Program-delta oriented debugging supporting method DMET

Abstract

SUMMARYResearch has been conducted on methods of automatically performing tests that focus on the delta (difference) between a revision that is known to function normally in advance and a revision that contains defects to identify the causes of the defects. However, since software had to be created from source programs for each test, an extremely long test time had been required. Also, since the emphasis had only been on the testing tasks and no consideration was given to debugging tasks, these methods were not practical. Therefore, in the current research, the authors propose the DMET debugging technique, which can be used for actual software maintenance. DMET, which reduces the test execution time compared with conventional techniques, supports the entire sequential processing flow from testing to debugging. Also, to verify the effectiveness of the proposed technique, the authors implemented the DSUS development support environment, which uses DMET, and performed comparison experiments. From the results, it was apparent that using DMET enabled the total debugging time to be shortened.

An overview and case study of a statistical regression testing method for software maintenance

“…DMET [26] proposed by Matsushita and colleagues is one debugging supporting method that could be applied to statistical regression testing. According to DMET, a failure found by the statistical regression testing method implies that there is a high probability that the cause (the bug) exists in the delta source code between the base software and the update software.…”