Ruby: Difference between Random#rand and Kernel#rand

Question

Is there is really difference between Random#rand and Kernel#rand?

From what i can see they use different 'C' functions.

Answer 1

They behave the same when called with a Range, but differently in several other cases.

1. When called with a negative integer -n (like -3), Random#rand raises ArgumentError, while Kernel#rand just behave as if you called it with n (by (-n).to_int.abs).
2. When called with 0, Random#rand raises ArgumentError, while Kernel#rand just behave as if you called it without any argument.
3. When called with Float n, Random#rand returns a float between 0 and n (as expected). Kernel#rand do an n.to_int.abs conversion, so for example rand(-1.9) is equivalent to rand(1), which always returns 0; rand(0.1) is equivalent to rand(0) and thus equivalent to rand.

It seems that Random#rand (and also Random::rand of course) is more robust than Kernel#rand for strange parameter. More information in documentation for Kernel#rand and Random#rand.

Answer 2

It looks like a slightly different API, but both seem to leave the actual generating to the genrand_real function.