Validation: add documentation and use CEL pre-processor #3333
Conversation
istio-testing
added
the
size/L
|
@howardjohn: The following tests failed, say
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. I understand the commands that are listed here. |
| * **default**. Usage: `default(self.x, 'DEF')`. | ||
| * **oneof**. Usage: `oneof(self.x, self.y, self.z)`. This checks that 0 or 1 of these fields is set. | ||
| * **index**. Usage: `self.index({}, x, z, b)`. This does `self.x.z.b` and returns `{}` if any of these is not set. |
There was a problem hiding this comment.
Is the difference 'DEF' must be a concrete value whereas {} could be an expression of some sort?
There was a problem hiding this comment.
Both 'DEF' and {} are just examples of the same thing -- the default value. In one example I return a string and the other an empty object, but there is not a difference.
| // +kubebuilder:validation:XValidation:message="only one of WorkloadSelector or Endpoints can be set",rule="oneof(self.workloadSelector, self.endpoints)" | ||
| // +kubebuilder:validation:XValidation:message="CIDR addresses are allowed only for NONE/STATIC resolution types",rule="!(default(self.addresses, []).exists(k, k.contains('/')) && !(default(self.resolution, 'NONE') in ['STATIC', 'NONE']))" | ||
| // +kubebuilder:validation:XValidation:message="NONE mode cannot set endpoints",rule="default(self.resolution, 'NONE') == 'NONE' ? !has(self.endpoints) : true" | ||
| // +kubebuilder:validation:XValidation:message="DNS_ROUND_ROBIN mode cannot have multiple endpoints",rule="default(self.resolution, '') == 'DNS_ROUND_ROBIN' ? default(self.endpoints, []).size() <= 1 : true" |
There was a problem hiding this comment.
nit it really doesn't matter for this logic but just looks a bit odd that we default to something which isn't actually the default 'NONE'
Depends on istio/tools#3050.
This PR adds support for a new cel pre-processor: https://github.com/howardjohn/celpp.
This avoids the need to write incredibly verbose and error prone boilerplate to workaround limitations in the CEL language.
This PR has a few commits: