Managing Variability Using Heterogeneous Feature Variation Patterns

Feature-driven variability is viewed as an instance of multi-dimensional separation of concerns. We argue that feature variation concerns can be presented as pattern-like entities – called feature variation patterns – cross-cutting heterogeneous artifacts. We show that a feature variation pattern, covering a wide range of artifact types from a feature model to implementation, can be used to manage feature-driven variability in a software development process. A prototype tool environment has been developed to demonstrate the idea, supporting the specification and use of heterogeneous feature variation patterns. We illustrate the idea with a small example taken from J2EE, and further study the practical applicability of the approach in an industrial product-line.