Abstract
In the field of Domain Languages Engineering, Semantic Variation Points are an important issue. This crucial information is often related to the dynamic semantics of systems. Identifying and understanding it is a requisite for all model-based activities (design, simulation, test, formal verification, etc.). Most of the time, semantic variation points are only informally identified in a documentation associated with a metamodel: they are not part of the metamodel itself, and there is currently no mechanism to capture them explicitly. We propose a template-based notation enabling semantic variation points to be clearly and explicitly identified within the metamodel, using template parameter definitions. Semantic variation points can then be intuitively fixed by parameter binding at both model and metamodel levels. We illustrate our proposal with a templated version of the UML 2 state machine metamodel. Finally, we describe a prototype implementation of our mechanisms in the context of the Eclipse Modeling Framework.