touch: Respect -h when getting metadata #5951
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
It would be possible to get them to run on FreeBSD by avoiding get_symlink_times, but the behavior we're testing is not platform-specific, so it's fine to not test it on FreeBSD.
ysthakur
force-pushed
the
fix-touch-stat-follow
branch
2 times, most recently
from
1adbe1e
to
2a11b2b
Compare
|
@tertsdiepraam I don't know why some of the workflows are failing, but the failures don't seem to be related to |
|
We'll see what this run does :) We do get unrelated errors quite often unfortunately. Recently especially on windows. But don't worry, we'll still look at the PR if there are some failures. |
|
Welp, still failing, but at least the failures on that run were only on Windows. |
|
Windows has been having problems recently. No need to worry about it (for you) 😄 |
|
GNU testsuite comparison: |
|
GNU testsuite comparison: |
|
Yay! Can we merge now? |
zhitkoff
added a commit
to zhitkoff/coreutils
that referenced
this pull request
Mar 12, 2024
* ci: use codecov token in CICD/GnuTests workflows * tee: fail test if string setup fails * sort: add skipped test for combined flags Now that clap#2624 has been resolved, we can and should test both variants. * cat: don't flake even on exotic pipe buffer sizes See also 9995c63. There is a race condition between the writing thread and the command. It is easily possible that on the developer's machine, the writing thread is always faster, filling the kernel's buffer of the stdin pipe, thus succeeding the write. It is also easily possible that on the busy CI machines, the child command runs first for whatever reason, and exits early, thus killing the pipe, which causes the later write to fail. This results in a flaky test. Let's prevent flaky tests. * numfmt: don't flake even on exotic pipe buffer sizes * split: don't flake even on exotic pipe buffer sizes * simulate terminal utility (squash) * workaround: run builds with retry (a) * added configurable terminal size * chore(deps): update rust crate rayon to 1.9 * cargo: fix feature = "cargo-clippy" deprecation * tests/printf: Fix char_as_byte test, add char and string padding tests * printf: Change get_char and write_padded to handle bytes instead of chars * uucore/format: add padlen to spell-checker:ignore * tests/printf: Verify the correct error behavior of printf when provided with '%0c' or '%0s' * printf: Raise error on '%0c' and '%0s' formats * cp: fix flaky test test_cp_arg_interactive_update, document adjacent bug * chore(deps): update rust crate walkdir to 2.5 * cat: permit repeating command-line flags * cat: fix -b and -n anti-symmetry * cat: ignore -u flag, just like GNU does * tests/common/util.rs: add cfg(feature = "env") * cat: prefix two test fns with "test_" * Bump mio from 0.8.10 to 0.8.11 * extend error message for case when writer instanciation fails second time * Bump chrono from 0.4.34 to 0.4.35 * ls: use chrono::TimeDelta::try_seconds instead of deprecated chrono::TimeDelta::seconds * touch: replace use of deprecated chrono functions * chmod: slightly adjust error message when preserve-root is triggered One of the GNU tests checks for the exact error message. * chgrp+chown: also trigger preserve-root during dirwalking, fix error message This is explicitly tested in the GNU tests. * uucore: drop unused function resolve_relative_path This function is by necessity ill-defined: Depending on the context, '..' is either the logical parent directory, sometimes the physical parent directory. This function can only work for the latter case, in which case `Path::canonicalize` is often a better approach. * split: close as much fds as needed for opening new one * use std::command::pre_exec() to set limits on child before exec * chore(deps): update softprops/action-gh-release action to v2 * dd: treat arg as bytes if it contains 'B' * Fix clippy warnings * tr: stream output instead of buffering This should lower memory consumption, and fixes OOM in some scenarios. * shuf: fix and test off-by-one errors around ranges * shuf: fix error message text on negative-sized ranges Found by @cakebaker: uutils#6011 (comment) * chcon: allow overriding between --dereference and --no-dereference * chcon: allow repeated flags and arguments * touch: Respect -h when getting metadata (uutils#5951) * Add tests that stat symlinks * Check follow first in stat * Don't run tests on FreeBSD It would be possible to get