Release behavior of Open Source projects

In this article we are taking a look at the release behavior of open source projects. The underlying data is coming from the 1000 most popular GitHub projects (by stars). Only projects which are using the GitHub release feature are considered, because git tags do not necessarily indicate a release.

Furthermore, GitHub releases are only considered when they were published before the 1st of January 2020 and the projects issued at least 5 releases before that time.

After applying these constraints, only 529 projects out of the 1000 most popular GitHub projects made it into the data sources for this article.

To get a better idea of what this means in practice, let’s start by taking a look at the 25 projects with the most releases.

In which months are releases happening?

When grouped by months, a trend at releasing towards the end of the year becomes visible. Things start off slowly and are set back during summer and winter holiday seasons.

At which weekdays are projects publishing releases?

Contrary to my initial fear, the majority of the work appears to be happening within the weekly work cycles and much less so on the weekends.

At which days of the year are releases published?

This chart is probably not as meaningful as the previous ones, but this is mostly due to the low amount of data. It does, however, nicely show the drop of activity during winter holidays.

At which time of the day are projects being released?

This chart is not particularly meaningful, because the timestamps from the GitHub API do not include timezone information. So these times do not reflect the local time of the project authors, but probably only the UTC time of the releases.

You can view the source code used to generate the diagram data at taig/github-release-stats