Opening the black-box of model transformation

Abstract

The automated execution of model transformation plays a key role within Model Driven Development. The software that executes a transformation, commonly known as a transformation engine, receives the meta-models of the source and destination, and a set of transformation rules as input. Then the engine can be used to convert instances of the source meta-model to produce a destination model. Transformation engines are often seen as black boxes. In order to be sure of the correct execution, it is crucial to understand how a transformation engine executes a given transformation. This paper presents a method of capturing and analysing the activities carried out within the transformation engine by elaborating on existing tracing mechanisms used by existing engines. We compare the tracing mechanisms involved in four popular, rule-based transformation frameworks and highlight their shortcomings. A new trace meta-model is presented to deal with some of these shortcomings. These processes can be applied to all existing frameworks; as a proof of concept we have extended an existing traceability framework, based on our earlier work, to implement these mechanisms.