Exceptional Transformations

Abstract

As model transformations are increasingly used in model-driven engineering, the dependability of model transformation systems becomes crucial to model-driven development deliverables. As any other software, model transformations can contain design faults, be used in inappropriate ways, or may be affected by problems arising in the transformation execution environment at run-time. We propose in this paper to introduce exception handling into model transformation languages to increase the dependability of model transformations. We first introduce a classification of different kinds of exceptions that can occur in the context of model transformations. We present an approach in which exceptions are modelled in the transformation language and the transformation designer is given constructs to define exception handlers to recover from exceptional situations. This facilitates the debugging of transformations at design time. It also enables the design of fault-tolerant transformations that continue to work reliably even in the context of design faults, misuse, or faults in the execution environment.