Ryan Zhang

I think these two cases are actually different. In the first "serviceTrait" case, the trait can discover the labels from the workload it attaches to. In the second "ingress" case,...

Agree With @captainroy-hy, using a trait to determine the behavior of a workload controller is the designed behavior. In addition, I don't know what to put in the childResources fields...

> But it seems no need to touch workload for now. Actually, I think there is real need to touch trait for now but we do need to touch workload....

I wonder why patch can't do "remove"?

just to be clear, the real problem is not just to remove the fields that are not present in the new CR but also preserve the fields that are not...

> @ryanzhang-oss Right. > > Let me add another example: > > ``` > # user applies > auto: false > rules: [a, b, c] > > # controller adds...

I am not sure if this is compatible with the current structure. as the workload itself is already declared as a RawExtention. It might be doable but pretty tricky and...

After further discussion, we have decided to take the format below but keep the API structure. So we will just parse the rawExtension as is and if there is a...

> ```yaml > workload.oam.dev/name:

Thanks, Glyn, I wonder if you can submit a PR?