一个划算的回归测试用例排序的成本-收益模型

AN ECONOMICAL COST-BENEFIT MODEL FOR TEST CASE PRIORITIZATION OF REGRESSION TESTING

回归测试是一个昂贵的测试过程,它验证修改的软件,确认是否给已经测试的代码引进了新的错误。因为在软件开发的过程中,测试人员给测试包中增加了新的测试用例,使软件确认更加困难。为了使回归测试更经济划算,提出了很多技术和方法。测试用例选择,测试包缩减,测试用例排序可以帮助我们减少需要运行的测试用例数目,同时能够更快地满足测试目标。测试用例排序是在满足某个性能目标的基础上,保证在测试过程中首先执行更重要的功能。以前的研究提出了一个性能目标:缺陷发现率(APFD),以及很多改善缺陷发现率的技术。但是这些技术运用起来非常昂贵,而且可能不会减少回归测试的成本。在以往研究的基础上,提出了一个成本-收益模型,为比较各种排序算法提供了一定依据。并在此基础上,提供了一个测试用例排序算法。

Regression testing is an expensive testing process. It validates modified software to ensure no new defects be introduced to the code having been tested. Because during the process of software exploring, testers add new test cases to test suite and make software validation even more difficult. Many techniques and methods have been presented to improve the cost-effective of regression testing. Test case selection, test case reduction,test case prioritization can help us reduce the number of test cases needed to execute and quickly meet testing objectives. Test case prioritization guarantees to execute earlier test cases with higher priority, on the basis of meeting specific performance object. Previous studies have presented such an object,weighted average percent of fault detected （APFD） and many techniques to improve it. But it is very expensive to execute these techniques and might not reduce the cost of regression testing. This paper presents a cost-benefit model on the basis of previous studies. The model provides a certain basis to compare different kinds of test case prioritization algorithms. It provides a regression test case prioritization algorithm based on the model.