Inconsistent arange behaviour with different int types

Given this code:

import numpy as np
np.arange(1,1,0)

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ZeroDivisionError: division by zero

Obviously step 0 is an invalid input for creating a range. So the division error is understandable, but might be more explicit. What surprised me, was the following behaviour:

np.arange(np.int64(1),1,0)
__main__:1: RuntimeWarning: invalid value encountered in long_scalars
>>> array([], dtype=int64)

The code is as wrong as before, but now instead of throwing any error, it just returns an empty list. Shouldn’t it throw exactly the same error as above? An error explicitly stating, that a range with a step of 0 can’t be specified?

[ml31415]

Now np.int32 and np.int64 produce a ValueError while int and np.int produce ZeroDivisionError. This is unfortunately still inconsistent.

[seberg]

@melissawm if we have a test that runs into the issue, no (i.e. raises the error probably). Otherwise, I cannot say without checking it closer.

That said, I suspect it is long fixed, sorry, I can’t do a full check right now.