Python 3 memoryview : unsupported format <d

[evanmason]

In moving some code from Python 2 to Python 3 I've run into a problem with memoryview. The code below works with Python 2, but not Python 3. Is there a workaround for this?

Reproducing code example:

import numpy as np
from multiprocessing import sharedctypes

x = sharedctypes.RawArray('d', 10)
memoryview(x)[:] = array(np.zeros_like(np.arange(10.), dtype=float64))

Error message:

NotImplementedError: memoryview: unsupported format <d

Numpy/Python version information:

Python 3.5.7 + numpy '1.16.3' and Python 2.7.15 + numpy '1.16.0'

[eric-wieser]

This error comes from CPython not numpy. It's raised by memoryview.__setitem__ when self.format == '<d', no matter what you assign:

>>> memoryview(x)[:] = 'forbidden'
NotImplementedError: memoryview: unsupported format <d

Since this worked in python 2, you should consider reporting it upstream