Hey -- um, anybody know what the heck this baby does? I'm actually just concerned with it in the context of:
I'm using it with the commons-id framework and wondering how likely it is to get two identical random numbers if they're called close enough together, or if Sun's implementation of this algorithm magically does something to make it "unlikely."
secure random is a pretty solid implementation of a PRNG. it's not based at all on system clock, so when you try to generate the random numbers should have nothing to do with the RNs you get. also, from the name it looks like it hashes the entropy value, so even if the two values were time-based (and close together) the hash would change that.