ApplicationCRD is indeed pretty new so we have no doc for it for now, please feel free to check its type structure here: https://github.com/oam-dev/kubevela/blob/master/apis/core.oam.dev/v1alpha2/application_types.go
We are experiencing timing issues with the application of changes resulting from trait and scope processing to application resources. In particular we have an example of workload type of which known customer applications can take several minutes to start or restart. The implementation model suggested by the oam-kubernetes-runtime is to update the application’s workload resource or its eventual children resources after they are created. This is causing the application to be restarted since the operator typically begins processing the resources when they are first created and may then need to restart them when a change is later detected. In addition we cannot guarantee that a user’s application will not lose data if it is restarted unexpectedly.
As a result, having an OAM standard way to apply changes for traits and scopes to workloads before the workload resources are actually created will be important for us.
How can this be done?