Re: [Doh] [Ext] New version: draft-ietf-doh-resolver-associated-doh-03.txt

[Paul Hoffman <[e-mail]>]

On Mar 25, 2019, at 4:05 PM, Ralf Weber <[e-mail]> wrote:
> I have some questions after reading -03 (and skipping -02 ;-)
> - The resolver IP addresses returned from DNS on section 4 are these Do53 server addresses or DoH server address or are they just the IP to start the process in section 2 (maybe 3)?

They are the IP addresses of the Do53 resolver that the computer is using. I'll make that clearer in the next draft.

> - On section 3 a non compliant current server will return NXDomain. What are we going to do with this answer (treat is as no DoH associated)?

Yes, that was the intention. I'll clarify that.

>> As for the late discussion of using the URI RRtype instead of TXT, I would not know what to put in the "priority" and "weight" values. That alone seems enough reason to leave this as a TXT record, but others might disagree. It's not a lot of effort to change the text to the URI RRtype, but I don't want to do so unless it is actually better than TXT.
> Another option could be a new RRType, given that it doesn’t have to be provisioned on auth servers it should not be that difficult to roll it out. We still could use the special name or just ask .

I have been told that some OSs have "send a TXT request" API call, but I have not verified that. If so, then TXT records will have an advantage. Otherwise, using a new RRtype is cleaner.

Can someone who from the application implementation world verify or refute my hazy memory of API calls for TXT records?

--Paul Hoffman