static typing

[jorenham]

If there's any interest in it, I could write some stubs, or inline annotations. Or in case there won't be any new releases, the stubs could also be distributed as a stand-alone stubs-only package, or as part of https://github.com/numpy/numtype.

Either way, I don't mean to flex or anything, but once everything is set up, annotating this project would probably take me less than an hour to annotate, so let me know if that would be of any help.

[jorenham]

I guess it still looks pretty active: https://pypistats.org/packages/numpy-financial

[Kai-Striega]

I'd be keen on this, however I don't have much time to review it at the moment

[jorenham]

How attached to startup time are you? Importing typing can already add ~25ms, depending on the Python version that's used. I'll also need to use @typing.overload quite a bit, which isn't free.

If that's a problem, then I'll go for .pyi stubs (2 to be exact), which have 0 runtime impact. The disadvantage of stubs are the additional maintenance burden, and the lack of support for runtime type-checkers like beartype. But on the other hand, stubs are a bit more flexible than inline annotations, especially in case of function @overloading, which I expect to use for almost every function in _financial.py.

[Kai-Striega]

As far as I know, of the code I've seen that use numpy-financial, it isn't used in performance critical code. So I'm not too attached to an additional ~25ms and would much rather optimize for maintainability.