Re: [dnssd] [DNSOP] QDCOUNT > 1 (a modest proposal)

[Abtin Keshavarzian <[e-mail]>]

Hi,

I think the proposal is to fully prohibit "QDCOUNT > 1" and require
resolvers to always reject such queries.
Is my understanding correct?

When there is ambiguity, it may be better to lean towards a more
accommodating clarification.

Ted mentioned the main use-case:
- With Thread (and Matter) more small IoT embedded devices can act as
client performing DNS-SD service resolution.
- These can be battery powered and resource-constrained sensors (limited
memory and computational power, low bandwidth and sleepy).
- For service resolution, they need to get SRV and TXT records for a
service instance name (most likely a local service).

It would be good to allow resolvers to respond to queries containing
questions for SRV and TXT for the same name.
This can be optional (not required but also not disallowed) and/or limited
to local names.

Without this, we require multiple message exchanges (adding inefficiencies)
on a less capable client device to accommodate a far more capable resolver.
Is there any concern with this (that I may be missing)?

Thanks,
Abtin.

On Thu, Feb 16, 2023 at 9:00 AM Ray Bellis <[e-mail]> wrote:

>
>
> On 16/02/2023 11:49, tjw ietf wrote:
>
> > This is a great idea Ray and the chairs look forward to your draft
> submission
>
> Joe Abley and I are going to write the "QDCOUNT is 1, dammit!" draft.
>
> I've also just revivied draft-bellis-dnsext-multi-qtypes, but since
> DNSEXT doesn't exist any more (boo!) I'm open to suggestions as to
> whether the draft should go to DNSOP or DNSSD.
>
> Ray
>
> _______________________________________________
> dnssd mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/dnssd
>