Closed Beronx86 closed 7 years ago
That will introduce a copy which is bad for performance.
I guess you could do data = [a if a.flags.aligned else numpy.require(a, requirements='A') for a in data]
?
Or you could use copy=False which avoids the problem entirely.
@bartvm , I checked, it works.
After transformation by start_server and DataServerStream, the numpy.ndarray may become unaligned.