Unusual Events in GitHub Repositories

TL;DR

This paper investigates which unusual events in GitHub repositories are most useful for developers to monitor, aiming to reduce information overload in active software projects.

Contribution

It identifies specific unusual events that are considered particularly useful by developers, providing insights to streamline project activity awareness.

Findings

Large code modifications are highly useful to developers.

Unusual reviewing activity attracts significant attention.

A subset of unusual events can effectively inform developers about important project changes.

Abstract

In large and active software projects, it becomes impractical for a developer to stay aware of all project activity. While it might not be necessary to know about each commit or issue, it is arguably important to know about the ones that are unusual. To investigate this hypothesis, we identified unusual events in 200 GitHub projects using a comprehensive list of ways in which an artifact can be unusual and asked 140 developers responsible for or affected by these events to comment on the usefulness of the corresponding information. Based on 2,096 answers, we identify the subset of unusual events that developers consider particularly useful, including large code modifications and unusual amounts of reviewing activity, along with qualitative evidence on the reasons behind these answers. Our findings provide a means for reducing the amount of information that developers need to parse in order to stay up to date with development activity in their projects.