A Comparative Study of Coarse - and Fine - Grained Safe Regression Test Selection

Regression test selection techniques reduce the cost of regression testing by selecting a subset of an existing test suite for reveri cation of a modi ed program. Over the past two decades, numerous regression test-selection techniques have been described in the literature. Initial empirical studies of a few of these techniques have shown that they can be bene cial, but so far, few studies have compared techniques by performing the analysis for di erent methods on the same set of subjects. This paper describes a comparative evaluation of di erent regression test-selection techniques on the same experimental subjects. In particular, the two methods DejaVu and TestTube, developed by the authors, are evaluated and compared in terms of a cost model that considers their relative precision and e ciency. The data reveal that in some instances TestTube's relative lack of precision does not prevent it from competing with the more precise DejaVu, yet in other instances, DejaVu's superior precision gives it a clear advantage over TestTube. Since TestTube requires less critical-phase time cost, such variations in performance can complicate the tester's choice of which technique to use. As a result, a hybrid technique is suggested, which combines the gains in e ciency from TestTube with the increased precision of DejaVu.