Refactoring of Model Transformations

Abstract

Model-to-model transformations between visual languages are often defined by typed, attributed graph transformation systems. Here, the source and target languages of the model transformation are given by type graphs (or meta models), and the relation between source and target model elements is captured by graph transformation rules. On the other hand, refactoring is a technique to improve the structure of a model in order to make it easier to comprehend, more maintainable and amenable to change. Refactoring can be defined by graph transformation rules, too. In the context of model transformation, problems arise when models of the source language of a model transformation become subject to refactoring. It may well be the case that after the refactoring, the model transformation rules are no longer applicable because the refactoring induced structural changes in the models. In this paper, we consider a graph-transformation-based evolution of model transformations which adapts the model transformation rules to the refactored models. In the main result, we show that under suitable assumptions, the evolution leads to an adapted model transformation which is compatible with refactoring of the source and target models. In a small case study, we apply our techniques to a well-known model transformation from statecharts to Petri nets.