Regresar

Towards Effective Mutation Testing for ATL

Abstract:

The correctness of model transformations is crucial to obtain high-quality solutions in model-driven engineering. Testing is a common approach to detect errors in transformations, which requires having methods to assess the effectiveness of the test cases and improve their quality. Mutation testing permits assessing the quality of a test suite by injecting artificial faults in the system under test. These emulate common errors made by competent developers and are modelled using mutation operators. Some researchers have proposed sets of mutation operators for transformation languages like ATL. However, their suitability for an effective mutation testing process has not been investigated, and there is no automated mechanism to generate test models that increase the quality of the tests. In this paper, we use transformations created by third parties to evaluate the effectiveness ATL mutation operators proposed in the literature, and other operators that we have devised based on empirical evidence on real errors made by developers. Likewise, we evaluate the effectiveness of commonly used test model generation techniques. For the cases in which a test suite does not detect an injected fault, we synthesize test models able to detect it. As a technical contribution, we make available a framework that automates this process for ATL.