_time_t
type in libgit2, which could be a different width than the system
one. Fixes a compile warning.
|
|
b444918a
|
2021-07-06T20:51:14
|
|
Limit ITimer usage to AmigaOS4
|
|
6c78fd06
|
2021-07-02T18:30:07
|
|
Merge pull request #5915 from novalis/executable
Consider files executable only if the user can execute them
|
|
39fd7646
|
2021-06-30T12:59:48
|
|
Merge branch 'main' into respect-force-flag-in-remote-fetch
|
|
a2cd10e5
|
2021-06-26T18:52:21
|
|
define WINHTTP_NO_CLIENT_CERT_CONTEXT if needed
|
|
4bf136b0
|
2021-06-23T16:53:53
|
|
config: fix included configs not refreshed more than once
If an included config is refreshed twice, the second update is not taken
into account.
This is because the list of included files is cleared after re-reading
the new configuration, instead of being cleared before.
Fix it and add a test case to check for this bug.
|
|
b0fd4cf8
|
2021-06-16T13:38:45
|
|
Consider files executable only if the user can execute them
This is what git.git does, so we should follow suit.
|
|
428f1acf
|
2021-06-15T17:33:03
|
|
submodule: git submodule dup out and source parameters have become mandatory.
|
|
6d2a6f3e
|
2021-06-15T00:22:30
|
|
Apply suggestions from code review
|
|
8cad7e62
|
2021-06-13T08:21:05
|
|
Update src/ignore.c
Co-authored-by: lhchavez <lhchavez@lhchavez.com>
|
|
9410f2b5
|
2021-06-13T08:20:59
|
|
Update src/ignore.c
Co-authored-by: lhchavez <lhchavez@lhchavez.com>
|
|
1713ab45
|
2021-05-02T13:41:59
|
|
Respect the force flag on refspecs in git_remote_fetch
|
|
d07a0dc1
|
2021-06-04T16:34:32
|
|
Remove useless condition
|
|
2934b447
|
2021-06-03T11:21:39
|
|
Fix memory leak in git_smart__connect
The call to git_proxy_options_dup will replace the url pointer of the
proxy.
But if the url pointer is already set, the old address will be lost
forever and will never be free.
|
|
cf9196bd
|
2021-05-30T10:42:25
|
|
Tolerate readlink size less than st_size
|
|
bf1616d1
|
2021-05-26T15:32:05
|
|
submodule: git submodule dup implementation has been reverted. Assert has been added.
|
|
11aa20b0
|
2021-05-26T14:06:31
|
|
submodule: git submodule dup object dup has been added.
|
|
d6f57b9c
|
2021-05-19T18:50:29
|
|
submodule: git submodule dup guard statement has been fixed.
|
|
1ee3c37f
|
2021-05-19T09:31:30
|
|
Merge branch 'pr/5853'
|
|
94571908
|
2021-05-18T16:07:04
|
|
branch: git branch upstream with format format name parameter has been added.
|
|
a6fb72a8
|
2021-05-18T11:51:25
|
|
tests: exit with error on win32 leakcheck
|
|
49f9941d
|
2021-05-16T23:54:25
|
|
submodule: git submodule dup has been added.
|
|
9836ca91
|
2021-05-16T18:02:23
|
|
branch: git branch upstream functions layouts have been fixed.
|
|
ea889b3e
|
2021-05-16T17:58:15
|
|
branch: git branch upstream format enum has been added.
|
|
b5dcdad3
|
2021-05-16T12:53:58
|
|
Merge pull request #5852 from implausible/httpclient/skip-entire-body
Fix issues with Proxy Authentication after httpclient refactor
|
|
f7c3d622
|
2021-05-16T13:58:51
|
|
branch: git branch upstream functions layouts have been fixed.
|
|
58944388
|
2021-05-16T11:11:56
|
|
Merge branch 'zero_oid_in_old'
Manually merging #5842
|
|
4c462bb6
|
2021-04-13T13:20:34
|
|
diff: Fix bug in git_diff_find_similar.
When a split src gets turned into a rename, it should
also lose the IS_RENAME_TARGET flag, so that it doesn't
get processed in a subsequent iteration as a rename target.
Doing so can cause an assertion failure because it no
longer has the SPLIT flag.
Fixes #5811.
|
|
3473a088
|
2021-05-12T11:48:23
|
|
httpclient: no proxy creds in requests if proxy is CONNECT type
|
|
049618ce
|
2021-04-30T15:11:54
|
|
httpclient: git_http_client_skip_body should drain socket of body
|
|
7d9599bd
|
2021-05-12T10:30:12
|
|
branch: git branch upstream merge has been exposed.
|
|
a2d953e5
|
2021-05-12T10:29:45
|
|
branch: git branch upstream with format has been added.
|
|
ff78aea6
|
2021-05-11T11:09:31
|
|
Merge pull request #5860 from libgit2/ethomson/buf_text
buf: remove unnecessary buf_text namespace
|
|
14f6950b
|
2021-05-10T23:14:17
|
|
buf: bom enum is in the buf namespace
Instead of a `git_bom_t` that a `git_buf` function returns, let's keep
it `git_buf_bom_t`.
|
|
d525e063
|
2021-05-10T23:04:59
|
|
buf: remove internal `git_buf_text` namespace
The `git_buf_text` namespace is unnecessary and strange. Remove it,
just keep the functions prefixed with `git_buf`.
|
|
31d9c24b
|
2021-05-06T16:32:14
|
|
filter: internal git_buf filter handling function
Introduce `git_filter_list__convert_buf` which behaves like the old
implementation of `git_filter_list__apply_data`, where it might move the
input data buffer over into the output data buffer space for efficiency.
