Abstract. The Object Constraint Language (OCL) is becoming more and more popular for model-based engineering, in particular for the development of models and model transformations. OCL is supported by a variety of analysis tools having different scopes, aims and technological corner stones. The spectrum ranges from treating issues concerning formal proof techniques to testing approaches, from validation to verification, and from logic programming and rewriting to SAT-based technologies. This paper is a first step towards a well-founded benchmark for assessing validation and verification techniques on UML and OCL models. The paper puts forward a set of UML and OCL models together with particular questions for these models roughly characterized by the notions consistency, independence, consequences, and reachability. The paper sketches how these questions are handled by two OCL tools, USE and EMFtoCSP. The claim of the paper is not to present a complete benchmark right now. The paper is intended to initiate the development of further UML and OCL models and accompanying questions within the UML and OCL community. The OCL community is invited to check the presented UML and OCL models with their approaches and tools and to contribute further models and questions which emphasize the possibilities offered by their own tools.
IntroductionModel-driven engineering (MDE) as a paradigm for software development is gaining more and more importance. Models and model transformations are central notions in modeling languages like UML, SysML, or EMF and transformation languages like QVT or ATL. In these approaches, the Object Constraint Language (OCL) can be employed for expressing constraints and operations, thus OCL plays a central role in MDE. A variety of OCL tools is currently available, but it is an open issue how to compare these tools and how to support developers in choosing the OCL tool appropriate for their project. This paper puts forward a set of UML and OCL models together with particular questions for these models. This set of models is intended to be a first version of an OCL analysis tool benchmark to be developed within the OCL and UML community.⋆ This research was partially funded by the Nouvelles Equipes program of the Pays de la Loire region (France).