Redesigning the Spoofax Testing Language

Volker Lanting

The Spoofax Testing Language (SPT) is the existing solution for testing in the Spoofax language workbench. It allows developers of domain specific languages to write their test cases declaratively. As it aims to be implementation agnostic, developers don’t need to concern themselves with the details of the artifacts generated by Spoofax, and can write their tests before implementing their language. However, the previous implementation has become slow and unusable for larger test suites and can not be executed programatically. This means it can’t be used for continuous integration and automated regression testing. As Spoofax was redesigned to become more robust and platform independent, the previous SPT is no longer compatible. We took this opportunity to redesign SPT.

In this thesis we will discuss the benefits of a testing approach like SPT, how far along it is on the path of testing any language, and what is required to make it usable by modern day developers. We will analyze the problems that SPT had to tackle, how it solved them, and which problems still remain. Finally, we present and evaluate our new design and implementation to solve some of these remaining problems. We created a platform independent, real-time performant, easily extendable architecture that allows SPT to be used for automated tasks such as continuous integration and the automated grading of students’ domain specific languages.

---

Previous: | Redesigning the Spoofax Testing Language
Next: Arjen Rouvoet | Discussion about shoulders of giants: related work in research and writing