Two WAM implementations of action rules

Two implementations of Action Rules are presented in the context of a WAM-like Prolog system: one follows a meta-call based approach, the other uses suspended WAM environments on the WAM heap. Both are based on a similar program transformation that clarifies the semantics of Action Rules, while using different primitives. Their implementation is compared experimentally to the TOAM-based implementation of Action Rules in B-Prolog. The suspension based approach is faster at reactivating agents on the instantiation event. The meta-call approach is easier to implement, performs overall very good and much better for synchronous events, and it is more flexible than the suspension based approaches.status: publishe