Code-based variability model extraction for software product line improvement

Successful Software Product Lines (SPLs) evolve over time. However, one practical problem is that during SPL evolution the core assets, especially the code, tend to become complicated and difficult to understand, use, and maintain. Typically, more and more problems arise over time with implicit or already lost adaptation knowledge about the interdependencies of the different system variants and the supported variability. In this paper, we present a model-based SPL improvement process that analyzes existing large-scale SPL reuse infrastructure to identify improvement potential with respective metrics. Since Conditional Compilation (CC) is one of the most widely used mechanisms to implement variability, we parse variability-related facts from preprocessor code. Then we automatically extract an implementation variability model, including product configuration and variation points that are structured in a hierarchical variability tree. The extraction process is presented with concrete measurement results from an industrial case study.