Detection and Resolution of Feature Interactions, the Early Light Way

The feature interaction problem has been recognized as a general problem of software engineering, whenever one wants to reap the advantages of incremental development. In this context, a feature is a unit of change to be integrated in a new version of the system under development, and the problem is that new features may interact with others in unexpected ways. We introduce a common abstract model, to be built during early requirement analysis in a feature oriented development. The model is common, since all the features share it, and is an abstraction of the behavioural model retaining only what is needed to characterize the features with respect to their possible interactions. The basic constituents are the abstract resources that the features access in their operations, the access mode (read or write), and the reason of each access. Given the model, the interactions between the features are automatically detected, and the goal oriented characterization of the features provides the developers with valuable suggestions on how to qualify them as synergies or conflicts (good and bad interactions), and on how to resolve conflicts. We provide evidence of the feasibility of the approach with an extended example from the Smart Home domain. The main contribution is a lightweight state-based technique to support the developers in the early detection and resolution of the conflicts between features.