Re:Synchronous languages
by p4bl0 on Friday October 05, 2018 @12:42PM
(#18149574)
Mh, you're right.
However, there is a notion of synchronous Kahn networks, but maybe it is only used by functional synchronous and dataflow languages (e.g., Lucid Synchrone, ReactiveML). Compilation of such languages involve clock calculus (i.e. typing for scheduling) which I think makes it possible to compute the maximum size of a queue in the network (or maybe at least guarantee that the size is bounded?). This would mean that what you call rendezvous points would be statically computable (in the case of pure synchronous languages it should be easy as the maximum queue size is zero, or one depending on how we count).
Anyway, you're right that I was mistaken to think that it was an easy problem.
However, there is a notion of synchronous Kahn networks, but maybe it is only used by functional synchronous and dataflow languages (e.g., Lucid Synchrone, ReactiveML). Compilation of such languages involve clock calculus (i.e. typing for scheduling) which I think makes it possible to compute the maximum size of a queue in the network (or maybe at least guarantee that the size is bounded?). This would mean that what you call rendezvous points would be statically computable (in the case of pure synchronous languages it should be easy as the maximum queue size is zero, or one depending on how we count).
Anyway, you're right that I was mistaken to think that it was an easy problem.
EDIT: Heh: https://www.lri.fr/~mandel/papiers/MandelPlateauPouzet-MPC-1... :).