On Wed, Aug 17, 2011 at 9:23 AM, John P. Hartmann <[email protected]> wrote:
> I suggest you look elsewhere for pseudo random numbers, or at least > verify the randomness using statistical methods. Lots have been > learnt about random numbers since the sixties. Think that latest models have some new instructions in the cryptic processors as well. Might be elegant if we could pick those up... But I am not writing SSL in Pipes, so my random requirements always have been pretty modest (like generate a new password or generate test data for some plumbing). PS I have developed a particular taste for random numbers, in that they fit a poisson distribution. But under the coverse that runs on the random stage as well.
