Metamodels contain errors.

Nevertheless, they are not tested systematically, yet.

Therefore, errors are found late. Dependent artifacts like models or tools may already have been created. Consequently, correcting erroneous meta-models and adapting the dependent artifacts requires additional effort. This effort can be saved if metamodels are tested independently of other artifacts.

MMUnit is the implementation of an approach for automated metamodel testing based on the specification of positive and negative example-models. MMUnit builds on the Eclipse platform and the JUnit framework.

What next?

Look at the screenshots,

read the (small) documentation,

download MMUnit, and

read about the ideas behind MMUnit.

News

2009-09-01: MMUnit 2.1.0 released. This version supports a number of coverage criteria for determining how well a set of test specifications covers a metamodel under test. Additional coverage criteria can be added via an extension point. This version is Galileo-ready.

2008-09-04: MMUnit 2.0.2 released. Bug fix release.

2008-07-15: MMUnit 2.0.1 released. Bug fix release.

2008-07-10: MMUnit 2.0.0 released. With this version, MMUnit is Ganymede-ready. The version 2.0.0 is a complete (and incompatible) reimplementation of MMUnit. Stability and test failure messages are drastically improved. Furthermore, we improved the graphical test specification editor: we added icons for the elements, and the conjunction symbol now contains a "^" instead of "+".