Improved Flexibility and Scalability by Interpreting Story Diagrams

Abstract

In this paper, we present an interpreter for Story Diagrams working on Eclipse Modeling Framework (EMF) models. The interpreter provides a more flexible and, under certain circumstances, a more scalable solution than the compiled Java code generated from Story Diagrams by Fujaba. of Dynamic EMF even allows the evolution of meta models at runtime. Story Diagrams can now be modeled and executed within Eclipse. They can be modified and re-executed by the Story Diagram interpreter immediately without recompiling the source code and restarting the application. Our implementation also supports higher-order transformations by using Story Diagrams to modify other Story Diagrams. generation is not applicable, like running systems. While interpretation obviously results in performance drawbacks, we demonstrate that the Story Diagram interpreter is able to improve the performance in certain worst-case situations compared to the average generated code. This is achieved by a dynamic ordering of the matching process, which considers the actual number of elements in an association at runtime. Such a dynamic ordering can minimize the matching effort considerably. In contrast, Fujaba generated code uses a static matching strategy. Whereas the Fujaba Story Diagrams have potentially high performance fluctuations, the performance of the Story Diagram interpreter is steadier and more scalable compared to the generated Java code.