Breaking out team update, feedstocks update, and feedstock listing update
See original GitHub issueIt is pretty silly we trigger a CI job to regenerate update teams, the feedstock repo, and the feedstock listing every time we merge here given what we are merging (pinnings, minutes, general docs, etc.) which have nothing to do with the feedstocks. We also then have to watch senseless build failures show up and get notifications about them; when, they are unrelated to the bulk of the content here. Not to mention we end up having to clone the feedstock repo in some cases to get the job done.
What we need is a better separation of concerns. The feedstocks repo should have CI and all of the feedstock info like this feedstock listing. As the feedstocks repo does not have the same activity as this one and that activity is strictly related to the feedstocks repo itself, this should cutdown on extra runs of these feedstock update actions when unintended. For now the relevant scripts can move over there, but we may want to break them out in the long run.
Initially it may seem concerning that part of the webpage would be moving (the feedstock listing). However, as we have seen by linking the status page in PR ( https://github.com/conda-forge/conda-forge.github.io/pull/210 ), linking of different repos webpages is seamless. We could simply turn feedstocks.html
into index.html
and thus ( http://conda-forge.github.io/feedstocks.html ) would become ( http://conda-forge.github.io/feedstocks ). In fact that latter link (sans .html
) works now. Thus this would actually work quite nicely.
In addition to these benefits, we would cutdown or CI usage because of lower activity. We would also no longer need a matrix for doing the update as all contents will live in the same repo. We would actually be able to use CI at this repo for anything we want as we will have freed up that resource. Our feedstock listing and feedstock submodules would remain in sync (unlike now). The overall complexity of our setup would be reduced and thus our management overhead will be reduced.
To summarize, we should break out feedstock maintenance stuff and move it to the feedstocks repo to simplify our workflow.
Issue Analytics
- State:
- Created 7 years ago
- Comments:9 (9 by maintainers)
Top GitHub Comments
The webpage listing has also been broken out now. The bulk of the work was in PR ( https://github.com/conda-forge/conda-forge-webservices/pull/142 ), but some follow-up PRs were necessary.
As such this is now totally complete. 🎉
Given everything else here is now done and the webpage listing is the only remaining task, which is already documented in issue ( https://github.com/conda-forge/conda-forge-webservices/issues/121 ), closing this issue out.