Pattern-based Testing of (Programmed) Graph Transformations – Coverage and Adequacy
A reoccurring motive in software engineering lies in an increase in size, amount and quality of software products. This commonly goes along by a growing degree of automization, e.g., by using code generator, an increased amount of reuse of software artifacts as well as a rising level of abstraction from a technology-centric view to a point of view rooted in the problem domain. One key enabling technology for this is Model-Driven Software Development (MDSD).
Model transformations are one fundamental building-block to MDSD in this regard, due to the fact that they represent the very basis for an automated processing of models. Consequently, model transformations – representing a special class of programs and sharing certain specifics – should be subject to a rigorous and systematic quality assurance process. Since formal verification techniques are expected to prove infeasible in real-world, industrial-size applications (e.g., due to problems related to acceptance, practicability, or scalability) there exists an increased demand for model transformation testing techniques. Unfortunately, existing techniques are not directly applicable, and, consequently, new concepts are needed.
The upcoming talk is thus about presenting new research results from the testing domain related to a distinct class of model transformation programs, namely programmed graph transformations. The focus will be on the presentation of a novel, implementation- and graph pattern-based coverage notion for programmed graph transformations. In addition to that, there will be a part on important properties of derived test suites obtained through using this coverage notion, rounded off by reporting on experiments based on mutation analysis.