Linking feature models to code artifacts using executable acceptance tests

Abstract

A feature model is a representation of the requirements in a given system abstracted at the feature level. Linking conceptual requirements in feature models to actual implementation artifacts provides for many advantages such as increased program comprehension, implementation completeness assessment, impact analysis, and reuse opportunities. However, in practice, as systems evolve, traceability links between the model and the code artifacts may become broken or outdated. In this paper, we contribute an approach to provide traceability links in a way that ensures consistency between the feature model and the code artifacts, enables the evolution of variability in the feature model, and supports the product derivation process. We do that by using executable acceptance tests as a direct traceability link between feature models and code artifacts. We evaluate our approach and present a brief overview of the tool support we provide. © 2010 Springer-Verlag Berlin Heidelberg.