Commit message (Collapse) | Author | Age | Files | Lines | ||
---|---|---|---|---|---|---|
... | ||||||
* | dir: New flag to control whiteout visibility | Tavian Barnes | 2023-10-17 | 2 | -3/+13 | |
| | ||||||
* | dir: Fix FreeBSD union mounts | Tavian Barnes | 2023-10-17 | 5 | -8/+74 | |
| | ||||||
* | dir: Add a flags parameter to bfs_opendir() | Tavian Barnes | 2023-10-17 | 6 | -21/+45 | |
| | ||||||
* | tests: Move closed_std* tests out of the POSIX group | Tavian Barnes | 2023-10-14 | 6 | -2/+8 | |
| | | | | | | | | | | | | | POSIX actually says > If the utility would be executed with file descriptor 0, 1, or 2 > closed, implementations may execute the utility with the file > descriptor open to an unspecified file. So we're not guaranteed to be able to detect the situation in the first place. Add a best-effort check for these platforms and skip the test. Link: https://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_09_01_01 | |||||
* | dir: Fix Android build | Tavian Barnes | 2023-10-14 | 1 | -1/+1 | |
| | | | | Android doesn't provide a getdents() wrapper, so use the syscall. | |||||
* | docs: Add a list of related utilities | Tavian Barnes | 2023-10-13 | 1 | -0/+42 | |
| | ||||||
* | tests: Always create the scratch directory | Tavian Barnes | 2023-10-13 | 1 | -0/+2 | |
| | | | | | | | | | | | | | | | | Otherwise, some tests that try to check for flag support like invoke_bfs scratch -quit -xattr || skip can be accidentally skipped if ./scratch doesn't exist yet: $ ./tests/tests.sh bsd/xattr tests skipped: 1 After this patch, we get the expected $ ./tests/tests.sh bsd/xattr tests passed: 1 | |||||
* | Merge tag '3.0.4' | Tavian Barnes | 2023-10-12 | 5 | -4/+16 | |
|\ | ||||||
| * | Release 3.0.43.0.4 | Tavian Barnes | 2023-10-12 | 3 | -2/+14 | |
| | | ||||||
| * | tests: Don't crash if unbuffer isn't installed | Tavian Barnes | 2023-10-12 | 1 | -1/+1 | |
| | | ||||||
| * | bfstd: Actually use the copied locale | Tavian Barnes | 2023-10-12 | 1 | -1/+1 | |
| | | | | | | | | This fixes a segfault in xstrerror() when using musl. | |||||
* | | build: Re-enable liburing | Tavian Barnes | 2023-10-12 | 1 | -1/+1 | |
|/ | | | | This reverts commit 735a2979a37251dc447274aa888ad77ded126457. | |||||
* | Release 3.0.33.0.3 | Tavian Barnes | 2023-10-12 | 3 | -13/+5 | |
| | ||||||
* | build: Disable liburing for now | Tavian Barnes | 2023-10-12 | 1 | -1/+1 | |
| | | | | liburing will be enabled for the next feature release. | |||||
* | docs: Start writing the 3.1 changelog | Tavian Barnes | 2023-10-12 | 1 | -0/+33 | |
| | ||||||
* | bench: Look for ./bfs if ./bin/bfs doesn't exist | Tavian Barnes | 2023-10-12 | 1 | -1/+5 | |
| | ||||||
* | bench: Check for bfs -j support before using it | Tavian Barnes | 2023-10-12 | 1 | -3/+13 | |
| | ||||||
* | bftw: Make sure we don't close a directory while we unwrap it | Tavian Barnes | 2023-10-12 | 1 | -2/+6 | |
| | | | | | | | bftw_cache_reserve() can lead to bftw_cache_pop(), which could close the directory we're trying to unwrap! If that happened, we would try dup_cloexec(-1), which would fail with EBADF, so there was no observable bug. But it's better to avoid the whole situation. | |||||
* | bftw: Fix to_close list corruption with !BFS_USE_UNWRAPDIR | Tavian Barnes | 2023-10-12 | 1 | -6/+13 | |
| | | | | | | | | | | It's possible for pincount to drop to zero, then get incremented and drop back to zero again. If this happens, we shouldn't add it to the to_close list twice. This should fix the intermittent hang on the macOS CI. Fixes: 815798e1eea7fc8dacd5acab40202ec4d251d517 | |||||
* | list: Assert that we're not inserting already-attached nodes | Tavian Barnes | 2023-10-12 | 2 | -21/+46 | |
| | ||||||
* | diag: Move enum debug_flags out of ctx.h | Tavian Barnes | 2023-10-12 | 5 | -53/+55 | |
| | ||||||
* | bftw: Don't force buffering for parallel dfs | Tavian Barnes | 2023-10-12 | 1 | -5/+30 | |
| | ||||||
