On Fri, 2008-10-17 at 14:56 +0100, Edsko de Vries wrote:
Hi,
(Note: assumes knowledge of pi-calculus.)
I am playing with writing a simple interpreter for the pi-calculus using STM. The implementation of most of the operators of the pi- calculus is straightforward, but I am unsure on how to implement the replication operator. The interpretation of !P should be the interpretation of the infinite number of parallel processes P | P | P ... . I am implementing parallel processes using the fork operation, but spawning an infinite amount of processes -- even if the (infinite) majority of them all immediately lock -- seems like a bad idea.
So I would prefer to spawn one P, and as soon as the thread that interprets P makes any progress at all, spawn another, and so on. I'm not too sure however how to achieve this.
Typically replication is restricted to replicated input for reasons akin to this.