Applications of Graph Transformations with Industrial Relevance

Abstract

In model-driven development, model transformations can be specified using an operational (imperative) or relational (declarative) approach. When using a relational approach, approved formal concepts are necessary to derive a conform operationalization. In general, though, it is not sure if implementations realize these formal concepts in an entirely correct way. Moreover, usually, available formal concepts neither cover every technicality, nor cover each additional optimization an implementation relies on. Consequently, conformance needs to be validated also on the implementation level. Conformance means that for each source model S and target model T related according to the relational specification, a corresponding implementation transforms S into T (and T into S in case that the specification is bidirectional). We present an automatic conformance testing approach for TGG implementations, where the Triple Graph Grammar (TGG) approach is an important representative of relational model transformation approaches. We show that the grammar character of TGGs is very convenient for the automatic generation of conformance test cases. In particular, test input models can be generated together with their expected result obtaining a complete oracle. We show how to measure test suite quality and evaluate our approach on our own TGG implementation. © 2012 Springer-Verlag.