Steven Hawkins

It's unfortunately harder than it sounds - the code generation isn't really setup to understand things that are already defined by a base class. The class would also almost exactly...

> where the api provides an outputstream (exec in stream), we can additionally provide a non-blocking write method The options I see here are: 1. Provide a blocking / non-blocking...

@lishalongfa that is not the same error. That will happen on watches occasionally if the server lacks bookmark support. You are supposed to handle this situation in the Watcher.onClose(WatcherException) method...

@lishalongfa please refer to https://kubernetes.io/docs/reference/using-api/api-concepts/ to learn more about how watches may terminate. Also please convert any further questions to a separate discussion topic so they are not associated with...

Relates to #3266

@adriansuarez ideally we'd like to not make much more of an investment in the crud mock server, and instead have users rely upon minikube, or other small k8 distribution, or...

> So I don't think switching to jenvtest would solve my problem That's unfortunate. > but maybe the right approach is to rewrite it as a controller, which would be...

> I'm unsure if/how this is going to impact the underlying REST API calls, maybe @shawkins already has an idea about it. listKind already defaults to kindList. I'm not sure...

> I don't think this is used often, but it might impact some edge cases: https://github.com/kubernetes/kubernetes/issues/87133 After looking things over more I think it's just for resolving kind conflicts -...

> BTW: overall I still think that the ability to specify watch namespaces per controller is introducing a horrible amount of complexity for cases which are not really used in...