Update JSON_ETF encoding rules in gnmi specification#238
Update JSON_ETF encoding rules in gnmi specification#238
Conversation
Added emphasis that JSON_IETF must use namespace qualified names where required as we observe a variety of implementations do not strictly adhere to section 4 of RFC7951.
| encodings. | ||
| encodings. Also note, data must be encoded using namespace | ||
| qualified names where required per | ||
| [RFC7951 Section 4]([https://tools.ietf.org/html/rfc7951](https://datatracker.ietf.org/doc/html/rfc7951#section-6.8)). |
There was a problem hiding this comment.
Agree but my only comment would be - is this necessary?
The encoding is JSON_IETF and that encoding should just dictate conformance to RFC7951. Is there any case where RFC compliance should be diverged here?
It's been a while since I read this entire section (will go back and read) but if it isn't already we should cover what the expectation is for prefix encoding as well w/ this value encoding when it comes to RFC7951 compliance
There was a problem hiding this comment.
Strictly speaking, this reminder in the doc should not be necessary since RFC7951 requires it and OC requires RFC7951. But it's observed there is not compliance with this, even within OpenConfig's own published tools. Given the miss, a little redundancy in the docs seems warranted.
FWIW, I did some searching within OpenConfig repositories and do see this open issue for ygot which is still accurate/unresolved: openconfig/ygot#107
There was a problem hiding this comment.
Ah, nevermind, ygot does implement prefixing:
https://github.com/openconfig/ygot/blob/98713a6f0eb19805b91a4fddec2fa6b2c152e71d/ygot/render.go#L1102
3 participants
Added emphasis that JSON_IETF must use namespace qualified names where required as we observe a variety of implementations do not strictly adhere to section 4 of RFC7951.