It's good to have types that we can believe in. Currently, some generated points have coordinates that are not ints but numpy.int64s. This somewhat subtle difference can matter e.g. when one tries to serialize to JSON.
This PR explicitly casts numpyints to standard ints and adds a test to ensure the type purity.
LGTM. Thought about removing the comment on lines 196-7 in generation, but since the reason for the int cast might not be totally obvious to everybody, I think it's better to keep the comment...
It's good to have types that we can believe in. Currently, some generated points have coordinates that are not
int
s butnumpy.int64
s. This somewhat subtle difference can matter e.g. when one tries to serialize to JSON.This PR explicitly casts
numpy
int
s to standardint
s and adds a test to ensure the type purity.@jeppetrost @petterbejo