Re: [core] Responses to NON with 4.02

[Achim Kraus <[e-mail]>]

Hello Christian,

I think, this discussion from April 2020 is at least related to your
question.

https://mailarchive.ietf.org/arch/msg/core/wd4LleVnnvt2WOXj28ouNZcCTXU/

best regards
Achim Kraus

Am 16.02.21 um 21:29 schrieb Christian Amsüss:
> Hi,
>
> reviewing new-block told me something new about RFC7252, that is that a
> response to a NON request containing an unknown critical option MUST be
> either RST or complete silence (whereas in the CON case, 4.02 is the
> logical response).
>
> Why is that? A NON 4.02 response would be just the same size, give
> better information, and (to me, most importantly) would not require the
> additional case distinction between "it's a 4.02 because some option
> value doesn't make sense" and "it's a 4.02 because I don't know the
> particular option" at a very different place than were the options are
> understood.
>
> (Or worse yet: A NON may arrive at a proxy that forwards it over TCP or
> as a CON, and by the time the proxy gets the 4.02, it doesn't know
> either).
>
> And will anything go Really Bad if I ignore that MUST? (I currently do
> -- up to now for ignorance, going forward because my libraries all work
> under the fiction that the library is an intermediary, and thus may do
> all things an intermediary may do).
>
> BR
> c
>
>
> _______________________________________________
> core mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/core
>