On the Usage of TGGs for Automated Model Transformation Testing

As model transformations are fundamental to model-driven engineering, assuring their quality is a central task which can be achieved by testing with sufficiently adequate and large test suites. As the latter requirement can render manual testing prohibitively costly in practice, a high level of automation is advisable. Triple Graph Grammars (TGGs) have been shown to provide a promising solution to this challenge as not only

test case generators

, but also generic

test oracles

can be derived from them. It is, however, unclear if such generated test suites are indeed adequate and, as different strategies can be used to steer the test generation process, a systematic means of comparing and evaluating such test suites and strategies is required.

In this paper, we extend existing work on TGG-based testing by(i) presenting a generic framework for TGG-based testing, (ii) describing a concrete instantiation of this framework with our TGG tool eMoflon, and (iii) exploring how the well-known technique of

mutation analysis

can be used to evaluate a set of test generation strategies by analyzing the generated test suites.