Constraint-oriented variability modeling

Traditional syntax-oriented variability modeling specifies the set of possible system variants by explicitly describing how variability is expressed by linguistic means and it concentrates on the set of features that may or may not be present in a product. In contrast, constraint-based variability modeling defines variability in a top-down way by restricting the set of possible compositions of reusable artifacts in terms of properties and by including in this declarative description also some behavioral knowledge the experts may have about the product. Concretely, we propose here to integrate constraintbased solution space variability modeling with feature-oriented problem space variability modeling. This new approach paves the way to significantly simplify feature-oriented software development of product lines: Each feature is described by a set of constraints capturing what the feature contributes to a product variant and expects from it, and, for a given feature selection, the set of associated feature constraints allows synthesizing the set of product variants satisfying the constraints automatically. We illustrate and evaluate the proposed approach on the concrete example of a family of workflows from the bioinformatics domain.