History-sensitive recovery of product line features

Since software product lines (SPLs) increasingly have to satisfy additional requirements, their designs might degenerate over time. The degeneration is caused by various reasons. For instance, the features suddenly start to be realized and they evolved in inconsistent ways across multiple products. In an extreme case, the SPL code is fully or partially replicated and individually changed across several evolving products. In order to regain control of the SPL assets, a key activity is the design recovery of features from existing applications. However, existing techniques for feature analysis are not effective as they tend to explicitly rely on a single project history. They do not take into consideration change histories of features across multiple evolving products of a SPL. This research proposes a novel technique for history-sensitive feature recovery while repairing degenerated SPL designs. Our technique is ought to encompass a set of heuristics for facilitating SPL design recovery. We aim at investigating to what extent exploiting multi-product change histories allow accurate identification of: (i) code elements that contribute to each feature realization; and (ii) variability properties of the feature code. The empirical evaluation of our technique will be based on two industry case studies.