Research On Automated Program Repair Based On Error Propagation Chain

As software become more complex,software defects are inevitable.In the event of a software error or software failure,fast and effective maintenance is required to detect and repair errors in time to reduce the loss caused by software failure.Automated program Repair(APR)can help maintenance personnel to fix program errors.At present,researchers have proposed many automated program repair methods.Firstly,we analyze the current situation of automated program repair and fault localization,and then analyze the challenges and problems of them,and focus on the feasible methods to improve the automated program repair efficiency.Then,we manually analyze the defect programs,and the error propagation chain(EPC)of the defect program is obtained by using the dynamic slicing technology.By analyzing the error propagation chain of the defect programs,we find that the scope of error propagation for most programs is a single class.So we can only consider the statements in the failure class as suspicious statements in fault localization.Finally,we design and implement the error propagation chain(EPC)-based repair system EPCNopol,which implements the error propagation chain-based fault localization method in the automated repair tool,and conducts experimental research on eight typical cases in Defects4 j.Compared with the experimental results of SFL-based Nopol,it shows that the method can effectively improve the automated program repair efficiency.The automated program repair can effectively help maintenance personnel to carry out software debugging,so the efficiency of the automated program repair can further reduce the loss caused by software defects.This paper proposes an new automated repair method based on error propagation chain,which can effectively improve the efficiency of automated program repair.