Lost in Translation: Forgetful Semantic Anchoring

Assigning behavioral semantics to domain-specific languages (DSLs) opens the door for the application of formal methods, yet is largely an unresolved problem. Previously proposed solutions include semantic anchoring, in which a transformation from the DSL to an external framework that can supply both behavioral semantics and apply formal methods is constructed. The drawback of this approach is that it loses the structural constraints of the original DSL along with the details of the transformation, which can lead to erroneous results when formal methods are applied. We demonstrate this problem of ¿forgetful¿ semantic anchoring using existing approaches through a translation from dataflow systems to interface automata. We then describe our modeling tool FORMULA and apply it to the same example, showing how forgetful semantic anchoring can be avoided.