ls: Add NotADirectory message

[danielzgtg]

This makes the error message for ls ~/.bashrc/test less scary. Two people had already complained about the phrasing "unknown io error".

Before

Some say the error message is scarier than the GNU coreutils one.

home@daniel-desktop3:~$ /bin/ls .bashrc/test
/bin/ls: unknown io error: '.bashrc/test', 'Os { code: 20, kind: NotADirectory, message: "Not a directory" }'
home@daniel-desktop3:~$ /bin/gnuls .bashrc/test
/bin/gnuls: cannot access '.bashrc/test': Not a directory

After

I got rid of the word "unknown". I changed the phrasing, because Peter Cordes said "it hasn't checked which path component isn't a directory; it's not the final one".

$ /usr/bin/cargo run --color=always --bin ls --profile dev --manifest-path /home/home/CLionProjects/uutils/src/uu/ls/Cargo.toml -- /home/home/.bashrc/test
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.11s
     Running `target/debug/ls /home/home/.bashrc/test`
ls: '/home/home/.bashrc/test': A path component was not a directory

Process finished with exit code 2

[xtqqczze]

We also need a French translation in src/uu/ls/locales/fr-FR.ftl

[sylvestre]

i can take care of the french translation
but could you please add a test to verify this error message, thanks

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/tail/retry. tests/tail/retry is passing on 'main'. Maybe you have to rebase?
Skipping an intermittent issue tests/tail/symlink (passes in this run but fails in the 'main' branch)
Note: The gnu test tests/cp/sparse-extents is now being skipped but was previously passing.

[danielzgtg]

but could you please add a test to verify this error message, thanks

Added:

#[test]
#[cfg(unix)]
fn test_ls_not_a_directory_errors() {
    let scene = TestScenario::new(util_name!());
    scene
        .ucmd()
        .arg("/etc/hosts/test")
        .fails()
        .code_is(2)
        .stderr_contains("not a directory");
}

[sylvestre]

dupe of https://github.com/uutils/coreutils/pull/11333/changes ?

[danielzgtg]

dupe of https://github.com/uutils/coreutils/pull/11333/changes ?

Same bug. However, the advantages of my PR are:

• The error message is not misleading
• My test uses /etc/hosts, which is compatible with macOS and Android. That PR uses /sbin/init, which only works on desktop Unix/Linux

However, I don't care if my PR gets closed and the other gets merged because I'm used to the 1st point, and I don't program on the platforms in the 2nd point.

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/date/resolution (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/pr/bounded-memory (passes in this run but fails in the 'main' branch)
Congrats! The gnu test tests/tail/retry is no longer failing!
Congrats! The gnu test tests/expand/bounded-memory is now passing!

[oech3]

Better message is one of at aim of this project. But, "A path component was" would add sed hack for GnuTests in the future. Do you consider suggesting the change to GNU too?

[oech3]

The message is coming from OS. Different might cause other issue at different command.

[danielzgtg]

Closing as I don't know how complex is "sed hack for GnuTests in the future."

[sylvestre]

Closing as I don't know how complex is "sed hack for GnuTests in the future."

it is trivial.
see: https://github.com/Ecordonnier/coreutils/blob/d6c793c09abb43078ef6ecff6b8f49eccb0cc015/util/build-gnu.sh#L342