Exhaustively testing every product of a software product line (SPL) is a di cult task due to the combinatorial explosion of the number of products. Combinatorial interaction testing is a technique to reduce the number of products under test. However, it is typically up-to the tester in which order these products are tested. We propose a similarity-based prioritization to be applied on these products before they are generated. The proposed approach does not guarantee to find more errors than sampling approaches, but it aims at increasing interaction coverage of an SPL under test as fast as possible over time. This is especially beneficial since usually the time budget for testing is limited. We implemented similarity-based prioritization in FeatureIDE and evaluated it by comparing its outcome to the default outcome of three sampling algorithms as well as to random orders. The experiment results indicate that the order with similarity-based prioritization is better than random orders and often better than the default order of existing sampling algorithms.

No abstract

No abstract

Abstract. Software product line (SPL) engineering provides a promising approach for developing variant-rich software systems. But, testing of every product variant in isolation to ensure its correctness is in general not feasible due to the large number of product variants. Hence, a systematic approach that applies SPL reuse principles also to testing of SPLs in a safe and efficient way is essential. To address this issue, we propose a novel, model-based SPL testing framework that is based on a deltaoriented SPL test model and regression-based test artifact derivations. Test artifacts are incrementally constructed for every product variant by explicitly considering commonality and variability between two consecutive products under test. The resulting SPL testing process is proven to guarantee stable test coverage for every product variant and allows the derivation of redundancy-reduced, yet reliable retesting obligations. We compare our approach with an alternative SPL testing strategy by means of a case study from the automotive domain.

No abstract