Antoine:
> However, if we want to think about a new subclassing API, it may be worth looking at the recent Numpy changes. Numpy added a new random generator API recently, with a design based on composition rather than inheritance (and also they switched from Mersenne Twister to another underlying PRNG!):
> https://numpy.org/doc/stable/reference/random/index.html

Yeah, sometimes composition is simpler. My BaseRandom base class (PR 19631) can be used with composition indirectly, since all you need is to implemented getrandbits(). Example:

---
from numpy.random import default_rng
import random

class NumpyRandom(random.BaseRandom):
    def __init__(self):
        self._rng = default_rng()

    def getrandbits(self, n):
        # FIXME: support n larger than 64 ;-)
        return int(self._rng.integers(2 ** n))

gen = NumpyRandom()
print(gen.randint(1, 6))
print(gen.random())
print(gen.randbytes(3))
---