conda-forge.github.io

conda-forge.github.io copied to clipboard

There have been repeated attempts at elaborating a list of "core ideas", "required improvements", "conda-forge wishlist" to give the community a better insight on what parts of the conda-forge ecosystem requires work.

These repeated attempts in my opinion signal that we were not able to find a model to document these things effectively with low-effort. I'd like to propose a couple of alternatives to see if we can get better at this. One of the motivations would be to let funding sources find actionable items easily. PSF has this repo, but I think that model won't work for us.

My proposal is more less based on using issues instead of documents or repositories.

For new and existing items that are not in progress:

• Use Github issues (with a template, if needed) to document fundable items.
• Label them as necessary, or add them to a milestone so they are easy to find.

For in-progress tasks:

• If a team or person is working on that task (even if partially), assign the Point of Contact person to the issue.
• Minimal report of advances made.

Once done or obsolete:

• Close the issue.
• OR: Report if other actions items are required, possibly in a new issue with further details.

---

This assumes the issues will be in this repository. Oftentimes, there's key work documented in other repos (infra, key feedstocks, etc). I'm not sure how to deal with those. Maybe:

• Create a "mirror" issue in this repo, linking to the other one. Probably will get out of sync at some point.
• Use a project board to link issues across repositories (maybe with the same template?). This will require enabling project boards in conda-forge. I'm not sure how we can handle permissions there because feedstock maintainers might be able to edit those. I need to double check.

---

Tangentially related: all funding streams that mention conda-forge as the target organization should be documented in the funding docs page, for transparency. I don't think we are currently enforcing this, but I wonder to what point we should and what the feelings are there.

cc @jakirkham because I think you brought this up a couple core meetings ago and I said I'd follow up. This is my crude attempt at getting the conversation started.

Label them as necessary, or add them to a milestone so they are easy to find.

Org-wide issue search with labels works pretty ok. We can use the label:fundable query and it would find all of them; e.g. this is how it looks for label:maintenance.

If we enabled Project boards in the organization we can also have some automation to have them added automatically, but this has proven tricky in the conda org because every repo needs that workflow and... we have way too many repos 😂 So I think labels it is.

I opened https://github.com/conda-forge/.github/pull/506, if someone wants to take a look! If I don't hear back, I'll merge by next week.

And https://github.com/conda-forge/conda-forge.github.io/pull/2296 for the docs.