Reflexion Models for State Diagrams

Abstract

State diagrams are commonly used for specification or illustration of behavior. They appear on all levels, from high-level system behavior down to behavior of a single function. However, state diagrams are often implemented manually, and there is no guarantee that the implementation corresponds to the specification. Furthermore, a state machine may be implemented in the code without having a specification in form of a state diagram, and it might be that the developer was not even aware that he is implementing one. We call this kind of implementations implicit state machines. Our practical experience has shown that such code is very hard to comprehend. In this paper, we describe our work on applying the idea of reflexion models on state diagrams.