Re: [Doh] signalling client expectations and server capabilities

["Mark Delany" <[e-mail]>]

On 02Jun19, Jim Reid allegedly wrote:
> 
> 
> > On 2 Jun 2019, at 16:54, Joe Abley <[e-mail]> wrote:
> > 
> > In the very general sense (ignoring protocol specifics relating to the DNS and HTTP) it feels like it might be handy to be able to specify a capabilities requirement with a query. Something like
> > 
> >  - I want an answer to the question PIR.ORG/IN/SOA with RD=1
> >  - I want query minimisation and secure transport for recursive queries
> >  - I want you to confirm explicitly that you understand the request for query minimisation

Yes, I had thought about mentioning capabilities but I doubted anyone would have
read below the fold of my email :-)

One thing I wondered about is how well capabilities work in a multi-hop and
multi-transport system that has DoH in the path. In the very general sense
should a stub resolver asking for capabilities have said query punch all the way
through to an auth server?

There is already a smattering of adhoc capability signals in DNS such as TC=1
and maximum response size which probably should have received some attention in
8484 so if nothing else maybe formalization will at least raise the question for
future protocol designers: "how do we deal with capabilities?".


> Hmmm. The add list might be a more appropriate place to discuss this. Or maybe dnsop. It???s not clear if these sorts of capabilities could/should be signalled in DoH or vanilla DNS.

It's certainly a big topic and probably risks the second-system syndrome.


Mark.