[Standards-JIG] Re: PROPOSAL for Live Chat
tmarkmann at googlemail.com
Fri Jan 5 16:43:14 UTC 2007
Yeah...since all this data goes through a server and not peer to peer. We
really should keep in mind not to push unneeded traffic to a XMPP server.
On 1/5/07, Alexander Gnauck <gnauck at ag-software.de> wrote:
> > Character-by-character updates are something that you send while in the
> > composing state, but your chat state doesn't change while you are
> > sending each character -- all the while you are still just composing as
> > far as chat state notifications are concerned. So there is no reason to
> > send another composing event (the "yep I'm still composing" event?).
> > IMHO a separate protocol is in order. Negotiate it separately from chat
> > states because it really is different and will have its own parameters
> > (do we agree to use it? do we send one char at a time or bunch them up?
> > etc.). Don't overload chat state notifications with a function it wasn't
> > intended to perform, and let those specialized systems that want "live"
> > chat have it in a granular manner.
> i would send one or multiple characters in a specific interval (500 or
> 1000ms), and only if there are outstanding chars or key actions to send.
> This would reduce the traffic if somebody is typing very fast.
> And as Ralph suggested some index for ordering would be good. In order
> delivery is required according to the RFC, but as we have seen in a
> previous discussion its currently a problem on some servers.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Standards