Ebben Aries

Honestly I think things got a bit murky over time wrt. the `interface-id` typedef cc: @aashaikh as the original commit went in > 8 years ago on 03/30/2016 From history,...

> It should be possible to name the key anything that one wants This defeats the purpose of the typedef and the key should then just be an opaque user-defined...

@dplore @gcsl - PTAL

While I understand the problem stmts that leaf-lists impose with the current gNMI operations, I can't help but feel we are breaching the boundaries of the RPC operation intent by...

Some additional consideration and commentary: This is an example of a seemingly subtle set of wording changes in a spec w/ no modification to the proto IDL. The spec is...

> bumped to v0.11.0 -- agreed that since this can cause breaking change in existing implementations due to the ambiguity clarification. The more I think about the cases where there...

Few generic comments/questions * Should verbiage/expectations be added around reboots surrounding image upgrades, ZTP re-instantiation, etc.. where such values may not be retained? * For POSIX compatible systems, see influence/structure...

I don't think (1) is feasible for the reasons already mentioned. The identities are tied to physical ethernet interface speeds and creating multiples to cover all aggregate use-cases is a...

Submitted https://github.com/openconfig/public/pull/1183 to put restrictions around physical vs. aggregate for ethernet related config/state

> The annotation that is used in OpenConfig models is [`telemetry-atomic`](https://github.com/openconfig/public/blob/master/release/models/openconfig-extensions.yang#L154-L165). > > `telemetry-on-change` is used to indicate that the data model expects that this value is sent as `ON_CHANGE`...