Well-Behaved Model Transformations with Model Subtyping

In model-driven engineering, models abstract the relevant features of software artefacts and model transformations act on them automating complex tasks of the development process. It is, thus, crucially important to provide pragmatic, reliable methods to verify that model transformations guarantee the correctness of generated models in order to ensure the quality of the final end product. In this paper, we build on an object-oriented algebraic encoding of metamodels and models as defined in the standard Meta-Object Facility and in tools, such as the Eclipse Modeling Framework, to specify a domain-specific language for representing the action part of model transformations. We introduce the big-step operational structural semantics of this language and its type system, which includes a notion of polymorphic model subtyping, showing that well-typed model transformations are well behaved. That is, that metamodel-conformant model transformations never go wrong.