Rethinking the bug tracking paradigm

Abstract

In modern development processes, the bug tracking tool is one of the most important means of tracking what tasks, features or bugs are currently active in a project. Such information is crucial, not only for maintaining developer workflow, but also to make informed decisions as to whether a product should go live in its current state. When a bug is identified, the reporter creates a report using a bug tracking tool. Bug reports typically include general information such as a summary of the issue, its expected due date, and a description. Once the report is created, teams can rank and prioritise it as they see fi t, and assign it to individuals for resolution. The main issue with this approach to tracking project progress is that the quality of bug reports can vary based on the person reporting the bug, their knowledge of the system, their cognitive workload, and other contextual factors. In this project, we set out to study and contribute an improvement to prevalent issues with bug tracking systems. This was done in two phases. Firstly, we reviewed the literature in order to gain insights about prevalent issues with bug-tracking. We also backed up this literature review with a survey amongst practitioners to identify what the main complaints were with regards to bug tracking systems. Results showed that most critical issues, according to the participants, were the lack of important information in bug reports, and miscommunication between bug reporters and developers. With this in mind, we designed and implemented a Jira plugin which departed from the usual 'compulsory field' approach to improving bug quality, and instead analysed bug description content in real time to provide recommendations for improving what the reporter was writing. As a visual aid, a suitability metre (akin to password strength metres) was also developed. This provides the reporter with an overall rating indicator of the quality of the bug report currently being created. The plugin was trialled by practitioners, and the resulting bug reports were reviewed and rated by a panel of two experts, who blindly rated a mixture of bug reports, 50% of which were written without the aid of the plugin, and 50% of which were written with the aid of the plugin. The expert review results indicate that the plugin motivated the authoring of objectively better bug reports, rated in terms of completeness, usefulness and overall quality.