Mutation Operators for the Atlas Transformation Language

Faults in model transformations will result in defective models, and eventually defective code. Correction of defects at the code level is considered very late and is often expensive. Uncorrected defects in the models will propagate to other artifacts; thus, adversely affecting the quality of the end product. Moreover, defect propagation may result in a system that does not meet the stakeholders´ requirements. Therefore, model transformations must be thoroughly tested to maintain product quality, while keeping development cost at reasonable levels. Existing literature on model transformation verification and validation has considered coverage based techniques. Mutation testing is a popular technique that has been extensively studied in the literature, and shown to perform better than coverage based techniques. To support the mutation testing of model transformations, this paper proposes a suite of mutation operators for the Atlas Transformation language (ATL). The effectiveness of the proposed operators is evaluated using a model transformation program, implemented in ATL, to transform Use Case Maps models to UML Activity Diagrams. The results show that the proposed operators can successfully detect inadequacies in an example test suite.