We overlooked one thing. Since hashing is defined in terms of
tobytes(), the equality-hash invariant is now broken:

>>> from _testbuffer import ndarray
>>> x = ndarray([1,2,3], shape=[3], format='f')
>>> y = ndarray([1,2,3], shape=[3], format='B')
>>> a = memoryview(x)
>>> b = memoryview(y)
>>> a == b
True
>>> hash(a) == hash(b)
False
>>> 


This is one problem with the new equality definition. It puts
"memoryview" firmly into array territory. I'm not saying that's
wrong, I even think it was the intention of the PEP authors to
have a zero copy "arrayview".

Both array.array and numpy.array sidestep the issue by not being
hashable.


I don't really see a way around all this except doing slow
element-wise hashing.