Currently memoryview.cast requires C-contiguous data when going from N-D to 1-D. As a result F-contiguous data is not supported. Please see example below:
>>> mv.c_contiguous False >>> mv.f_contiguous True >>> mv.cast("B") Traceback (most recent call last): File "<stdin>", line 1, in <module> TypeError: memoryview: casts are restricted to C-contiguous views
However as F-contiguous data is still contiguous and can be viewed in 1-D, it seems possible to support this type of data as well. PickleBuffer's raw method does appear to support both C & F contiguous data. Building off the example above:
>>> from pickle import PickleBuffer >>> pb = PickleBuffer(mv) >>> pb.raw() <memory at 0x1105cec70>
This would be useful when interacting with other libraries that work with this kind of data. That said, don't know if there is other context behind the C-contiguous constraint that I may be missing here.