Composition Analysis of Multiple Aspects in Aspect-oriented Programs

This paper presents a classification of widely studied approaches that focus interaction analysis and composition of multiple aspect(s) in aspect-oriented programs (AOP). It is evident that AOP has an ability to distort semantics of base-programs and aspects themselves due to advice-method or advice-advice interactions in terms of control now and data sharing features. Thus, it entails a comprehensive analysis in order to pinpoint ambiguities at semantic level; especially in case of multiple aspects that remains focus of this paper. One possible interaction (method-advice) may inviolate a predicate of another advice from another aspect, residing in the base-program. We narrow our analysis to understand such intricacies and present a bunch of rules to understand weaving semantics (in particular, advice-advice interaction from two different aspects) and also propose precedence laws for aspects to be woven. In this paper discrete computation of aspects using operational semantics enables better interaction analysis of aspects and specifying their behavior by excluding the dependencies of any specific AOP language implementation. The contributions are threefold: (1) a survey based on an informal classification of proposed approaches targeted towards interfering aspects (2) our proposed formal definitions for composing aspects through evaluation rules using operational semantics (3) addressing precedence issues in aspects to some degree and offering a preliminary solution based on proactive execution order of aspects.