Supporting constraint impact analysis in software processes at design-time

Context: In safety-critical domains such as automotive and aerospace, regulatory standards impose strict traceability constraints. These constraints are formalized during process design and followed throughout the software lifecycle. While such rigorous processes ensure software quality, they can hinder process adaptability in evolving engineering contexts. To address this, organizations adopt flexible processes that support process customization, such as adding or removing process steps or step constraints from a base process. However, modern engineering demands fine-grained dynamic process flexibility. Dynamic process flexibility allows individual constraints to be flagged as optional constraints ( OCs ) at design-time, enabling engineers to skip them at execution time based on contextual needs. Flagging a constraint as optional without identifying its impact on downstream constraints may introduce costly rework at execution time. Objective: To provide dynamic process flexibility by enabling individual process constraints to be flagged as optional constraints ( OCs ) at design-time while mitigating costly rework caused by not identifying their downstream impacts. Method: We present a design-time mechanism that automatically identifies downstream impacted constraints ( ICs ) when certain constraints are flagged as optional (OCs). The mechanism leverages dependency analysis to support engineers in identifying the impacts of constraints. Results: We evaluated the mechanism’s effectiveness through a controlled experiment with 42 participants, most of whom had beginner-level OCL knowledge. Participants using guidance support outperformed those without guidance in terms of accuracy (94.35% vs. 67.15%) and time efficiency (398.71s vs. 478.51s). Further statistical analysis using the Wilcoxon signed-rank test confirmed that these differences are statistically significant, with large effect sizes ( r > 0.5 ) observed across key performance measures (accuracy, precision and recall). The guidance support also reduced cognitive load and helped participants better understand the rationale behind constraint impacts. Conclusion: The proposed mechanism enhances dynamic process flexibility by allowing engineers to flag constraints as optional while identifying their downstream impacts. This contributes to more adaptable yet reliable software engineering processes.