Abstract :
Domain-specific modeling languages (DSMLs) show promise in improving model-based testing and experimentation (T&E) capabilities for software systems. This is because its intuitive graphical languages reduce complexities associated with error-prone, tedious, and time-consuming tasks. Despite the benefits of using DSMLs to facilitate model-based T&E, it is hard for testers to capture many variations of similar tests without manually duplicating modeling effort. This paper therefore presents a method called parameterized attributes that is used to capture points-of-variation in models. It also shows how parameterized attributes is realized in an open-source tool named the Generic Modeling Environment (GME) Template Engine. Finally, this paper quantitatively evaluates applying parameterized attributes to T&E of a representative distributed software system. Experience and results so show that parameterized attributes can reduce modeling effort after an initial model (or design) is constructed.
