The discrepancy in strides between numpy and dpnp arrays

[chudur-budur]

I think the stride computation in dpnp arrays is not correct.

>>> import numpy as np
>>> import dpnp
>>> a = np.arange(0, 10, dtype=dpnp.int64)
>>> a.strides
(8,)
>>> b = dpnp.arange(0, 10, dtype=dpnp.int64)
>>> b.strides
(1,)
>>> c = np.arange(0, 10, dtype=dpnp.int32)
>>> c.strides
(4,)
>>> d = dpnp.arange(0, 10, dtype=dpnp.int32)
>>> d.strides
(1,)
>>> p = a[::2]
>>> p.strides
(16,)
>>> q = b[::2]
>>> q.strides
(2,)
>>> r = c[::2]
>>> r.strides
(8,)
>>> s = d[::2]
>>> s.strides
(2,)

Why is that?