Usability is one of the software qualities attributes that is subjective and often considered as a less critical defect to be fixed. One of the reasons was due to the vague defect descriptions that could not convince developers about the validity of usability issues. Producing a comprehensive usability defect description can be a challenging task, especially in reporting relevant and important information. Prior research in improving defect report comprehension has often focused on defects in general or studied various aspects of software quality improvement such as triaging defect reports, metrics and predictions, automatic defect detection and fixing. In this paper, we studied 2241 usability and non-usability defects from three open-source projects-Mozilla Thunderbird, Firefox for Android, and Eclipse Platform. We examined the presence of eight defect attributes-steps to reproduce, impact, software context, expected output, actual output, assume cause, solution proposal, and supplementary information, and used various statistical tests to answer the research questions. In general, we found that usability defects are resolved slower than non-usability defects, even for non-usability defect reports that have less information. In terms of defect report content, usability defects often contain output details and software context while non-usability defects are preferably explained using supplementary information, such as stack traces and error logs. Our research findings extend the body of knowledge of software defect reporting, especially in understanding the characteristics of usability defects. The promising results also may be valuable to improve software development practitioners' practice.
|Number of pages||8|
|Journal||International Journal on Advanced Science, Engineering and Information Technology|
|Publication status||Published - 2020|
- Defect report
- Software defect repository
- Software repository mining
- Usability defects