Empirical studies of a safe regression test selection technique

摘要

Regression testing is an expensive testing procedure utilized tonvalidate modified software. Regression test selection techniques attemptnto reduce the cost of regression testing by selecting a subset of anprogram's existing test suite. Safe regression test selection techniquesnselect subsets that, under certain well-defined conditions, exclude nontests (from the original test suite) that if executed would revealnfaults in the modified software. Many regression test selectionntechniques, including several safe techniques, have been proposed, butnfew have been subjected to empirical validation. This paper reportsnempirical studies on a particular safe regression test selectionntechnique, in which the technique is compared to the alternativenregression testing strategy of running all tests. The results indicatenthat safe regression test selection can be cost-effective, but that itsncosts and benefits vary widely based on a number of factors. Innparticular, test suite design can significantly affect the effectivenessnof test selection, and coverage-based test suites may provide testnselection results superior to those provided by test suites that are notncoverage-based