Since cljdocs inception I've always been very eager to make it easy for folks to contribute. I wrote almost two dozen Architecture Decision Records (ADR) hoping it would help future contributors navigate the codebase. I encouraged people to contribute and tried to spec out smaller "good first issue" type contributions so folks could start with something small.
Once people started contributing I of course wanted to encourage them to stick around and help maintain the project. Since there's so many parts to maintaining a project like this even small non-code contributions like participating in a discussion or reviewing a PR can be tremendously helpful.
To give a sense of ownership I added anyone who contributed as a contributor on GitHub and gave them push access to the repository. The understanding was that we'd still use PRs for most contributions but it did remove me as a primary gatekeeper. Any previous contributor is empowered merge a PR. I think the first time I read about this approach as in Felix Geisendörfers Pull Request Hack article.
As the project grew we added continuous deployment. Any commits to the main branch are automatically deployed. It might seem a bit wild to just let anyone deploy to cljdoc.org but it works for this project.
cljdoc isn't a library. It doesn't have an API. We can just let people improve it in ways they see fit. Sure maintainability is still a concern, but that is much less challenging than maintaining backwards compatible APIs or just designing a good API in the first place. Historically I think most contributions have actually improved the codebase.
I'm really happy I'm not maintaining a popular library with lots of contributors. I'm helping maintain a thing where people can play around and explore their ideas. Where it's ok to make a mistake. We're all here to learn. I learned so much from creating cljdoc and I couldn't be happier that others can do the same.
Shoutout to Lee Read who prompted these reflections. 🙌