W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2009

Re: #172 (take over HTTP Upgrade Token Registry) httpbis

From: Julian Reschke <julian.reschke@gmx.de>
Date: Sat, 15 Aug 2009 22:13:23 +0200
Message-ID: <4A8716E3.2050703@gmx.de>
To: Mark Nottingham <mnot@mnot.net>
CC: HTTP Working Group <ietf-http-wg@w3.org>
Mark Nottingham wrote:
> 
> On 06/06/2009, at 5:49 AM, Julian Reschke wrote:
>> Open Questions:
>>
>> - are we happy with the details of the registration procedure (if not, 
>> we should treat that as separate issue)?, and
> 
> My first impulse upon reading this was to suggest it be changed to a 
> designated expert procedure (which IMO is good policy, as it allows a 
> sanity check on registrations without raising the bar too high). 
> However, AFAICT there aren't as many risks in registering a new upgrade 
> token as there would be for a new header (for example), and the 
> considerable effort required to deploy a new upgrade token successfully 
> suggests that the registry won't be inundated.
> 
> So, I'm not fussed either way.

OK, so let's leave things as they are for now.

>> - is the registry supposed to take just product tokens, or 
>> product/version pairs? The text in RFC 2817 is unclear, and the one 
>> value it registers contains both.
> 
> My reading of the text is that the term "token" was an unfortunate 
> choice, and the intent was to register both.

RFC 2817 registers "TLS/1.0" (token + product-version), while RFC 2616 
reserves "HTTP" (just token).

So we need to decide how to populate the existing registry 
(<http://www.iana.org/assignments/http-upgrade-tokens/>) which currently 
  is empty -- actually it isn't anymore, it now has a broken entry for 
"WebSocket":

> Value  Description       Reference
> -----  ----------------  ---------
> 1      WebSocket         [draft-hixie-thewebsocketprotocol]

...broken in that the value to be registered isn't "1", right?


Citing the registration procedure:

>    6.  The responsible party for the first registration of a "product"
>        token MUST approve later registrations of a "version" token
>        together with that "product" token before they can be registered.

...which licenses the registration of just a token.

So it seems the registry can take both.

My proposal thus is to instruct IANA to change the registry to:


Value      Description                  Reference

HTTP       Hypertext Transfer Protocol  [RFC2616]
TLS/1.0    Transport Layer Security     [RFC2817]
WebSocket  WebScocketProtocol           [draft-hixie-thewebsocketprotocol]


BR, Julian
Received on Saturday, 15 August 2009 20:14:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:08 GMT