W3C home > Mailing lists > Public > ietf-http-wg@w3.org > October to December 2012

Re: on DNS records

From: Roland Zink <roland@zinks.de>
Date: Thu, 15 Nov 2012 12:03:01 +0100
Message-ID: <50A4CBE5.9020107@zinks.de>
To: ietf-http-wg@w3.org
On 15.11.2012 01:56, Mark Nottingham wrote:
> On 15/11/2012, at 11:54 AM, Patrick McManus <pmcmanus@mozilla.com> wrote:
>> - Are we advertising that port 80 is capable of HTTP/2, or an alternate port for HTTP/2, or (capable of) both?
>>
>> imo it has to be an arbitrary port.. if you send anything other than http/1 on port 80 across the general internet you'll get some breakage.
> Agreed... however, I don't see a reason to make it impossible to say "I support HTTP/2 on port 80 too", so the client can optimistically try to use it first (as long as it can handle breakage).
>
> Cheers,

When it can be determined from the first few bytes on the wire if it is 
HTTP/1 or HTTP/2 then servers and network elements could handle both 
protocols on the same port. The first bit may be enough for this (0 for 
ASCII HTTP and 1 for HTTP/2 control message?). However in the beginning 
there could be unnecessary breakages and a HTTP/2 default port may be a 
way around it. For example traffic may routed around HTTP/1 only network 
elements.

>
> --
> Mark Nottingham   http://www.mnot.net/
>
>
>
>
Roland Zink   http://home.zinks.de/
Received on Thursday, 15 November 2012 11:03:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 15 November 2012 11:03:26 GMT