The effectiveness of using non redundant test cases with program spectra for bug localization

In this paper, we present our approach of using non redundant test cases with program spectra (one of the automated bug localization techniques) to locate software bugs in a program. We evaluate several spectra metrics (functions mapped from program spectra) using the non redundant test cases. Extensive evaluation on Siemens Test Suite and subset of Unix datasets shows the effectiveness of locating bug using non redundant test cases with program spectra. In this paper, we also show that by adding duplicates of non redundant test cases, the stability and performance of spectra metrics are affected.