NEP-18: Calls on Dask+CuPy arrays cause "object __array__ method not producing an array" #4563
Description
Consider the following example:
import numpy as np
import cupy
import dask.array as da
x = cupy.random.random((5000, 1000))
d = da.from_array(x, chunks=(1000, 1000), asarray=False)
u, s, v = np.linalg.svd(d)This gives the following traceback:
Traceback (most recent call last):
File "svd.py", line 9, in <module>
u, s, v = np.linalg.svd(d)
File "/home/pentschev/.local/lib/python3.5/site-packages/numpy/core/overrides.py", line 152, in public_api
implementation, public_api, relevant_args, args, kwargs)
File "/home/pentschev/.local/lib/python3.5/site-packages/numpy/linalg/linalg.py", line 1598, in svd
a, wrap = _makearray(a)
File "/home/pentschev/.local/lib/python3.5/site-packages/numpy/linalg/linalg.py", line 117, in _makearray
new = asarray(a)
File "/home/pentschev/.local/lib/python3.5/site-packages/numpy/core/overrides.py", line 152, in public_api
implementation, public_api, relevant_args, args, kwargs)
File "/home/pentschev/.local/lib/python3.5/site-packages/numpy/core/numeric.py", line 555, in asarray
return array(a, dtype, copy=False, order=order)
File "/home/pentschev/.local/lib/python3.5/site-packages/dask/array/core.py", line 1002, in __array__
x = np.array(x)
ValueError: object __array__ method not producing an arrayI don't immediately see what's the cause for this issue. In this last function of the trace above, x is a CuPy array. Is that the case that we should add explicit support for __array_function__ in Dask Array?