On Fri, 2017-04-28 at 23:37 +0300, Ilari Liusvaara wrote: > On Fri, Apr 28, 2017 at 12:44:19PM -0700, Joe Touch wrote: > > > > The key, however, is that this proposal is really redefining HTTP > > ports > > 80 and 443 (if that's the direction you go), and you need to get > > consensus on that. It's not enough to simply say "we want to do it > > and > > it works as currently specd". > > The properties this draft exploits for detection are completely > fundamential to HTTP/1.x and HTTP/2. There is no way anything that > breaks those properties can be deployed on mass scale without a new > ALPN, which should be ample warning to the server that new things > are going on. > > This isn't "unextendable because of middleboxes", it is "unextendable > because the endpoints can't negotiate". > > > -Ilari
If we accept DKG's proof, it works on what *is*. "Cant negotiate"? There is no negotiation. The "server" demuxes on published, known, accepted, delineated protocol standards. Cant fix future changes and cant fix non-end-points screwing with the stream. Am I missing something? /Hugo _______________________________________________ dns-privacy mailing list [email protected] https://www.ietf.org/mailman/listinfo/dns-privacy
