A growing share of products expose sophisticated configurability to handle variations in user and context requirements. Configurators usually rely on variability models, like feature models (FMs), that structure and constrain the available options. However, most feature-based configuration techniques do not scale well to collaborative contexts. They offer limited mechanisms to determine responsibilities and access rights, to schedule configuration tasks, and to resolve conflicts. Fragments of solutions already exist but a unified and formal foundation for collaborative feature-based configuration is still missing. To provide enhanced control and guidance, we specify responsibilities and rights with views on the FM. Views establish insulated spaces in which users can safely configure the part of the FM assigned to them. The configuration of these views is regulated by a workflow that defines the configuration process. The result of that combination is a new formalism called feature-based configuration workflow. Finally, to handle conflicts between user decisions, we develop a range fix generation algorithm. These concepts and their properties are integrated into a sound mathematical framework. Our contribution is motivated and illustrated through several real-world applications: a product line of meeting management applications (PloneMeeting), a product line of communication protocols used in the aerospace industry (CFDP), the Linux kernel, and an operating system for embedded applications (eCos). The definitions and algorithms are implemented in a toolset that extends SPLOT, an open source configuration environment, and YAWL, a comprehensive workflow management environment. This toolset demonstrates the efficiency and applicability of our contribution.
Feature-based configuration: collaborative, dependable, and controlled
Hubaux, A. (Author). 9 Jan 2012
Student thesis: Doc types › Doctor of Sciences