* cksum: fix parsing error with tagged cheksum files
When passed the '-c'/'--check' flag, and parsing a checksum file in
the "tagged" format, cksum (symlinked to sha256sum, etc...) expects a
line that looks like this:
ShA256 (file.bin) = da39a3ee5e6b4b0d3255bfef95601890afd80709
If the hash algorithm at the beginning of the line (in the above case
SHA256) is missing, then cksum panics because it is attempts to use the
value of an array at index -1. This fix causes cksum to instead consider
the line a syntax error and ignore it, just as GNU cksum does. I also
added unit and integration tests to check for the above behaviour.
---------
Co-authored-by: Sylvestre Ledru <sylvestre@debian.org>
* tail: fix panic on write error in bounded_tail
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* tail: increase test file size to ensure bounded_tail path
---------
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* refactor uu_ls so that crate users can call the command without having
the internal print stdout output
* redo changes
* clippy
* help cspell understand dired
* rebase
* try and optimize for memory and speed performance.
update output handling to add streaming mode.
* work on ci issues, optimize uid and gid cache handling, improve dired output calculations
* test fx: change read_dir parameter to mutable reference in collect_directory_entries and enter_directory functions
* update documentation
* more perf tweaks
* fmt
* tweak perf
* Add back (and shorten) comment for 31_556_952
* Update src/uu/ls/src/ls.rs
Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com>
* Update src/uu/ls/src/ls.rs
Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com>
---------
Co-authored-by: Sylvestre Ledru <sylvestre@debian.org>
Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com>
The test_env_arg_ignore_signal_valid_signals test was failing on macOS
CI because 100ms was not always enough for the process to terminate
after receiving a fatal signal. Increase the delay to 500ms on macOS
only, keeping 100ms on other platforms.
Fixes flaky failure seen in: https://github.com/uutils/coreutils/actions/runs/24442074142/job/71409227191
The `+1` accounting for the ACL/context indicator was applied once per
item inside the padding loop, so N files with ACLs inflated the
link-count column width by N. Apply it a single time after the loop.
Fixes#10980
test_single_quote_pair_as_delimiter_is_invalid is intermittent on
i686: the parent writes "a''b\n" to cut's stdin while cut validates
args and exits immediately with "delimiter must be a single
character". On a fast child exit, the kernel closes the pipe
read-end before the parent's write() runs, so write() returns
EPIPE and the pipe_in thread panics.
The race is inherent: draining stdin in cut before exiting on an
arg error would hang on terminals and read unbounded data from
long pipes, contradicting GNU behavior. Use the test framework's
ignore_stdin_write_error() flag, which exists for exactly this
"parent writes to child that exits fast" pattern.
Upstream commit 9f50c8b42 ("tee: fix input with sleep") already
contains the functional fix for short reads from a paused writer.
Keep the regression coverage so we do not reintroduce the bug.
The test writes one small chunk, waits long enough for a buggy tee to
exit, then writes a second chunk and asserts both stdout and the output
file receive both writes.
* Fixed panic in who > /dev/full
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* More sctrictly test
---------
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Josh French
can1357
Ibrahim Burak Yorulmaz
oech3
Devel08
Aleksandar Janicijevic
Lorenzo Rossi
Darren Schroeder
Sylvestre Ledru
xtqqczze
Eyüp Can Akman
Daniel Hofstetter
Kevin Burke
Ivan Shanygin