Re: [Ohttp] Discovery

[Eric Rescorla <[e-mail]>]

On Fri, Jun 25, 2021 at 1:50 AM Andrew Campling
<[e-mail]> wrote:

>
> Moving back to the topic of discovery, experience with DoH highlights the
> dangers of a working group only completing part of the work where the lack
> of a discovery mechanism has impaired take up.  There was a delay of around
> 18 months before the ADD working group could be chartered and work started
> on discovery; repeating that approach here seems both unnecessary and
> inefficient.


At the risk of relitigating a discussion which has already been discussed
ad nauseum. I do not agree with this assessment.

To the contrary, I think the ADD process ought to lead one to the
conclusion that in the current setting discovery for DoH is of relatively
limited applicability. Having attempted to solve it before standardizing
the transport would have delayed the fairly substantial amount of
deployment we are already seeing.


Why not include discovery in the scope of the charter so that the working
> group can consider whether a mechanism is required and, if so, provide a
> suitable solution?
>

There are two kinds of discovery that one might imagine being in scope:

1. Discovery of proxies.
2. Discovery that a server is O-HTTP capable

The former is extremely difficult, for many of the same trust reasons that
DoX discovery is difficult, but even moreso. The second is cleanly
separable. Given that there is real value in the protocol with neither form
of discovery, I don't think it's useful to charter them, especially in the
absence of any real proposals. If someone comes up with such a proposal, we
can consider a recharter then.

-Ekr