Avoiding Redundant Testing in Application Engineering

摘要

Many software product line testing techniques have been presented in the literature. The majority of those techniques address how to define reusable test assets (such as test models or test scenarios) in domain engineering and how to exploit those assets during application engineering. In addition to test case reuse however, the execution of test cases constitutes one important activity during application testing. Without a systematic support for the test execution in application engineering, while considering the specifics of product lines, product line artifacts might be tested redundantly. Redundant testing in application engineering, however, can lead to an increased testing effort without increasing the chance of uncovering failures. In this paper, we propose the model-based ScenTED-DF technique to avoid redundant testing in application engineering. Our technique builds on data flow-based testing techniques for single systems and adapts and extends those techniques to consider product line variability. The paper sketches the prototypical implementation of our technique to show its general feasibility and automation potential, and it describes the results of experiments using an academic product line to demonstrate that ScenTED-DF is capable of avoiding redundant tests in application engineering.