Reverse engineering feature-aware commits from software product-line repositories

Abstract

Version control systems such as Git and Subversion are widely used. Even though software often has to allow for variability, the commonly used version control systems do not provide functionality to handle variability explicitly in a commit. This lack of functionality for software with high variability, like software product lines, results in time intensive development processes, reduced clarity of the commit history, and no possibility to propagate or revert individual changes to features. In this thesis, we introduce our operator FeatureSplit which addresses these problems, by reducing the complexity of the commit history of software product line repositories. FeatureSplit achieves complexity reduction through separation and regrouping of committed changes by feature. In an empirical evaluation, we show that the resulting feature-aware commits simplify the managing and referencing of variability in version control system. FeatureSplit simplified on average 88% of commits in the evaluated repositories.