Syntax-preserving program slicing for C-based software product lines

Abstract

Program slicing is a well-established technique for identifying a reduced subset of a program based on pre-defined criteria, leading to complexity reduction in subsequent activities. Despite extensive study over the past 40 years, slicing techniques for software product lines (SPLs) remain notably scarce. The absence of dedicated SPL slicing approaches hinders their efficient analysis and maintenance, limiting the ability to focus only on relevant parts of the SPL. One reason for this deficiency is the complex nature of a common variability implementation: the use of C preprocessor #ifdef-annotations within C code. A slicing approach for C-based SPLs must address the intricate interplay between the C code and the functionality introduced by the C preprocessor. Effectively handling these intricacies will unleash the full potential of SPL analysis. In this paper, we present a novel syntax-preserving program slicing approach for C-based SPLs. Unlike existing methods, our approach enables the computation of program slices through an integrated analysis of both C and CPP code, while preserving the original program syntax (no element of its syntax is disregarded or changed). This preservation ensures that the resulting program slices remain authentic subsets of the SPL, making them suitable inputs for variability-aware analyses. Additionally, we demonstrate the practical applicability of these slices in the context of software transplantation, showcasing their potential for facilitating functionality transfer between different program versions. In contrast to existing transplantation approaches, our solution works without test cases, removing the need for product configuration and execution. Consequently, the variability implementation (along with all other contained preprocessor code) is preserved during the transplantation. We empirically evaluate our approach on four distinct open-source SPLs, showcasing its effectiveness in generating diverse program slices tailored to different slicing criteria. We asses the accuracy of our code representation, the time required for slicing and transplantation, the size reduction achieved through the slices, and the functionality of our variability-aware transplantation approach.