Detection of Syntactic Aspect Interaction in UML State Diagrams Using Critical Pair Analysis in Graph Transformation

Zaid Altahat, Tzilla Elrad, Luay Tahat, Nada Almasri

Aspect Oriented Modeling separates crosscutting concerns by defining Aspects and composition mechanisms at the model level. Composition of multiple Aspects will most likely result in more than one Aspect matching the same join points. Consequently, Aspects do not always interact in a predictable manner when woven together. Intended interaction among aspects is designed by the system designer. Unintended interaction (or interference) must be automatically managed. When the woven aspect demonstrates a behavior that is different than its autonomous behavior, then this is a potential interference. Interference has been recently reported in Aspect Oriented Software Development (AOSD) by the industry. Leaving this problem unsolved may result in erratic software behavior and will hinder the adaptation of AOSD by the industry. This identified problem is similar to a phenomenon that exists in graph transformation systems where multiple Graph Transformation rules share some conflicting elements, it is referred to as Critical Pair Analysis and it provides an algebraic-based mechanism to detect and analyze the interaction of the rules. In this paper we propose a framework to detect unintended Aspect interaction at the model level. The proposed framework transforms Aspects modeled in UML State Diagram to Graph Transformation Rules, and then it applies Critical Pair Analysis to detect unintended interactions among aspects. This will enable developers to specify only the order of precedence for intended interaction among aspects without the need to manually investigate unintended interactions for the combinations of every Aspect to every other Aspect in the system. The proposed interaction detection solution is automated, modular, and independent of the base model; which adds the advantage of not having to re-evaluate the interaction each time the base model changes.

Knowledge Graph



Sign up or login to leave a comment