An Eye Tracking Research on Debugging Strategies towards Different Types of Bugs

Abstract

Debugging is one of the important links in software quality assurance. Generally, the debugging performance of people adopting different debugging strategies varies enormously. Although there are studies discussing debugging strategies, little research analyzes the impact of these strategies towards different types of bugs. In this paper, the experiments conducted on 20 participants suggest that there do exist differences on the eye movement data of those successful and failed debugging samples. Specifically, concerning data flow bugs, it is beneficial to pay attention to the changes of variables, Nevertheless, it is more important to watch the code and understand their logical structure when dealing with control flow bugs. We believe it can help programmers find defects more efficiently by combining this conclusion and the error message provided by the compiler.