> 
> I'm not entirely sure what the right way to do this is, though.

The code that itojun has just dumped into several different source
files does attempt to do this (but I think can be broken with a small
number of values from the sequence).

OTOH what is actually needed is some generic library code that can
generate different types of 'random' numbers depending on what it
is asked to do. Eg
- repeateable sequences with reasonable statistical randomness
- sequences where values are guaranteed not to be reproduced
- random values

The actual algorithms being absent from the ABI, they might even
be kernel options.

	David

-- 
David Laight: david@l8s.co.uk