RE: <device> proposal (for video conferencing, etc)

On Fri, 18 Dec 2009, Nick Lothian wrote:
> > > >
> > > > However, in this case we need to even more, because otherwise 
> > > > there's no guarantee that a user with one browser could chat to a 
> > > > user with another browser, which makes the whole exercise 
> > > > pointless.
> > >
> > > Won't that go via a server, which could potentially transcode?
> >
> > When we do go through a server, it would certainly be better if the 
> > server didn't have to transcode every frame. That's a lot of load on 
> > the server. But I think ideally we'd eventually come up with a 
> > client-to-client protocol, and so the server couldn't transcode even 
> > if it wanted to.
> 
> Do you consider the client-to-client protocol to be in scope at the 
> moment, or is that far-future work?

I don't know. If we do introduce a client-to-client protocol, I'd imagine 
we'd want to reuse an existing one, so that we don't have to resolve all 
the NAT traversal issues. (Simplicity wouldn't be as high a priority in 
this case as in the client-server case, since we'd expect far fewer 
independent implementations, so there'd be less to gain by developing a 
whole new protocol than there was with, e.g., Web Sockets.) I'm not 
familiar with existing protocols for this, though.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Thursday, 17 December 2009 21:46:52 UTC