[netconf] Some clarifications would help

[Mahesh Jethanandani <[e-mail]>]

NETCONF WG,

The chairs met to discuss the poll that was initiated on transaction-id draft on the question of variance. Before we resume that poll, we felt some clarification around what is being proposed would help.

Going back to the presentation in IETF 117 on the transaction-id draft, it is unclear to the chairs, what the true benefit of etag + seq txid mechanism is. For instance, on slide 6 of the presentation, how is it that the Client NC Etag is not getting the representation similar to RC Etag? As we understand it, it is an opportunity for the server to return less than the full representation, like a diff, and hence the benefit is inversely promotional to the size of the diff.

The WG has over the years tried to keep RC and NC consistent. This is in part because some implementations build one on top of the other, or share a common infrastructure. The original scope of this draft was to have NC achieve feature parity with RC. If NC is to be better than HTTP, then RC should be similarly improved to maintain feature parity.

If the proposal to add Etag + Seq to RC, it is unclear if the HTTP specification would allow for it. See specifically discussions regarding Etag caching, and Weak versus Strong validation (https://datatracker.ietf.org/doc/html/rfc7232#section-2.1 <https://datatracker.ietf.org/doc/html/rfc7232#section-2.1>).

Thanks.

Mahesh & Kent (as co-chairs)