| Popis: |
We describe some progress towards a new common framework for model driven engineering, based on behavioral programming. The tool we have developed unifies almost all of the work done in behavioral programming so far, under a common set of interfaces. Its architecture supports pluggable event selection strategies, which can make models more intuitive and compact. Program state space can be traversed using various algorithms, such as DFS and A*. Furthermore, program state is represented in a way that enables scanning a state space using parallel and distributed algorithms. Executable models created with this tool can be directly embedded in Java applications, enabling a model-first approach to system engineering, where initially a model is created and verified, and then a working application is gradually built around the model. The model itself consists of a collection of small scripts written in JavaScript (hence "BPjs"). Using a variety of case-studies, this paper shows how the combination of a lenient programming language with formal model analysis tools creates an efficient way of developing robust complex systems. Additionally, as we learned from an experimental course we ran, the usage of JavaScript make practitioners more amenable to using this system and, thus, model checking and model driven engineering. In addition to providing infrastructure for development and case-studies in behavioral programming, the tool is designed to serve as a common platform for research and innovation in behavioral programming and in model driven engineering in general. |
