Análise comparativa de algoritmos de técnica combinatória para testes em linhas de produtos de software

Abstract

Software Product Line (SPL) is an approach that aims to develop a family of products that have differences. These differences are called commonality and variability of SPL, and the feature model can represent both. The increased variability and the number of products make SPL test exhaustive and in some cases not feasible. Generating tests from models of features, i.e., generating a valid and representative test of all possible product configurations subset, is still an open challenge. Thus, this paper describes a comparative analysis of combinatorial testing algorithms, an approach used to generate test configurations that cover all possible interactions between features. These techniques were applied in six SPL and the results were evaluated by a set of specific measures for software product lines. Given the results of the evaluation, it is highlight the main challenges of combinatorial testing and the main advantages of the algorithms discussed. One challenge consists in selecting a small subset of all possible configurations for testing. This sample should cover relevant characteristics of the feature model, while staying as small as possible. Using the algorithms is possible to select a small number of settings in large models features in reasonable time.