BUG: numpy.roll() does not work with unsigned shift value #27282
Description
Describe the issue:
Calling numpy.roll() will raise an unexpected exception if the shift argument is an unsigned int or a tuple with an unsigned int.
This issue occurs with numpy v2.0.1 and 2.1.0, but does not occur on numpy 1.26.4 (edit: does not with a uint64, but works with other unsigned int types).
Reproduce the code example:
import numpy as np
x = np.arange(10)
y = np.roll(x, np.uint64(1)) # raises an exceptionError message:
C:\Python312-64\Lib\site-packages\numpy\_core\numeric.py:1282: RuntimeWarning: overflow encountered in scalar negative
rolls[ax] = ((slice(None, -offset), slice(offset, None)),
C:\Python312-64\Lib\site-packages\numpy\_core\numeric.py:1283: RuntimeWarning: overflow encountered in scalar negative
(slice(-offset, None), slice(None, offset)))
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "C:\Python312-64\Lib\site-packages\numpy\_core\numeric.py", line 1265, in roll
return roll(a.ravel(), shift, 0).reshape(a.shape)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Python312-64\Lib\site-packages\numpy\_core\numeric.py", line 1288, in roll
result[res_index] = a[arr_index]
~~~~~~^^^^^^^^^^^
ValueError: could not broadcast input array from shape (10,) into shape (9,)Python and NumPy Versions:
Python: 3.12.5 (tags/v3.12.5:ff3bc82, Aug 6 2024, 20:45:27) [MSC v.1940 64 bit (AMD64)]
Numpy: 2.1.0
Runtime Environment:
No response
Context for the issue:
No response