them to run on FreeBSD by avoiding get_symlink_times, but the behavior we're testing is not platform-specific, so it's fine to not test it on FreeBSD. --------- Co-authored-by: Sylvestre Ledru <sylvestre@debian.org> * pr: fix deprecation warnings & remove comment * chgrp: fix clippy warning * cut: allow non utf8 characters for delimiters (uutils#6037) * cp: improve the support of --attributes-only (uutils#6051) * cp: improve the support of --attributes-only * remove useless comments Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com> --------- Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com> * cp: Split the copy_file function a bit * parser: if closing square bracket not found, stop looking for it again This solves uutils#5584, where the fuzzing would take hours without this. * Fix install: invalid link at destination also remove some FixMEs for FreeBsd * Bump nix from 0.27 to 0.28 * uucore/pipes: adapt to new return type of nix fn nix 0.28 changed the return type of unistd::pipe() from Result<(RawFd, RawFd), Error> to Result<(OwnedFd, OwnedFd), Error> * tty: unistd::ttyname takes AsFd instead of RawFd change introduced by nix 0.28 * stty: remove ofill output flag flag was removed from nix::sys::termios::OutputFlags in nix 0.28 * cat: adapt to type change of unistd::write() nix 0.28 changed "write(fd: RawFd, buf: &[u8]) -> Result<usize>" to "write<Fd: AsFd>(fd: Fd, buf: &[u8]) -> Result<usize>" * chore(deps): update rust crate blake3 to 1.5.1 --------- Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com> Co-authored-by: Ben Wiederhake <BenWiederhake.GitHub@gmx.de> Co-authored-by: Ulrich Hornung <hornunguli@gmx.de> Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> Co-authored-by: Sylvestre Ledru <sylvestre@debian.org> Co-authored-by: Dimitris Apostolou <dimitris.apostolou@icloud.com> Co-authored-by: Dorian Péron <dorianperon.i@gmail.com> Co-authored-by: Terts Diepraam <terts.diepraam@gmail.com> Co-authored-by: mhead <mtrxhead@protonmail.com> Co-authored-by: Yash Thakur <45539777+ysthakur@users.noreply.github.com> Co-authored-by: Zoltan Kiss <121870572+cj-zoltan-kiss@users.noreply.github.com>
zhitkoff
added a commit
to zhitkoff/coreutils
that referenced
this pull request
Mar 12, 2024
* ci: use codecov token in CICD/GnuTests workflows * tee: fail test if string setup fails * sort: add skipped test for combined flags Now that clap#2624 has been resolved, we can and should test both variants. * cat: don't flake even on exotic pipe buffer sizes See also 9995c63. There is a race condition between the writing thread and the command. It is easily possible that on the developer's machine, the writing thread is always faster, filling the kernel's buffer of the stdin pipe, thus succeeding the write. It is also easily possible that on the busy CI machines, the child command runs first for whatever reason, and exits early, thus killing the pipe, which causes the later write to fail. This results in a flaky test. Let's prevent flaky tests. * numfmt: don't flake even on exotic pipe buffer sizes * split: don't flake even on exotic pipe buffer sizes * simulate terminal utility (squash) * workaround: run builds with retry (a) * added configurable terminal size * chore(deps): update rust crate rayon to 1.9 * cargo: fix feature = "cargo-clippy" deprecation * tests/printf: Fix char_as_byte test, add char and string padding tests * printf: Change get_char and write_padded to handle bytes instead of chars * uucore/format: add padlen to spell-checker:ignore * tests/printf: Verify the correct error behavior of printf when provided with '%0c' or '%0s' * printf: Raise error on '%0c' and '%0s' formats * cp: fix flaky test test_cp_arg_interactive_update, document adjacent bug * chore(deps): update rust crate walkdir to 2.5 * cat: permit repeating command-line flags * cat: fix -b and -n anti-symmetry * cat: ignore -u flag, just like GNU does * tests/common/util.rs: add cfg(feature = "env") * cat: prefix two test fns with "test_" * Bump mio from 0.8.10 to 0.8.11 * extend error message for case when writer instanciation fails second time * Bump chrono from 0.4.34 to 0.4.35 * ls: use chrono::TimeDelta::try_seconds instead of deprecated chrono::TimeDelta::seconds * touch: replace use of deprecated chrono functions * chmod: slightly adjust error message when preserve-root is triggered One of the GNU tests checks for the exact error message. * chgrp+chown: also trigger preserve-root during