int.denominator, int.imag: More precise return types

[AlexWaygood]

No description provided.

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉

[JelleZijlstra]

Would this fall under #7258? I can't imagine much real-world code being affected though.

[AlexWaygood]

Would this fall under #7258? I can't imagine much real-world code being affected though.

I'd argue "no", since the problems raised in that issue are mainly to do with class-level and module-level constants (#7258 (comment)). So the risk seems small imo... but yeah, the benefits are also probably small as well.

I don't have a particularly strong opinion on this — I was just looking at int because of all the renewed talk about typing & numbers recently, and noticed that fractions.Fraction has a much more precise annotation here:

typeshed/stdlib/fractions.pyi

Line 151 in 29c17ff

def imag(self) -> Literal[0]: ...

[srittau]

Please note that the moratorium is for literal constants, not for all literal values. Especially return types should be fairly uncontroversial.