Measurement of the Complexity of Variation Points in Software Product Lines

Feature models are used in member product configuration in software product lines. A valid product configuration must satisfy two kinds of constraints, multiplicity of each variation point and dependencies among the variants in a product line. The combined impact of the two kinds of constrains on product configuration should be well understood. In this paper we propose a measurement, called VariationRank, that combines the two kinds of constraints to assess the complexity of variation points in software product lines. Based on the measurement we can identify those variation points with the highest impact on product configurations in a product line. This information could be used as guidance in product configuration as well as for feature model optimization in software product lines. A case study is presented and discussed in this paper as well.