Nic Cope
More prior art: https://docs.fedoraproject.org/en-US/council/policy/ai-contribution-policy/
With server-side apply Crossplane patches every composed resource on every reconcile. This is expected behavior - we don't check whether the resource needs updating and only patch it if it...
> Nothing is changing - all patch events are "empty" - but still there I understand that the XR is making no-op patches with no changes. What I'm trying to...
Yeah, I was trying to fix this recently. It's a similar issue to https://github.com/crossplane/crossplane/issues/6761, where the new provider dependency conflicts with the existing provider dependency over ownership of resources. I...
We (@upbound, my employer) also have a test implementation: https://docs.upbound.io/manuals/cli/howtos/testing/#test-your-project-locally It's focused around a concept called "control plane projects" though. A control plane project is an on-disk "software project" that...
Seems reasonable to me. Off the top of my head I think this'd mostly be a case of updating the real and render fetchers. As well as the protobufs and...
This proposal has a bunch of implications, e.g.: * Would we have a provider package up a bunch of Managed Resource Definitions (MRDs) instead of CRDs? * Would we need...
> Is there a reason we need type-level differentiation rather than labeling, other than RBAC point above? It's mostly RBAC. The only other potential pros I could think of are:...
Dropping another note here - while there are many other ways to skin this cat, I could see an MRD being a good place to add additional Crossplane specific metadata...