Repository navigation
BUG: Numpy Typing does not recognize a deque as a valid input for np.max #24265
Description
Describe the issue:
When running mypy on the example code one gets the following message:
error: No overload variant matches argument type "deque[int]" [call-overload]
Nonetheless, the code executes correctly. Should collections' deques be a valid input for both np.min and np.max?
When trying to convert a deque to a numpy array the conversions is successful; therefore, shouldn't deques be considered as array_like?
Reproduce the code example:
import numpy as np
from collections import deque
a = deque([0])
np.max(a)Error message:
error: No overload variant matches argument type "deque[int]" [call-overload]Runtime information:
1.23.2
3.10.6 (main, May 29 2023, 11:10:38) [GCC 11.3.0]
Context for the issue:
No response
Activity
BvB93 commentedon Jul 26, 2023
The issue here stems from
dequefailing to properly implement the interface as specified by theSequenceabstract base class (which we rely on in our typing definition of array-like), the former lacking support forslice-based indexing. Loosening the__getitem__annotations down below via removal of theslice-based overload might be worth a shot here:numpy/numpy/_typing/_nested_sequence.py
Lines 65 to 72 in 7e806a1