Static Detection of Redundant Test Cases: An Initial Study

As software systems evolve, the size of their test suites grow due to added functionality and customer-detected defects. Many of these tests may contain redundant elements with previous tests. Existing techniques to minimize test suite size generally require dynamic execution data, but this is sometimes unavailable. We present a static technique that identifies test cases with redundant instruction sequences, allowing them to be merged or eliminated. Initial results at ABB show that 7%-23% of one test suite may be redundant.