Enabling Efficient Automated Configuration Generation and Management

Abstract

Creating and managing valid configurations is one of the main tasks in software product line engineering. Due to the often complex constraints from a feature model, some kind of automated configuration generation is required to facilitate the configuration process for users and developers. For instance, decision propagation can be applied to support users in configuring a product from a software product line (SPL) with less manual effort and error potential, leading to a semi-automatic configuration process. Furthermore, fully-automatic configuration processes, such as random sampling or t-wise interaction sampling can be employed to test or to optimize an SPL. However, current techniques for automated configuration generation still do not scale well to SPLs with large and complex feature models. Within our thesis, we identify current challenges regarding the efficiency and effectiveness of the semi- and fully-automatic configuration process and aim to address these challenges by introducing novel techniques and improving current ones. Our preliminary results show already show promising progress for both, the semi- and fully-automatic configuration process.