Fault localization using execution traces
Margaret Ann Francel, Spencer Rugaber · 1992
This paper describes an approach to the automatic localization of program faults using execution traces. The traces are used to construct a directed graph that models the propagation of values during execution. Knowledge of which output values are incorrect is then used to narrow the region that must be examined by the person debugging the program. Algorithms to construct and analyze the graph have been designed and implemented.