On Jan 7, 2015, at 10:19 AM, Andrew Sullivan <a...@anvilwalrusden.com> wrote: > For these purposes, a separate resolver is needed anyway. So a > different class is just as feasible.
You seem to be missing Tony's point. The point is that there is already a UI for typing, for example, "foo.onion". There is no UI for typing "foo" and saying "and that should be resolved using the 'onion' domain class." So a resolver that supports using a different protocol to resolve .onion can work with an application that isn't aware that this is happening under the covers. Whether that's a good or bad thing can of course be discussed, but that's the key point that your response doesn't address. Except for this part, which is correct: > I'd argue that mDNS is in the same boat. This technique, > really, is using the namespace as a clue that the protocol has shifted. Of course, that boat has already sailed... :) _______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
