Using Synthetic Coverage Information to Evaluate Test Suite Prioritizers
Gregory M. Kapfhammer
Allegheny College, USA.
Regression testing techniques reorder and reduce a test suite in an attempt to ensure that test execution is both efficient and effective. The empirical evaluation of an approach to either test prioritization or reduction often involves the use of software tools to test real-world case study applications. Since many regression testing techniques are coverage-based, this presentation describes a method for using synthetic coverage to experimentally assess test suite prioritizers. As an alternative to directly testing case study applications, the use of automatically generated synthetic coverage supports the systematic evaluation of regression testing schemes. To this end, we present algorithms that automatically generate test coverage and timing information while varying both the amount of coverage overlap and the anticipated difficulty of reordering the test suite. This presentation concludes with an examination of empirical results that measure the efficiency and effectiveness of the synthetic coverage generators.