Pylint: numpy.logical_or gives false error assignment-from-no-return

[mhooreman]

Steps to reproduce

Execute pylint on the following code:

"""Demo of the assignment-from-no-return pylint false positive"""

import numpy as np


def demo():
    """Demo of the false positive, at line z = ..."""
    x = np.array([0, 1, 0, 1], dtype=np.bool)
    y = np.array([0, 0, 1, 1], dtype=np.bool)
    z = np.logical_or(x, y)
    print(np.array([x, y, z]))


if __name__ == "__main__":
    demo()

Current behavior

assignment-from-no-return(E1111): test.py:10:4: demo: Assigning result of a function call, where the function has no return

Expected behavior

No issue

pylint --version output

pylint 2.2.2
astroid 2.1.0
Python 3.6.8 |Anaconda, Inc.| (default, Dec 30 2018, 01:22:34) 
[GCC 7.3.0]

[PCManticore]

Thanks for the report!

[arekfu]

For the record, this affects other functions too, such as numpy.equal.

[sroet]

and numpy.divide

[pxl-th]

numpy.log as well

[jordyjwilliams]

Any update as to the cause of this? or what versions this didn't occur on? could be a numpy issue?

[sroet]

@jordyjwilliams It has to do with the release of numpy==1.16 pinning numpy<1.16 has fixed CI linting for me for now, but I like this to be fixed for the newer numpy

[janosh]

Does anyone know if this needs fixing on pylint's or numpy's part?

[PCManticore]

This is on pylint's side.

[jordyjwilliams]

@PCManticore any idea on when this will be fixed?

[raaaaaymond]

Can confirm same problem for a lot more Numpy functions, for example deg2rad(), arctan(), and more presumably. Moreover, array item assignment (such as arr[i] = x) suffers from unsupported-assignment-operation.

[jordyjwilliams]

+1 to all these issues

[PCManticore]

Should be fixed with the latest master of astroid.

[janosh]

@PCManticore That was quick! The commit says release date tba. Are we talking days or weeks?

[PCManticore]

Weeks most likely, I'll do a dev release in a bit, but the final one might be somewhere in February.