Another Master’s Thesis within TRUCONF
Christian Burghard has written his Master’s Thesis as part of the TRUCONF project. The thesis describes the domain-specific language which was designed specifically for the modelling of AVL measurement device state machines, as well as the associated modelling tool and test case generation toolchain.
Christian Burghard: “Model-based Testing of Measurement Devices Using a Domain-specific Modelling Language”. Graz University of Technology, Institute of Software Technology, 2018. (PDF)
The work was supervised by Bernhard K. Aichernig. The defense was scheduled on May 24th, 2018.
The practice of model-based testing finds increasing application in industry, due to its potential to cope with the ever rising complexity of technical systems. For this reason, the AVL List GmbH is introducing a model-based testing methodology for the application to its portfolio of automotive measurement devices. In a previous project by AVL, the Graz University of Technology and the Austrian Institute of Technology, a model-based mutation testing approach has been developed. While this approach has been successfully validated in terms of functionality, it was rejected by AVL’s test engineers as they deemed its UML-based modelling front-end too difficult to use in their specific industrial setting. In the thesis at hand, we examine the tool composition-, usability- and experience-related reasons which have lead to the rejection of this modelling approach. We present a textual domain-specific language which we specifically tailored to the sole purpose of modelling AVL’s measurement device state machines. To ensure the intended improvement in the user experience of the modelling formalism, we developed the language in close and frequent collaboration with the test engineers. The resulting domain-specific language, called MDML, turned out to be very easy to learn and to be able to efficiently encode measurement device state machine models. In conjunction with MDML, we further developed a dedicated modelling tool, based on the well-known Eclipse-IDE. As we did with the language, we tailored this modelling tool to our use case and we also enriched it with a number of features providing user guidance and direct connection to AVL-internal data sources. Most importantly, we integrated a test case generation toolchain which we built around the pre-existing MoMuT test case generator. This toolchain involves a model transformation from MDML into object-oriented action systems to serve as input for the generator. It further involves the concretion of MoMuT’s abstract test case into executable test code. Lastly, we show that the capabilities of our model-based testing methodology are at least on-par with those of the previous UML-based one by means of a case study involving the generation and execution of tests for one of AVL’s measurement devices.
Model-Based Mutation Testing; Domain-Specific Language; State Machines; UML vs. DSL; Test Case Generation; User Experience; Model Transformation; OOAS