/cc @apelisse @jiahuif @alexzielenski

So you're using the test code to make a CRD and the CRD is invalid? From a very cursory look, that seems normal and expected, but the examples included in the tests should probably be valid. Alternative, we could also detect if the default is not part of the enum at generation time, and fail at that time, that'd probably be better.

I'm using code generated by kube-openapi with an enum to make a CRD and the CRD can't be made for having an invalid default value, because the default zero value is not one of the enum values. I didn't provide the default value and the generator chose a zero value based on datatype. The example included in tests is not valid for the same reason, and that's why I showcased it using the example in tests.

Looked at it again. It looks like the value is REQUIRED anyway, the default is kind of forced because the field is not a pointer, so I would say this works as expected. There's not really any other great way to do that.

Uhm I worked around it by removing the Default: "", line and it worked. Is this "REQUIRED" not enforced?

Yeah, that's a little weird I agree. Also not entirely sure depending on your exact situation. (though required should ALWAYS be respected tbh).

Can we have the default value use the first enum value then?

Looked at it again. It looks like the value is REQUIRED anyway, the default is kind of forced because the field is not a pointer, so I would say this works as expected. There's not really any other great way to do that.

IMO the schema is invalid if the default value wouldn’t pass validation itself. The schema generator should throw an error for that

yeah, I don't think that's wrong. I'm not exactly sure why we have a default here, I need to double check.

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

/remove-lifecycle stale

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

/remove-lifecycle stale