Understanding the Characteristics of Visual Contents in Open Source Issue Discussions: A Case Study of Jupyter Notebook

TL;DR

This study analyzes visual content in open source issue discussions, revealing their types, purposes, and impact on discussions, with implications for improving issue tracking systems.

Contribution

It provides the first empirical characterization of visual contents in OSS issue discussions, identifying types, purposes, and their effects on discussion dynamics.

Findings

Over 25% of issues include visual contents.

Visual contents in comments correlate with longer discussions.

Approximately 60% of visual contents are screenshots or mockups.

Abstract

Most issue tracking systems for open source software (OSS) development include features for community members to embed visual contents, such as images and videos, to enhance the discussion. Although playing an important role, there is little knowledge on the characteristics of the visual contents to support their use. To address this gap, we conducted an empirical study on the Jupyter Notebook project. We found that more than a quarter of the issues in the Jupyter Notebook project included visual contents. Additionally, issues that had visual contents in the comments but not in the issue posts tended to have a longer discussion and to involve a larger number of discussants. In our qualitative analysis, we identified eight types of visual contents in our sample, with about 60% including screenshots or mockups of the main product. We also found that visual contents served diverse purposes, touching both problem and solution spaces of the issues. Our effort serves as an important step towards a comprehensive understanding of the characteristics of visual contents in OSS issue discussions. Our results provided several design implications for issue tracking systems to better facilitate the use of visual contents.