[Icfp04-discuss] random function and simulator dumps

[Detlef Pleiss]

just for comparing the different approaches:
at least in the beginning, with the lightning division in mind, I
concentrated on getting to results as fast as possible.
For that reason I dumped the idea of rebuilding the original random number
generator and just used the one that comes with the programming language ...
as the quality of ant brain programs should not depend on which random
number generator is used, as long as it is pseudo-random enough.
As a result of not using the original random number generator I didn't spend
any time with simulator dumps, as mine would be different anyway.
Of course that leaves the risk of having some unknown bugs in the simulator,
so I don't know for sure, but I believe it to work correctly, as the
routines themselves are based on the algorithms from the task description.
Taking the named risk, I got the simulator working in 5h + 1h debugging.

ciao - Det.