[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [hybi] Reliable message delivery (was Re: Technical feedback.)






----- Original Message ----
> From: Greg Wilkins <gregw at webtide.com>
> To: "hybi at ietf.org" <hybi at ietf.org>
> Sent: Wed, 3 February, 2010 7:40:20 PM
> Subject: Re: [hybi] Reliable message delivery (was Re: Technical feedback.)
> 
> 
> I don't think you can assume that most websocket using software
> and/or subprotocols will have the same developer working on
> the client side and the server side.

+1
Cant agree more !
We have already seen this in action with the long poll based spec in xmpp.
I am sure hybi will end up being much more aggressively developed and deployed due to the greater reach and interest.

Regards,
Mridul

> 
> I can see separate impls for opera, firefox, jetty, tomcat, httpd etc.
> for some a standard subprotocol for multiplexing ws connections.
> 
> I can also see standards emerge for ws connection management that
> supports reconnects and retransmission.  These might be implemented
> in server frameworks and in js libraries like dojo and jquery.
> 
> Even if the dev team is the same client and server, it's just
> as likely that different developers will work on the javascript
> vs server language portions.    Even if the same person wrote
> both parts, it sounds like fragile design to assume implementation
> details of the other part.
> 
> Websockets is meant to be a significant change in web application
> development and new app models are likely to emerge.   I think it
> is very dangerous to say anything is "typical" of a protocol
> not yet deployed.  Even if there are typical usage patterns, I
> don't think it is good to design a protocol that only works for
> them.
> 
> 
> 
> regards
> 
> 
> 
> 
> Ian Hickson wrote:
> > On Wed, 3 Feb 2010, Graham Klyne wrote:
> >> Ian Hickson wrote:
> >>> The server and the client are typically written by the same person.
> >> If this is the basis on which the protocol is being specified, I don't
> >> understand why it's being done in a standards body.
> > 
> > So that the platform on top of which the client is written can be 
> > interoperably implemented such that it is possible for the clients to run 
> > on any implementation of the platform interoperably.
> > 
> > To clarify the terminology, from the perspective of the subprotocol 
> > running on Web Sockets, the platform is the browser with the Web Socket 
> > API, and the client is the JS code invoking the API, and implementing the 
> > client part of the subprotocol. The server implements both the Web Socket 
> > protocol and the subprotocol running over it.
> > 
> > The subprotocol is like HTTP, IRC, FTP, etc; Web Sockets is like TCP, and 
> > the platform is like the OS's network stack. Except everything is moved up 
> > one level.
> > 
> > HTH,
> 
> _______________________________________________
> hybi mailing list
> hybi at ietf.org
> https://www.ietf.org/mailman/listinfo/hybi



      The INTERNET now has a personality. YOURS! See your Yahoo! Homepage. http://in.yahoo.com/

Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.