Scalable and Efficient Sampling for Product-Line Testing

A software product line (SPL) is a family of software products that share a set of common features. Exhaustively testing every product of an SPL is a difficult task due to the combinatorial explosion of the number of products. Several sampling approaches have been proposed to select a set of products that can be used to test SPL. However, these approaches do not scale very well, especially for large SPLs such as the Linux Kernel. In addition, it is typically up-to testers in which order these products are tested. The testers may wish to order these products to detect faults as soon as possible. The products have been prioritized based on domain knowledge or feature model criteria, but not much attention has been paid to criteria at code-based level. We plan to use evolutionary testing approaches with different inputs to fitness functions to explore the configuration space of SPL feature models. Using the criteria from feature-model and code as inputs to evolutionary testing approaches, we want to investigate whether we can increase the efficiency of SPL testing w.r.t. finding more faults. Furthermore, we want to investigate code-based metrics which can be used to enhance SPL testing. In this proposal, we present research questions, research methods and a concrete working plan to investigate how we can employ evolutionary testing approaches to increase the efficiency of SPL testing.