RIVAR · Rich Interfaces for Verifiable Aspect Reuse

Aspect-Oriented Programming (AOP) has been recognised as an important contribution to the research on modularisation techniques for software development, especially of large and complex systems. However, to date once written aspects cannot be reused safely in new contexts and applications. The proposed project aims to study techniques for extending aspect specification so that they can be safely and verifiably reused across application borders. In particular, current aspect specifications define an aspect's interface (using so-called pointcut definitions) only based on syntactic information about the structure of a base program. To enable verifiably safe aspect reuse, these interface specifications need to be extended to provide semantically rich means of expressing aspect developers’ assumptions about the context in which an aspect is to be used and the guarantees that the aspect will make to the program into which it is woven. The proposed project first uses empirical and ethnographic techniques for building a taxonomy of assumptions and guarantees aspect developers typically make. Based on this taxonomy, extensions to common AOP languages are developed and integrated into existing aspect tooling from AOSD Europe. Where possible, also automatic or semi-automatic verification techniques for such aspect specifications will be studied.