Preliminary study on predicting version propagation in three-level component-based architectures

Abstract : Keeping a trace of the evolution of software archi-tectures is an important issue for architects. This paper is a summary of a submitted paper which states that the versioning activity does not propose solutions that fit software architectures, especially when dealing with co-evolution of different architecture representations that may be produced during the development process. This work is based on a three-level architecture description language (ADL) named Dedal, which gives a representation of architectures at the main stages of a software life-cycle. Dedal also performs co-evolution through change propagation within these representations. Another advantage of Dedal is that it has been formalized and thus provides a formal ground for studying version propagation. We based our study on substitutability relations that exist when a component is replaced at any of the three architecture levels. We aim at predicting compatibility of versioned artifacts in terms of impact on the different architecture levels.