On 07/23/10 16:39, Pádraig Brady wrote: > Note I needed to comment out the ASSERT check > when linking with the old rand-isaac. > Is that a bug in the old implementation or just a change?
It's a change to the API for rand-isaac. Combining the new test program with the old implementation would lead to undefined behavior, though it sounds like it sort of worked since the change was just to the seeding code. I changed all callers, so the API change shouldn't be a problem. Ah well, I was hoping 32-bit would be a tiny bit faster, but perhaps next time. Thanks for checking it.
