Keith Wall
Keith Wall
Anyone else been having similar ideas? Anybody think it is a good/bad idea? Could we stomach a pre-processing (InputDecorator like) based implementation? I'd would like to keep the implementation beneath...
> I'm intrigued by the assertion it needs to support numeric and boolean types, it is conceivable filters might been boolean and numeric typed configuration parameters. it may be useful...
> Whatever we implement we should only interpolate values, not keys, in objects. That is, we don't want to allow this, I think: > > ``` > ${foo}: bar >...
Closing, the operator will probably eclipse the need for this feature.
I was actually thinking about cases where the user creates the wrong sort of KEK in the KMS (say, a asymmetric key). The idea of this issue was to make...
My embryonic idea: if the Vault KMS encounters a 403 from a normal API call, it should probe the token using lookup-self. If the token is invalid, the error should...
> Do we want a separate KmsAuthenticationException (subclass of KmsException) to cover this? I wonder whether `KmsAuthenticationException` is quite the right abstraction. If you were authenticating using a TLS certificate,...
We touched on this issue at triage today. We reconfirmed a desire to get this done. @k-wall said he'd take a look.
Given the immaturity of metrics, I think I'd close my eyes to the API incompatibility that this change would make and just rename the existing labels.