An improved static program slicing algorithm using stack trace

Abstract

In order to improve the accuracy of the static program slicing which is used to locate the faults that cause an exception, we propose a new approach that uses the stack trace which is stored when an exception occurred. The proposed approach consists of three steps. First, it uses stack trace to infer the execution trace and figure out those methods and statements that have not been ever executed. Then, these methods which are not executed will be ignored when building the system dependence graph. Finally, the accurate slice will be got by adopting the improved program slicing algorithm. We perform an empirical study of our new approach by using four reported bugs of open-source project Ant. And the result shows that using our approach the slice is 8 percent less than using the general static program slicing algorithm on average.