Feather: A Feature Model Transformation Language

Abstract: Feature modeling has been a very popular approach for variability management in software product lines. Building a feature model requires substantial domain expertise, however, even experts cannot foresee all future possibilities. Changing requirements can force a feature model to evolve in order to adapt to the new conditions. Feather is a language to describe model transformations that will evolve a feature model. This article presents the structure and foundations of Feather. First, the language elements, which consist of declarations to characterize the model to evolve and commands to manipulate its structure, are introduced. Then, semantics grounding in feature model properties are given for the commands in order to provide precise command definitions. Next, an interpreter that can realize the transformations described by the commands in a Feather script is presented. Finally, effectiveness of the language is discussed using two realistic examples, where one of the examples includes a system from a dynamic environment and the other employs a system that has a large feature model containing 1,227 features.