Detecting and Evaluating Semantic Influences of Aspect Weaving in Aspect Oriented Models

摘要

Weaving an aspect may introduce undesired impacts on behavior of the base model or other afore-woven aspects. An approach to detect and evaluate the semantic influences of an aspect weaving on a pattern is presented. The pattern specifies a desired behavior that should stay unaltered or occur in the woven model. The detection and evaluation are based on the semantic relationship between a pattern and its projection that represents its actual semantics in the woven model. Five types of aspect weaving influences are identified. The approach has the Process Algebras (PA) as the underlying formalisms and has been implemented by mapping to the Concurrency Workbench (CWB) tool. An example shows the effectiveness.