Allow PEP 646 variadic generics syntax?

[burnpanck]

PEP 646 Variadic Generics have landed in python 3.11, which allow generics to be written that accept a variable number of arguments, and the prime use-case is for numpy-like arrays. As far as I can tell however, neither numpy itself, nor the python standard library ship actual typings making use of that. Given that nptyping already supports that use-case just through the use of strings instead of variadic generics, it may be reasonable to extend to the new syntax. Is that something which aligns with the library's vision?

[ramonhagenaars]

Hi Burnpanck,

Variadic generics are indeed on this library's radar. I would use it to support more generic types, for example indices in DataFrames. This is how it could look like:

from nptyping import Structure as S, Index  # note: Index does not exist yet

DataFrame[S["a: Int, b: Float"]]  # hint a structured DataFrame
DataFrame[Index["a"]]  # only hint an index
DataFrame[S["a: Int, b: Float"], Index["a"]]  # hint a structure and an index

However, mypy support is still in progress (see python/mypy#12280) and nptyping wants to stay mypy-compliant. Once mypy has implemented support, nptyping will follow.