Solly Ross

/kind feature /priority important-longterm

We also need to come up with a reasonable fallback approach, and detection. We should *be* able to do a dry-run patch to check if server-side apply is supported (there...

cool, that's way better.

I think we may just have to explicitly fail if server-side apply is not available -- the chance for really subtle bugs may be too high.

/remove-lifecycle rotten /lifecycle frozen

FWIW, basic support is in right now -- it looks like ```go cl.Patch(ctx, obj, client.Apply, client.ForceOwnership, client.FieldOwner("my-controller")) ``` but that's a bit less elegant than it could be

I'm thinking the eventual interface should not need the patch type, and should just keep the field manager in some internal state so that you don't have to type it...

basically: ```go type Applier struct { patcher client.Patcher name string } func (a *Applier) Ensure(ctx context.Context, obj runtime.Object, opts ...client.PatchOption) error { // put the default options first so that...

(that's why I have the `name` field)

/kind feature I do kind-of like this idea. Want to put together a PoC?