Analyzing Variability in 25 Years of Industrial Legacy Software: An Experience Report

In certain domains, safety-critical software systems may remain operational for decades. To comply with changing requirements, new system variants are commonly created by copying and modifying existing ones. Typically denoted clone-and-own, software quality and overall maintainability are adversely affected in the long-run. With safety being pivotal, a fault in one variant may require the entire portfolio to be assessed. Thus, engineers need to maintain legacy systems dating back decades, implemented in programming languages such as Pascal. Software product lines (SPLs) can be a remedy but migrating legacy systems requires their prior analysis and comparison. For industrial software systems, this remains a challenge. In this paper, we introduce a comparison procedure and customizable metrics to allow for a fine-grained comparison of Pascal modules to the level of individual expressions. By that, we identify common parts of while also capturing different parts between modules as a basis for a transition towards anSPLs practice. Moreover, we demonstrate the feasibility of our approach using a case study with seven Pascal modules totaling 13,271 lines of code with an evolution-history of 25 years and show our procedure to be fast and precise. Furthermore, we elaborate on the case study and detail peculiarities of the Pascal modules, which are characteristic for an evolution-history of a quarter century.