HASEL members, Haris Mumtaz and Kelly Blincoe, in collaboration with Carlos Paradis (University of Hawaii), Fabio Palomba (University of Salerno), Damian Tamburri (JADS), and Rick Kazman (University of Hawaii), recently studied the usage of a GitHub feature and its relationship with social aspects of software teams.
GitHub is a de-facto choice of millions of software developers because it offers features that assist collaborative software development. One of the features is “assign issue to issue commenters”. This feature aims to add new contributors to the project to help the existing development team. When new contributors are added to a project, it could change the social structure of the team because of their experience and expertise. However, how do we know if software teams find this feature useful or not? What are the indicators that suggest the feature’s usefulness? How do we know if the social structure of software teams is changed for good or for worse?
Our study investigated the adoption of this GitHub feature (assign issue to issue commenters) by examining the feature’s relationship with socio-technical aspects. We used social smells as the indicators that reflect the sub-optimal socio-technical patterns in the organizational and communication structure.
To achieve our objective, we empirically analyzed 13 open-source Apache projects. We collected feature usage data using GitHub’s APIs to filter the contributors who started contributing to the projects through the usage of this feature. In addition, we computed social smells for a period of two years (one year before the feature’s initial use and one year after) using the Kaiaulu tool. To analyze whether the feature was adopted by the project’s community or not, we calculated the ratio of the issue assignments achieved using the feature to the total issue assignments. To examine the feature’s relationship with socio-technical aspects, we employed statistical t-test and temporal analysis (using pre- and post-feature social smells).
We found that the usage of the GitHub feature varied across the analyzed projects. For instance, some projects only used the feature in a quarter of the post-feature year, while some projects used the feature more frequently (i.e., using it throughout the post-feature year). Overall, our empirical findings suggested that the social structure of software teams has improved in the projects that used the feature frequently, indicating that the feature could be useful for software teams. Our t-test showed that the social structure of the teams changed significantly after the feature adoption. Moreover, our temporal analysis showed that the social structure of software teams improved in the post-feature year.
The results from this study were accepted and presented at the 15th International Conference on Cooperative and Human Aspects of Software Engineering (CHASE) in May 2022. You can read the full paper here.