[joppy]

Because of how floating-point numbers are distributed, you should expect a random number drawn uniformly from [0,1] to basically never hit the smallest floating-point number: hitting something in the range [0, 2^-n] should happen with probability 2^-n.

I agree there is much more one could want out of a uniform random float generator than just k/N where k is a integer drawn uniformly from {0, 1, …, N}, but hitting arbitrarily small floating point numbers is not on this list.

If you are intending to sample numbers in the range [0,1] such that each float occurs with equal probability then that’s fine, but it’s certainly not uniformly random on [0,1].

[hansvm]

Steelmanning their argument, an ideal output would consist of every possible float in [0,1] occurring with an appropriate probability -- some of those vanishingly unlikely. Kind of like how randint() should be capable of returning every possible integer rather than, say, every multiple of 64.

It's fine to counter that most of the time the difference wouldn't matter or that it might be problematic to compute for some reason or another, but I don't think it's reasonable to critique the idea on the grounds of the result not being uniform when they didn't actually ask for each float to have equal probability.