Implementation of a regression test selection technique for C++ software. (c2001)

Abstract

Regression testing is an important but expensive software maintenance activity performed with the aim of providing confidence in modified software. Regression test selection techniques reduce the cost of regression testing by selecting tests for a modified program from a previously existing test suite. Many researchers have addressed the regresion test selection problem for procedural language software, but few have addressed the problem for object-oriented software. We present an implementation and experimental evaluation of a technique proposed by Rothermel, Harrold, and Dedhia. This technique is a regression test selection technique for object-oriented C++ software. The technique constructs graph representation for software, and uses these graphs to select tests, from the original test suite, that execute code that has been changed for the new version of the software. The technique is strictly code based, and requires no assumptions about the approach used to specify or test the software initially. The technique applies to applications programs that use modified classes. The results indicate that the implemented technique can reduce the number of regression tests that must be run.