speeding up lint_mathlib.lean

Stream: general

[Arthur Paulino]

Would it be reasonable to run the linters solely on declarations from files that were modified wrt master?

I'm (naively?) assuming that unchanged files won't have different lint outputs

[Yury G. Kudryashov]

This was discussed earlier. The simp_nf linter can break elsewhere.

[Arthur Paulino]

Is it the only linter that does so?

[Rob Lewis]

No. unused_arguments can break in untouched files if you change dependencies.