Attribute Handling for Bidirectional Model Transformations: The Triple Graph Grammar Case

Abstract

When describing bidirectional model transformations in a declarative (relational) way, the relation between structures in source and target models is specified. But not only structural correspondences between source and target models need to be described. Another important aspect is the specification of the relation between attribute values of elements in source and target models. However, most existing approaches either do not allow such a relational kind of specification for attributes or allow it only in a restricted way.

We consider the challenge of bridging the gap between a flexible declarative attribute specification and its operationalization for the triple graph grammar (TGG) specification technique as an important representative for describing bidirectional model transformations in a relational way. First, we present a formal way to specify attributes in TGG rules in a purely declarative (relational) way. Then, we give an overview of characteristic barriers that bidirectional model transformation tool developers are confronted with when operationalizing relational attribute constraints for different TGG application scenarios. Moreover, we present pragmatic solutions to overcome the operationalization barriers for different TGG application scenarios in our own TGG implementation.