Recovering Variability Information from Source Code of Clone-and-Own Software Systems

Clone-and-own prevails as an ad-hoc reuse strategy that addresses changing requirements by copying and modifying existing system variants. Proper documentation is typically not cherished and knowledge about common and varying parts between individual variants, denoted their variability information, is lost with a growing system family. With overall maintainability impaired in the longrun, software product lines (SPLs) or concepts thereof, can be a remedy. However, migrating a system family towards structured reuse requires a prior recovery of the systems' variability information. For software systems resulting from clone-and-own, this information is not explicitly available and recovering it remains an open challenge. We aim to bridge this gap and propose a fine-grained metric and analysis procedure, which compares software systems to the extent of individual statements including their nesting. By that, we recover variability information from software systems written in imperative programming languages. Moreover, we create a software family representation of all analyzed systems, called a 150% model, which contains implementation artifacts and their identified variability information. We demonstrate the feasibility of our approach using two case studies implemented in Java and show our approach to exhibit a good performance and the 150% model to precisely capture variability information of the analyzed systems.