PLanD: An expressive lightweight language for describing static and dynamic software product line architectures

Model fragmentation adversely affects Software Product Line engineering by causing inconsistencies, increased maintenance costs, and hindered traceability across variability models, architectures, and assets. This paper proposes PLanD, a lightweight language that unifies feature modeling, architectural structure, asset mappings, and runtime reconfiguration rules in a single textual specification, while providing analyzable semantics for configuration validity and runtime adaptation. The evaluation combines two illustrative case studies from the healthcare and automotive domains, an analytical comparison with representative variability modeling approaches, and a controlled classroom study with 30 graduate students. In the user study, participants completed the modeling task faster with PLanD than with Clafer (15 vs. 25 min on average) and rated its structure as easier to follow. These results provide initial evidence that PLanD can reduce short-task modeling effort and improve traceability in small-to-medium examples. However, broader industrial validation, richer quantitative benchmarking, and large-scale scalability experiments remain future work.