tsort: add sequential access hint to applicable platforms for performance

[anastygnome]

On unix plateforms, we advise the OS of the sequential access to the graph file.
There's no reason not to do it on windows too

The change to the code is minimal touse File::Options to open the file and there's virtually no cost for this
NOTE : we now open the input file read-only explicitely, but that's arguably better.

waiting for #14384 to be merged. Already includes it.

[sylvestre]

you did much more than the title suggests, no ? :)

[anastygnome]

@sylvestre please merge #14384 it's an important security fix, I just included it here to do the tests as I don't have access to a windows machine ^^'

[oech3]

Is it able to apply this also for target_os = "cygwin"?

[anastygnome]

Is it able to apply this also for target_os = "cygwin"?

It should, it's now exposed on cygwin by rustix, and it's cfg unix, so it should work, I added it

[anastygnome]

@sylvestre ready to merge ;)

[codspeed-hq]

Merging this PR will not alter performance

✅ 367 untouched benchmarks
⏩ 50 skipped benchmarks¹

---

_{Comparing anastygnome:tsort2 (ed3682c) with main (5f9e12e)}

Footnotes

1. 50 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. ↩

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/cut/bounded-memory (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/date/date-locale-hour (passes in this run but fails in the 'main' branch)
Skipping an intermittent issue tests/pr/bounded-memory (passes in this run but fails in the 'main' branch)
Skipping an intermittent issue tests/tail/retry (passes in this run but fails in the 'main' branch)
Skipping an intermittent issue tests/tail/symlink (passes in this run but fails in the 'main' branch)

[oech3]

Too complex and most targets are out of scope of uutils. Would you list supported targets instead?

[anastygnome]

This is a not clause though. I did this to have extensibility, and not having to align with rustix whenever we'll add a target

[anastygnome]

One day rustc will support is_function_accsssible and this hack won't be needed

[oech3]

It might better to add a open wrapper doing fadvise on many unix, FILE_FLAG_SEQUENTIAL_SCAN on Windows, and no-op on others to uucore::io

[anastygnome]

It might better to add a open wrapper doing fadvise on many unix, FILE_FLAG_SEQUENTIAL_SCAN on Windows, and no-op on others to uucore::io

Cannot be done cleanly, sadly, as on winfows it must be done before opening, and on Linux afterwards, it will still be convoluted, and we still have to manage the isdir case on windows which is hard to get right. It's not that hard to read for now with a comment