A joint study of the TU Delft and SIG meets the MSR Challenge of discovering interesting facts in historical software repositories. This case study shows that a number of straightforward instruments for visualisation and quantification can be used in concert to assess the efficiency of issue handling both at a high abstraction level and in detail.
Assessment of Issue Handling Efficiency
by Bart Luijten (TU Delft), Joost Visser (SIG), and Andy Zaidman (TU Delft)
Published in the proceedings of the 7th IEEE Working Conference on Mining Software Repositories (MSR 2010), May 2-3, 2010, Cape Town, South Africa, 2010.
Abstract: We mined the issue database of GNOME to assess how issues are handled. How many issues are submitted and resolved? Does the backlog grow or decrease? How fast are issues resolved? Does issue resolution speed increase or decrease over time? In which subproject are issues handled most efficiently? To answer such questions, we apply several visualization and quantification instruments to the raw issue data. In particular, we aggregate issues into four risk categories, based on their resolution time. These categories are the basis both for visualizing and ranking, which are used in concert for issue database exploration.