At 21:40 08/07/04 +0200, paolo veronelli wrote:
Most of my imperative pieces of software find their answers by touching around in some space of solutions and my favourite approximation algorithms use random distributions.
Is it haskell the wrong languages for those, as I'm obliged to code them inside Monads loosing the benefits of lazyness?
I don't see any reason to sacrifice the benefits of laziness. I can imagine a program that uses internally a random sequence of values. Given a suitable pseudo-random value generator, it would be quite possible to write a pure function that accepts as one its parameters a seed for the P-R generator. For many randomized algorithms, I think this should work fine, though I don't recommend this approach for cryptographic work. #g ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact