A comparison of 10 sampling algorithms for configurable systems

摘要: Almost every software system provides configuration options to tailor the target platform and application scenario. Often, this configurability renders analysis of individual infeasible. To address problem, researchers have proposed a diverse set sampling algorithms. We present comparative study 10 state-of-the-art algorithms regarding their fault-detection capability size sample sets. The former is important improve quality latter reduce time analysis. In nutshell, we found that with larger sets are able detect higher numbers faults, but simple small sets, such as most-enabled-disabled, most efficient in contexts. Furthermore, observed limiting assumptions made previous work influence number detected ranking Finally, identified technical challenges when trying avoid assumptions, which questions practicality certain