An Action-Stack Based Selective-Undo Method in Feature Model Customization

摘要

Feature models are widely adopted in domain-oriented software reuse. The feature model customization is a process to determine which features are selected into or removed from an application. Auto-propagation is a practical strategy to assist customization but has a side-effect that may select or remove unintended features. To handle such a side-effect, a mechanism called selective-undo is needed that eliminates unintended selections and removals while preserving other selections and removals. This paper presents a selective-undo method based on action-stack by using the intrinsic relationships in a feature model, and investigates the implementation of the corresponding algorithms.