Towards a new understanding of advice interference

Aspect-Oriented Programming (AOP) is an established programming paradigm for addressing crosscutting concerns that cannot be modularized using traditional dominant decomposition paradigms. However, the new language constructs introduced in AOP also raise new problems. One of such problems is Advice Interference, where interferences between aspects' advices by data and control flow dependencies jeopardize the correctness of an AOP program. Advice Interference has been qualitatively defined, but the cumulative effects of an aspect's advice on program contexts are overlooked in the definition. The definition also obscures the differences between Advice Interference and Weaving Interference, another well recognized type of AOP problems. Additionally, for Advice Interference, there is a fundamental flaw in the definition of relevant advices, between which interferences may occur. This paper investigates cumulative advice effects on a program context to provide a basis for a new understanding of Advice Interference. It also proposes a new semi-formal definition of Advice Interference that clarifies its differences from Weaving Interference. Finally, the flaw in the current definition of relevant advices for Advice Interference is addressed. The results of this research would provide AOP practitioners with a deepened understanding and fundamentally enhanced definition of Advice Interference and hence will inspire more robust solutions.