TypeError: 'numpy._DTypeMeta' object is not subscriptable

[opoyc]

Hello!

It seems that this typing is not compatible with latest versions of numpy. in typing.py:

NDArray = Union[np.ndarray]
Float64Array = np.ndarray[Any, np.dtype[np.double]] # pragma: no cover <----------
Int64Array = np.ndarray[Any, np.dtype[np.longlong]] # pragma: no cover
Int32Array = np.ndarray[Any, np.dtype[np.intc]] # pragma: no cover

[bashtage]

Is this an error you see when typing or running code? Which version of Numpy and Python?

[opoyc]

When running the code. It was working fine until yesterday. It's used in tsfeatures. This is our current env:

numpy==1.20.0
tsfeatures==0.4.5

I've raised a PR for your revision.

[bashtage]

Your numpy is too old for the latest release. The minimum is 1.22.3 now. Which arch version are you trying to use?

[bashtage]

If you are stuck with NumPy < 1.22 then you should really pin arch to be < 7. The reason for this is that NumPy 7+ is build using NumPy 2.0.0+ for the C api. This is only backward compat back to NumPy 1.22.3.