The thesis focuses on the model-driven testing of Complex Event Processing (CEP) applications. A toolset, which was created earlier in the department, facilitates the development of CEP applications by using a platform-independent model (CEDL) to describe the event processing logic and transforming it to a platform-specific one that can be directly processed by an event processing engine.
The thesis extends this toolset with a model-driven test method which builds up a platform-independent model consisting of test specifications, after which it generates Java-based unit tests assuming a given platform for event processing, the Drools rule engine. The unit tests are responsible for submitting event sequences to the engine which trigger the rules defined in the original model, thus enabling the test to check for the expected effects. This way, the generated tests indirectly validate whether the transformation of the platform-independent CEDL model into the platform-specific one is correct or not.
The presented method follows the model-driven approach; that is the platform-independent model of the application is transformed to another platform-independent model helping in the specification of test event sequences. The latter can be further utilized in platform-specific models. The use of the approach increases both the modularity and the reusability of the method.
The document is organized as follows. After introducing the background, the thesis describes the steps of test case generation in detail, and demonstrates the presented method via two case studies.