What happens in Triage?: An Empirical Study of Bug Triage in Software Product Evolution?

This paper investigates and evaluates the role of bug Triage in software evolution and maintenance. Traditionally, Triage decision-making has been based on bug reports. Decisionmaking concerns whether a bug is to be fixed and, if so, when and by whom. Research in this area focuses on automation of some aspects of bug fixing, enhancing information on bugs reports, and most significantly, automating Triage through Machine Learning (ML) techniques. Our paper is based on an ethnographic study of a software team, and includes Triage and Stand-up meeting
observation, analysis of bug report documents, study of the development environment and ad-hoc meetings. The framework of Distributed Cognition for Teamwork served as a theoretical lens for this study. Based on the analysis 33 complex bugs, the paper argues that Triage was used as a major information hub for discussing a wide range of information (e.g. about organisational processes and development of the software), allowing knowledge development that is valuable in software evolution beyond bug fixing alone.