Re: [hybi] New mailing list: hybi (HTTP "long poll" and related protocols)

[Mark Nottingham <[e-mail]>]

On 31/03/2009, at 6:41 AM, Ian Hickson wrote:

> On Tue, 31 Mar 2009, Mark Nottingham wrote:
>>
>> As discussed in the APPS area meeting last week, a new mailing list  
>> has
>> been created to discuss the standards implications and actions  
>> related
>> to HTTP "long poll" techniques (e.g., COMET, BOSH) as well as new
>> protocols that serve similar use cases (e.g., WebSockets, rHTTP).
>
> Thanks!
>
> I notice the description of the list is "Server-Initiated HTTP",  
> while at
> least some of the techniques proposed, in particular Web Socket, are  
> only
> tangentially related to HTTP (WebSocket is a client-initiated Web- 
> aware
> bidirectional socket, it's not server-initiated and the HTTP aspect is
> just an optional bootstrapping mechanism). Is the scope intended to
> include or exclude long-term discussion of such solutions?

Ah, the original name of the list was si-http; that's a bit of a  
leftover from that discussion. The scope is definitely to include such  
discussions, at least until we figure out otherwise.


> Also, what is the process by which discussion on this list can  
> result in a
> standards-track document? Browser vendors have started asking me about
> implementing WebSocket or whatever replaces it, so a clear picture  
> of how
> we get to something stable sooner rather than later would be very  
> helpful.
> (It is an unfortunate reality that once we have multiple  
> implementations
> of a feature, the spec becomes stable by necessity, regardless of  
> whether
> we have had the proper time to review it or not.)


Either something being an individual submission, or by a Working Group  
forming.

AIUI the bar is relatively high for individual submission standards- 
track documents these days; i.e., if there are contentious issues,  
it's likely that it'll need a WG.


--
Mark Nottingham     http://www.mnot.net/