Just wanted to share my thoughts about agenda and topics: I agree with Tim that everything that will be worked on is going to be interesting :-). IMO the core devs should leverage the time to their best use, so the rest of this post is to be taken as less important as the “main BinderHub” developments.
From my side, I want to build two projects with BinderHub:
- To make a web application from a research project (the o2r reference implementation) and API we have developed “Binder-ready” so that I have a scalable solution without reimplementing 80% of stuff that BinderHub already does, cf. https://github.com/o2r-project/nbo2rproxy and Enabling root execution in repo2docker
- Create a Binder-based workflow to power CODE CHECK, which hopefully boils down to making several features work with a BinderHub: https://github.com/codecheckers/binderhub/labels/breakthrough (tl;dr: save the image of an execution to a tarball and push that tarball along with the repo contents to a data repository, run an image imported from a Binder instead of building one with r2d)
[ For all of this I need to install my BinderHub on OpenStack, and am not sure I’ll manage to do that before the meeting. ]
I’d love to discuss “my topics” in person for just short bits of time if possible, or selfishly work on them a few hours and get some jump-starting from the experts. I’m also open to take the time and contribute big or small features! Here’s some that I came up with and that could be fun and completable:
- I’d really like more DOIs to be supported, e.g. https://github.com/jupyterhub/binderhub/issues/216 or https://github.com/jupyterhub/binderhub/issues/938 or b2share (API docs)
- Plaudit for Binders: https://github.com/jupyterhub/binderhub/issues/898
- What’s the current state of pinning a specific r2d version for a specific binder? (can’t find the issue, but relevant for reproducibility)