If this bug is reproducible as a nac3artiq/demo demo, I would put that there as well.
Actually, could you explain why the previous approach does not work? Is it solely because it doesn't take the type hint into consideration when parsing the type of the list elements?
FYI it works for
KernelInvariant[np.uint32]now but not forKernelInvariant[list[np.uint32]]. I'm still trying to figure out why. Maybe I need more time to understand how the whole thing…
numpy.array(list, copy=False)
I do agree though that this is the wrong format
That is my major concern about the documentation in its current form. Moving a lot of these back into the Rustdoc and perhaps compacting it…
anyhow
After looking at this again, I don't think this "cleanup" makes sense.
Treating all primitive-named functions as constructors breaks the fundamental assumption that a constructor creates…