cp: error on -s with relative source and destination outside cwd

[henrimarconni]

Removes one of the inconsistencies listed in #15254

cp -s b.txt sub created sub/b.txt -> b.txt, a dangling self-referential symlink, and exited 0. GNU refuses with:

cp: sub/b.txt: can make relative symbolic links only in current directory

and exits 1. A relative symlink target is resolved from the link's own directory, so it can't point back to ./b.txt from inside sub/.

Changes

• In handle_copy_mode (CopyMode::SymLink), return an error when the source is not absolute and the destination's parent is not the current directory.
• Add the cp-error-relative-symlink-not-in-cwd message to en-US.ftl and fr-FR.ftl.
• Add tests for the failing case and for the cases that must still succeed (relative source with destination in cwd, absolute source into a subdirectory).

Behavior

Before:

exit: 0; files: sub/b.txt l mode 777

After:

exit: 1; cp: sub/b.txt: can make relative symbolic links only in current directory

[henrimarconni]

Thanks for the review :)
I have modified the code as you said:

1. Moved the check to the start of copy_file function to prevent file destruction bugs when using --remove-destination flag
2. Added more test cases
3. Fixed WASI CI/CD failure (tests that pass an absolute host path are ignored on the WASI runner, since the sandbox cannot see the host tmpdir)

[henrimarconni]

Thanks for the review, all addressed:

• Added cp -rs dir out with a relative dir (refused) and the
  absolute-source counterpart (still works).
• ../<cwd>/b2: good catch, that was a bug. GNU compares
  dir_name(dest) with "." by name and then by device/inode, so it
  accepts this. Fixed to match, with tests for it and for a symlink
  to the cwd.
• Merged the "./" test into test_cp_symlink_relative_source_in_cwd,
  which now loops over "b2.txt" and "./b2.txt".

[codspeed]

Merging this PR will improve performance by 3.56%

⚠️ 118 benchmarks spent significant time in system calls

System calls cannot be consistently instrumented, so they are not included in the measure, which understates the real cost. Please switch to the Walltime instrument to accurately measure system calls.

Measurement and system calls

• sort_reverse_utf8_locale
• sort_unique_utf8_locale
• sort_mixed_utf8_locale
• cksum_multiple_files
• dd_copy_partial
• du_all_wide_tree[(5000, 500)]
• cksum_raw_output
• cksum_sysv
• cksum_default
• cksum_crc
• …and 108 more

ℹ️ 1 benchmark has no flame graph

The flamegraph could not be generated because too much time was spent in runtime internals, making the visualization not useful for performance analysis.

• sort_dictionary_order[500000]

⚡ 1 improved benchmark
✅ 400 untouched benchmarks
⏩ 70 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡ ⚠️	Simulation	sort_reverse_utf8_locale	82 ms	79.2 ms	+3.56%

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing henrimarconni:main (350371a) with main (80fd308)²}

Footnotes

1. 70 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

2. No successful run was found on main (9c28184) during the generation of this report, so 80fd308 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