Software Fault Localization Based on Testing Requirement and Program Slice

A heuristic approach is proposed to locate a fault according to the priority. To a given test case wt, fault localization has to be proceeded when its output wrong. Firstly, four assistant test cases, one failed and three successful test cases, are selected out according to the biggest cardinality of Req(wt,t_i), which stand for the common testing requirements both covered by wt and t_i. Then, code prioritization methodology is put forward based on program slice technique. Dynamic slice technique is taken for wt and execution slice technique for four assistant test cases. Some dices are constructed with different priority which means the possibility of containing bug and is evaluated according to the occurrences in the selected slices. Thirdly, the key algorithm including two procedures, refining and augmenting, is followed here to fault localization based on priority. In the refining phase, the most suspicious codes am checked step by step; in the augmenting phase, more codes will be gradually considered on the basis of direct data dependency. At last, experimental studies are performed to illustrate the effectiveness of the technique.