Closing the Gap Between Object - OrientedModeling of Structure

Due to hardware developments, strong application needs and the overwhelming innuence of the net in almost all areas, distributed systems, especially software systems, have become one of the most important topics for nowadays software industry. Unfortunately, distribution adds its share to the problems of developing complex software systems. Heterogeneity in both, hardware and software, concurrency, distribution of components and the need for in-teroperability between diierent systems complicate matters. Although subject to permanent changes, distributed systems have high requirements w.r.t. dependability and performance. Moreover, new technical aspects like resource management, load balancing and deadlock handling put an additional burden onto the developer. Nowadays programming languages are able to deal with these problems but work on a level of abstraction which is much too low. This makes complex systems hard to implement and much harder to maintain and change. Nowadays design methods fail to describe the speciic properties of distributed systems in an adequate manner. One of the essential drawbacks of the methods which are around for some time now, is the lack of support for the dynamic aspects of distributed software systems. Based on object-oriented analysis and design, our approach adds the aspects of architectural modeling to the usual structural modeling of distributed software systems. Moreover, behav-ioral modeling w.r.t. concurrency, competition for resources and coordination is made more intuitive by means of an adequate modeling language. We use a speciic kind of high-level petri-nets, so-called Object Coordination Nets (OCoNs). The result is an integrated design language which seamlessly combines the well-known approaches of object-oriented analysis and development with methods which work well for the speciic aspects of distributed software systems. Our long term goal is an integrated development methodology for distributed software systems. Currently, we are working on a sound theoretical framework for contract-based reasoning, advanced tool support for modeling, analysis and code generation as well as non-toy case studies to evaluate our approach. Abstract The UML as standardized language for visual object-oriented modeling allows to capture the requirements as well as the structure and behavior of complex software systems. With the increasing demands of todays systems, behavior aspects like concurrency, distribution and reactivity become more important. But the language concepts of the UML for describing behavioral aspects are weak compared to its concepts for describing structures. Besides a lack of visual expressiveness, a deeper integration with the structure speciication is missing. In order to close this gap, an expressive language …