Restructuring Variability in Software Product Lines using Concept Analysis of Product Configurations

Abstract

The management of variability plays an important role in successful software product line engineering. As the set of products that is derived from the product line and their requirements are constantly changing, the variability in the product line needs to evolve as well. A typical problem in in such an evolution scenario is that the number of variable features and variants will explode, and thus become unmanageable. One of the reasons for this explosion is that obsolete variable features are not removed. In order to address this problem, we present a new method for restructuring and simplifying the provided variability in a software product line. Our method is based on concept analysis. It analyzes the realized variability in a software product line, and constructs a lattice that provides a classification of the usage of variable features in real products derived from the product line. We show how this classification can be used to derive restructuring strategies for variability that solve the problem of variability explosion. The effectiveness of our method is demonstrated by presenting a case study of restructuring the variability in a large industrial software product line