dirwalking, fix error message This is explicitly tested in the GNU tests. * uucore: drop unused function resolve_relative_path This function is by necessity ill-defined: Depending on the context, '..' is either the logical parent directory, sometimes the physical parent directory. This function can only work for the latter case, in which case `Path::canonicalize` is often a better approach. * split: close as much fds as needed for opening new one * use std::command::pre_exec() to set limits on child before exec * chore(deps): update softprops/action-gh-release action to v2 * dd: treat arg as bytes if it contains 'B' * Fix clippy warnings * tr: stream output instead of buffering This should lower memory consumption, and fixes OOM in some scenarios. * shuf: fix and test off-by-one errors around ranges * shuf: fix error message text on negative-sized ranges Found by @cakebaker: uutils#6011 (comment) * chcon: allow overriding between --dereference and --no-dereference * chcon: allow repeated flags and arguments * touch: Respect -h when getting metadata (uutils#5951) * Add tests that stat symlinks * Check follow first in stat * Don't run tests on FreeBSD It would be possible to get them to run on FreeBSD by avoiding get_symlink_times, but the behavior we're testing is not platform-specific, so it's fine to not test it on FreeBSD. --------- Co-authored-by: Sylvestre Ledru <sylvestre@debian.org> * pr: fix deprecation warnings & remove comment * chgrp: fix clippy warning * cut: allow non utf8 characters for delimiters (uutils#6037) * cp: improve the support of --attributes-only (uutils#6051) * cp: improve the support of --attributes-only * remove useless comments Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com> --------- Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com> * cp: Split the copy_file function a bit * parser: if closing square bracket not found, stop looking for it again This solves uutils#5584, where the fuzzing would take hours without this. * Fix install: invalid link at destination also remove some FixMEs for FreeBsd * Bump nix from 0.27 to 0.28 * uucore/pipes: adapt to new return type of nix fn nix 0.28 changed the return type of unistd::pipe() from Result<(RawFd, RawFd), Error> to Result<(OwnedFd, OwnedFd), Error> * tty: unistd::ttyname takes AsFd instead of RawFd change introduced by nix 0.28 * stty: remove ofill output flag flag was removed from nix::sys::termios::OutputFlags in nix 0.28 * cat: adapt to type change of unistd::write() nix 0.28 changed "write(fd: RawFd, buf: &[u8]) -> Result<usize>" to "write<Fd: AsFd>(fd: Fd, buf: &[u8]) -> Result<usize>" * chore(deps): update rust crate blake3 to 1.5.1 --------- Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com> Co-authored-by: Ben Wiederhake <BenWiederhake.GitHub@gmx.de> Co-authored-by: Ulrich Hornung <hornunguli@gmx.de> Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> Co-authored-by: Sylvestre Ledru <sylvestre@debian.org> Co-authored-by: Dimitris Apostolou <dimitris.apostolou@icloud.com> Co-authored-by: Dorian Péron <dorianperon.i@gmail.com> Co-authored-by: Terts Diepraam <terts.diepraam@gmail.com> Co-authored-by: mhead <mtrxhead@protonmail.com> Co-authored-by: Yash Thakur <45539777+ysthakur@users.noreply.github.com> Co-authored-by: Zoltan Kiss <121870572+cj-zoltan-kiss@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently, if the
statfunction in thetouchcrate is used on a symlink, it will first try to get the metadata of the destination, even if thefollowparameter isfalse. It will not get the metadata of the symlink itself if the symlink is not broken. It should now checkfollowfirst.Minor change: the
failed to get attributes for {path}context isn't currently being added to the error if the firstfs::metadatacall fails andfollowistrue. But this PR makes it so all errors have that context added to them. However, at some point in the past, it seems like all errors had that context added to them, so I figured it'd be fine (@tertsdiepraam tracked down this commit fixing an earlier issue).The tests I added should maybe not run on FreeBSD because there's apparently an issue with nanoseconds or something? I have no idea what the difference there is, but I figured I'd set the nanoseconds field to 0 and let the CI run, just in case that works.Never mind, I added#[cfg(not(target_os = "freebsd"))]since there's not much chance of the specific behavior being tested here being different on FreeBSD.