- From: Patrick McManus <pmcmanus@mozilla.com>
- Date: Tue, 21 Aug 2012 08:30:51 -0400
- To: James M Snell <jasnell@gmail.com>
- Cc: ietf-http-wg@w3.org
Hi James, On Mon, 2012-08-20 at 21:59 -0700, James M Snell wrote: > That is, I as an implementor am not going to be required to do this > upgrade dance with every new connection, regardless of how it actually > happens... right? If we define something on port 80 that uses Http Upgrade, then likely you will - like websockets does. That's an argument against http upgrade. I would like to think some mechanism that lets you explicitly know that http/2 is in use on another port would be preferable and let you skip that dance once discovery is complete (i.e. dns srv and/or alternate-protocol). I don't think sending http/2 on port 80 without upgrade would ever be viable due to proxies etc.. > http2://www.example.com that operates on a different default port is a new scheme is out of scope of the proposed charter as far as I understand it as the charter expects HTTP/2 to retain the existing URI semantics of HTTP/1.1.
Received on Tuesday, 21 August 2012 12:31:26 UTC