On Thu 2017-04-27 17:34:27 +0000, Hugo Maxwell Connery wrote: > Should we mandate that all future protocols are "demuxible" from > all previous?
fwiw, i'm not prepared to make any kind of statement anywhere near that
grand, and i'd hope that the specific proposal under discussion doesn't
hinge on us having to make such sweeping analysis. What are "all
previous protocols", for example? does it need to be demuxable from
tcpmux itself? or what about RFC 862? ;)
In particular, the approach i've described in the draft only works for
client-speaks-first, stream-based protocols. It doesn't work at all for
server-speaks-first protocols like SMTP or SSH, for example, and it
doesn't address non-stream-based protocols at all.
The proposed approach is situated in the particular historical
circumstance we find the Internet in today. It'd be great if it wasn't
necessary to consider it (if there were no networks actively hostile to
user privacy and "surprising" ports, for example), and hopefully future
Internet transport protocols can be demuxed at the server in a way that
is opaque to the network operator in the first place, so that problems
like this don't arise.
I'm happy to have the bigger-picture discussion about how to plan for
the future, but i hope this particular draft can be evaluated on its own
merits (or lack thereof), largely independent of that discussion.
--dkg
signature.asc
Description: PGP signature
_______________________________________________ dns-privacy mailing list [email protected] https://www.ietf.org/mailman/listinfo/dns-privacy
