Towards Automated Testing of Abstract Syntax Specifications of Domain-Specific Modeling Languages

The abstract syntax of domain-specific modeling languages (DSMLs) can be defined with meta-models. Meta-models can contain errors. Nevertheless, they are not tested systematically and independently of dependent artifacts like models or tools. Consequently, errors are found late—not before the dependent artifacts have been created. Since all dependent artifacts must be adapted when an error is found, this results in additional error correction effort. This effort can be saved if the meta-model of a DSML is tested early. We propose an approach for automated meta-model testing that is based on a test meta-model, whose instances specify positive and negative examplemodels. Automated meta-model testing helps to find errors in a DSML’s meta-model early and, thus, saves time and money when developing a DSML with accompanying tool support.