This new implementation will do so in a more predictible way, always
freeing the given input buffer (either moving it to the output buffer or
filtering it into the output buffer first).
Convert internal users to it.
|
|
68b9605a
|
2021-05-06T15:37:31
|
|
filter: deprecate git_filter_list_apply_to_data
Deprecate `git_filter_list_apply_to_data` as it takes user input as a
`git_buf`. Users should use `git_filter_list_apply_to_buffer` instead.
|
|
5309b465
|
2021-05-06T15:24:30
|
|
filter: introduce git_filter_list_apply_to_buffer
Provide a filter application mechanism that takes a user-provided string
and length, instead of a `git_buf`.
|
|
26846f4c
|
2021-05-06T15:19:58
|
|
filter: remove git_buf sharing in `git_filter_list_apply_to_data`
The API `git_filter_list_apply_to_data` shares data between its out and
in parameters to avoid unnecessarily copying it when there are no
filters to apply. However, it does so in a manner that is potentially
confusing, leaving both `git_buf`s populated with data. This is risky
for end-users who have to know how to deal with this. Instead, we
remove this optimization - users who want to avoid unnecessary copies
can use the longstanding streaming API or check the filter status before
invoking the filters.
|
|
9869f1e5
|
2021-05-06T02:19:49
|
|
filter: deprecate git_filter_list_stream_data
`git_filter_list_stream_data` takes user input in a `git_buf`.
`git_buf` should only be used when libgit2 itself needs to allocate data
and returned to a user that they can free when they wish. Replace it
with `git_filter_list_stream_buffer` that takes a data buffer and
length.
|
|
4bd17208
|
2021-05-06T02:28:14
|
|
Merge pull request #5858 from libgit2/ethomson/memleaks
test: clean up memory leaks
|
|
b31795ef
|
2021-05-06T01:46:19
|
|
test: clean up memory leaks
|
|
69c65b9a
|
2019-12-31T12:38:03
|
|
path: bump most Win32 unicode buffer sizes from MAX_PATH to GIT_PATH_MAX
|
|
c44f5688
|
2021-05-02T10:27:56
|
|
Merge pull request #5823 from libgit2/ethomson/path_validation
Working directory path validation
|
|
ed94f549
|
2021-05-01T20:26:49
|
|
diff:add option to ignore blank line changes
|
|
c15ed350
|
2021-04-26T12:23:25
|
|
repo: validate repository paths
Ensure that a repository's path (at initialization or open time) is
valid. On Windows systems, this means that the longest known path
beneath the repository will fit within MAX_PATH: this is a lock file for
a loose object within the repository itself.
Other paths, like a very long loose reference, may fail to be opened
after the repository is opened. These variable length paths will be
checked when they are accessed themselves. This new functionality is
done at open to prevent needlessly checking every file in the gitdir
(eg, `MERGE_HEAD`) for its length when we could instead check once at
repository open time.
|
|
3589587d
|
2021-04-24T13:11:25
|
|
repo: factor the commondir detection
|
|
e5851c62
|
2021-04-24T09:44:45
|
|
refs: ensure loose refs adhere to path validation
On Windows, we need to enforce MAX_PATH for loose references and their
reflogs. Ensure that any path - including the lock file - would fit
within the 260 character maximum.
We do not honor core.longpaths for loose reference files or reflogs.
core.longpaths only applies to paths in the working directory.
|
|
1016ad4f
|
2021-04-15T01:31:12
|
|
path: don't join paths in git_path_find_dir
Let `git_path_find_dir` simply take a `git_buf` that contains a
directory or a file, instead of trying to both join a path AND then deal
with prettifying it or its basename. This allows consumers to join
paths themselves (and apply any necessary rules - like fitting within
MAX_PATH).
|
|
717df1a4
|
2021-04-15T01:23:07
|
|
worktree: validate worktree paths
Worktree paths need to fix within MAX_PATH always, regardless of
`core.longpaths` setting.
|
|
f3bcadd2
|
2021-04-05T10:57:49
|
|
submodule: validate path lengths
Validate that working directory paths honor `core.longpaths` where
appropriate. Paths to the submodule gitdirs must always honor the
operating system length restrictions; `core.longpaths` does not affect
gitdir paths.
|
|
b457fe27
|
2021-04-04T22:18:55
|
|
iterator: validate workdir paths
Supply the repository for the filesystem and workdir iterators - for
workdir iterators, this is non-null and we can lookup the core.longpaths
configuration option. (For regular filesystem iterators, this is NULL,
so core.longpaths does not apply.)
|