Abstract

An issue report describes a bug or a feature request for a software system. When resolving an issue report, developers may discuss with other developers and/or the reporter to clarify and resolve the reported issue. During this process, questions can be raised by developers in issue reports. Having unnecessary questions raised may impair the efficiency to resolve the reported issues, since developers may have to wait a considerable amount of time before receiving the answers to their questions. In this paper, we perform an empirical study on the questions raised in the issue resolution process to understand the further delay caused by these questions. Our goal is to gain insights on the factors that may trigger questions in issue reports. We build prediction models to capture such issue reports when they are submitted. Our results indicate that it is feasible to give developers an early warning as to whether questions will be raised in an issue report at the issue report filling time. We examine the raised questions in 154,493 issue reports of three large-scale systems (i.e., Linux, Firefox and Eclipse). First, we explore the topics of the raised questions. Then, we investigate four characteristics of issue reports with raised questions: (i) resolving time, (ii) number of developers, (iii) comments, and (iv) reassignments. Finally, we build a prediction model to predict if questions are likely to be raised by a developer in an issue report. We apply the random forest, logistic regression and Naïve Bayes models to predict the possibility of raising questions in issue reports. Our prediction models obtain an Area Under Curve (AUC) value of 0.78, 0.65, and 0.70 in the Linux, Firefox, and Eclipse systems, respectively. The most important variables according to our prediction models are the number of Carbon Copies (CC), the issue severity and priority, and the reputation of the issue reporter.

Bibtex

@article{huang2018empirical,
  title={An empirical study on the issue reports with questions raised during the issue resolving process},
  author={Huang, Yonghui and da Costa, Daniel Alencar and Zhang, Feng and Zou, Ying},
  journal={Empirical Software Engineering},
  pages={1--33},
  year={2018},
  publisher={Springer}
}