Patrick Ohly
Patrick Ohly
I ended up faking the "pass" in the affected benchmark by emitting the output that `test2json` turns into "pass": https://github.com/pohly/kubernetes/commit/9099f094d45b5ea6850fdf94d7df0af19e6d7a46 This is not a proper solution. I'm still not sure...
/remove-lifecycle stale /lifecycle frozen
`CompileCELExpression` can probably collect attribute names by walking over the AST, but I don't know how. So yes, definitely needs some exploration.
> > Detect attribute or capacity domains and/or identifiers which don't match the validation constraints for those. Could be done by analyzing the AST of the CEL expression. > >...
> Is there a the DNS subdomain check while creating the resources(if that is the name of other resources, it might exist while API validation) or is it guaranteed that...
> I'm assuming scheduling time events/in the pod resource is too late, is the idea to have them happen when putting in a bad selector for ResourceClaimTemplate? It cannot be...
> I guess the question is do we need extra event/status for problem report? That depends on user experience based on the kind of mistakes they make and what information...
/test pull-kubernetes-apidiff Do we care enough about "controller-runtime continues to build" that we should try that in the `pull-kubernetes-apidiff` job? It's advisory, so it would be okay if it fails...
/test pull-kubernetes-apidiff Now with trial build against controller-runtime :smile: