Carrying Ideas from Knowledge-Based Configuration to Software Product Lines

摘要

Software variability modelling (SVM) has become a central concern in software product lines - especially configurable software product lines (CSPL) require rigorous SVM. Dynamic SPLs, service oriented SPLs, and autonomous or pervasive systems are examples where CSPLs are applied. Knowledge-based configuration (KBC) is an established way to address variability modelling aiming for the automatic product configuration of physical products. Our aim was to study what major ideas from KBC can be applied to SVM, particularly in the context of CSPLs. Our main contribution is the identification of major ideas from KBC that could be applied to SVM. First, we call for the separation of types and instances. Second, conceptual clarity of modelling concepts, e.g., having both taxonomical and compositional relations would be useful. Third, we argue for the importance of a conceptual basis that provides a foundation for multiple representations, e.g., graphical and textual. Applying the insights and experiences embedded in these ideas may help in the development of modelling support for software product lines, particularly in terms of conceptual clarity and as a basis for tool support with a high level of automation.