A Process for Design-Level Detection of Concern Interactions in Aspect-Oriented Systems
The authors present a process for detecting concern interactions in Aspect-Oriented (AO) designs expressed in the UML and their domain specific Weaving Rule specification Language (WRL). The process consists of two tasks: a light-weight syntactic analysis of the AO model to reveal advice overlaps (e.g., instances where multiple advice applies to the same join point) as potential sources of interaction. Verification of desired model properties before and after weaving to confirm/reject findings of task 1 and/or to reveal new interactions. At the heart of task 2 is a weaving process that maps an unwoven Aspect-Oriented model to a behaviorally equivalent woven Object-Oriented model.