* | bftw: Fix unbuffered depth-first searches | Tavian Barnes | 2023-10-12 | 1 | -15/+41 | |
| | | | | | | | | | | | | | | | bftw() implements depth-first search by appending files to a batch, then prepending the batch to the queue. When we switched to separate file/ directory queues, this was only implemented for the file queue. Unbuffered searches don't use the file queue, so they were all breadth- first in practice. This meant that iterative deepening (-S ids) was actually "iterative deepening *breadth*-first search," a horrible strategy with no advantage over regular breadth-first search. Now it performs iterative deepening *depth*-first search, which at least limits its memory consumption. Fixes: c1b16b49988ecff17ae30978ea14798d95b80018 | |||||
* | bftw: Let iterative deepening work depth-first when sorting | Tavian Barnes | 2023-10-12 | 1 | -1/+1 | |
| | ||||||
* | bench: Default to --strategies=rust | Tavian Barnes | 2023-10-10 | 1 | -1/+1 | |
| | ||||||
* | bench: New --jobs benchmark | Tavian Barnes | 2023-10-10 | 1 | -4/+64 | |
| | ||||||
* | bench: Group by strategy in comparison mode | Tavian Barnes | 2023-10-10 | 1 | -4/+13 | |
| | ||||||
* | bench: Don't print paths in the --strategies benchmark | Tavian Barnes | 2023-10-10 | 1 | -1/+1 | |
| | ||||||
* | bench: Allow format specifiers in (sub)group headers | Tavian Barnes | 2023-10-10 | 1 | -10/+10 | |
| | ||||||
* | bar: Replace volatile sig_atomic_t with actual atomics | Tavian Barnes | 2023-10-06 | 1 | -8/+12 | |
| | ||||||
* | alloc: Test allocation size overflows | Tavian Barnes | 2023-10-06 | 2 | -4/+12 | |
| | ||||||
* | tests/bfs/status: Try to test SIGWINCH | Tavian Barnes | 2023-10-06 | 1 | -1/+1 | |
| | ||||||
* | tests: Check for expect_unbuffer too | Tavian Barnes | 2023-10-06 | 1 | -2/+8 | |
| | | | | That's where FreeBSD installs it. | |||||
* | tests: New bfs_pty wrapper for unbuffer | Tavian Barnes | 2023-10-06 | 4 | -7/+17 | |
| | ||||||
* | ci: Install expect for unbuffer | Tavian Barnes | 2023-10-05 | 2 | -0/+6 | |
| | ||||||
* | tests/bfs: Add a test that runs -status | Tavian Barnes | 2023-10-05 | 1 | -0/+3 | |
| | ||||||
* | tests/bfs: Add tests for color autodetection | Tavian Barnes | 2023-10-05 | 4 | -0/+65 | |
| | ||||||
* | tests: Respect $NO_COLOR | Tavian Barnes | 2023-10-05 | 1 | -9/+15 | |
| | ||||||
* | README: Realign badges | Tavian Barnes | 2023-10-05 | 1 | -4/+3 | |
| | | | | A recent GitHub change broke the nice alignment of the badges to the top heading, so just put the badges underneath it for now. | |||||
* | bfstd: Add a thread-safe wrapper for strerror() | Tavian Barnes | 2023-10-05 | 11 | -13/+69 | |
| | ||||||
* | thread: Define thread_local | Tavian Barnes | 2023-10-05 | 2 | -0/+16 | |
| | ||||||
* | config: Fold !__FreeBSD__ into BFS_USE_SYS_CAPABILITY_H | Tavian Barnes | 2023-10-05 | 2 | -2/+2 | |
| | ||||||
* | dstring: Limit the special dchar typedef to lint builds | Tavian Barnes | 2023-10-04 | 1 | -7/+2 | |
| | | | | | Clang still thinks that alignof(dstr[1]) == 2, so out of an abundance of caution, don't mess with dchar alignment in normal builds. | |||||
* | build: New lint flag | Tavian Barnes | 2023-10-04 | 1 | -1/+8 | |
| | ||||||
* | thread: s/call_once/invoke_once/ | Tavian Barnes | 2023-10-03 | 3 | -3/+3 | |
| | | | | call_once() is a reserved identifier from C11. | |||||
* | ci/freebsd: Use tailscale ssh | Tavian Barnes | 2023-10-03 | 1 | -10/+3 | |
| | ||||||
* | ci/macos: Don't install coreutils | Tavian Barnes | 2023-10-03 | 1 | -4/+0 | |
| | | | | | Since we build our own touch(1) implementation, we no longer need to work around the macOS bug. | |||||
* | tests: Fix make_deep() on FreeBSD | Tavian Barnes | 2023-10-02 | 1 | -9/+10 | |
| | | | | I keep forgetting that PATH_MAX is only 1024 there. | |||||
* | Merge branch 'io-uring' | Tavian Barnes | 2023-10-02 | 8 | -53/+310 | |
|\ | ||||||
| * | ioq: Use io_uring | Tavian Barnes | 2023-10-02 | 2 | -42/+288 | |
| | | | | | | | | Closes #65. |