As far as making simulations.  You can provide test operations I think, but
freenet presents difficulties that can't be addressed in any simulation.  I
think it best to assume that the current use is the ultimate simulation and
that everyone push the system as far is it will go.  The larger the node
set, the larger the complexities and the more likely the bugs.

I guess a question could be:
At what rate of growth does probability of complexity errors grow as
NUM_NODES goes to Infinity.

Has this type of problem been addressed yet?

Anyway, my solution is to keep tring to grow the node population as fast as
possible to determine if there might be chaos error which never goes away
anyway, but you must be aware of.

You need to get feed back from the other nodes for this.
I propose a tempory debug envoronment that will report errors to a central
site or some freenet key.

If complexity errors grow at log(n) where n is the number of nodes, then
problems will be managable.

However if error grows exponentially, the system is doomed in complexity
theory.

IMHO
Jay Ferguson


_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl

Reply via email to