Bridging the gap between software variability and system variant management: Experiences from an industrial machinery product line

Autoren Stefan Fischer
Lukas Linsbauer
Roberto E. Lopez-Herrejon
Alexander Egyed
Rudolf Ramler
Editoren
TitelBridging the gap between software variability and system variant management: Experiences from an industrial machinery product line
Buchtitelroceedings of the 41st Euromicro Conference series on Software Engineering and Advanced Applications (SEAA 2015)
Typin Konferenzband
VerlagIEEE
ISBN978-4673-7585-6
DOI10.119/SEAA.2015.57
MonatAugust
Jahr2015
Seiten402-409
SCCH ID#1522
Abstract

Companies that develop complex systems often do so in the form of product lines, where each product variant can be configured to a certain degree to fit a customer’s specific requirements. Features cannot be combined arbitrarily in a product line. The knowledge which features require or exclude each other is represented in form of variability models. Unfortunately, in practice, such variability models do not exist or they are oriented towards the needs and viewpoints of specific organizational units, e.g. sales, manufacturing, hardware engineering, or software development. In this paper we present our experiences in building a variability model for the highly configurable software part of a complex mechatronic system produced by one of our industrial partner companies. The company already had support and processes for product variant management in place for sales and hardware manufacturing. However, the corresponding variability model was at the level of the overall system and excluded the variability of the software part. The paper discusses the resulting problems and challenges and describes the approach we selected to bridge the gap that existed between product variants and software configurations. The goal and driving motivation for our work was the improvement of the software development process and specifically the testing of software variants. The paper also shows how software configuration and testing activities can benefit from an appropriate variability model.