On 26/12/2025 20:29, Philip Rowlands wrote:
Is there any appetite for coreutils to be shellcheck-clean?
For example, in release 9.9 tests/ there are 28 errors, 700 warnings and 2479
notes. Some of these errors are clearly false positive, and some due to
incorrect syntax under test, but these obscure what may be legitimate bugs or
shell compatibility issues.
It might be worth splitting these into milestones of "errors first", and there
are diminishing returns as syntax patterns are fixed.
I'm not suggesting that shellcheck should be a hard block on releases, but I
find it a useful tool, and would be happy to contribute patches as long as
maintainers were generally in favour.
It's a good point,
But I find shellcheck is very picky.
I reviewed all the "errors" and they were all false positives.
So TBH I'd avoid churning the tests to be shellcheck clean,
but I wouldn't be against tweaking them to be "error" clean.
cheers,
Padraig
