What is convergence debugging? Convergence Debugging is an automated debugging method that isolates a set of test cases that converge on the internal root cause of a failure. Convergence Debugging gives means to measure the debugging effectiveness of a set of test cases. It also can select a set of test cases that maximize debugging effectiveness.

Convergence Debugging Isolate effective debug testcases Measure debug effectiveness Select effective debug testcases

Why convergence debugging? Converging test cases are closer to the fault with respect to program control and data. Analyzing differences among converging test cases is more effective in locating faults than diverging test cases.

Convergence Debugging

Convergence Debugging is a fully automated debugging method which measures the debugging effectiveness of a subset of the executed test set with respect to its convergence on the internal control and data of the failure-causing test case. The highly-concentrated testing involved in Convergence Debugging could be seen as the opposite of Diversity Testing. In particular, Diversity Testing is used to find dispersed faults in the program, while the test cases selected in Convergence Debugging are used to find closely-related test cases to the one that detected a fault. However, the converging test cases should also show diversity of control and data in the vicinity of the control and data involved in the failed test case, in order to expose related but distinct failures, and also, related but distinct successful executions. Without assuring diversity in the vicinity of a failure, the internal control and data of the program could be identical for all of the debug test cases. Such undiversified debug test cases would not provide any significant debugging information. The differences between the failure-causing test case and the diversified, converging test cases could then be used to analyze the root cause of failures.

For more information on Convergence Debugging go to convergence.