Issue-based variability management

Context: Variability management is a key activity in software product line engineering. This paper focuses on managing rationale information during the decision-making activities that arise during variability management. By decision-making we refer to systematic problem solving by considering and evaluating various alternatives. Rationale management is a branch of science that enables decision-making based on the argumentation of stakeholders while capturing the reasons and justifications behind these decisions. Objective: Decision-making should be supported to identify variability in domain engineering and to resolve variation points in application engineering. We capture the rationale behind variability management decisions. The captured rationale information is useful to evaluate future changes of variability models as well as to handle future instantiations of variation points. We claim that maintaining rationale will enhance the longevity of variability models. Furthermore, decisions should be performed using a formal communication between domain engineering and application engineering. Method: We initiate the novel area of issue-based variability management (IVM) by extending variability management with rationale management. The key contributions of this paper are: (i) an issue-based variability management methodology (IVMM), which combines questions, options and criteria (QOC) and a specific variability approach; (ii) a meta-model for IVMM and a process for variability management and (iii) a tool for the methodology, which was developed by extending an open source rationale management tool. Results: Rationale approaches (e.g. questions, options and criteria) guide distributed stakeholders when selecting choices for instantiating variation points. Similarly, rationale approaches also aid the elicitation of variability and the evaluation of changes. The rationale captured within the decision-making process can be reused to perform future decisions on variability. Conclusion: IVMM was evaluated comparatively based on an experimental survey, which provided evidence that IVMM is more effective than a variability modeling approach that does not use issues.