Re: [pntaw] TURN over websockets or just TURN.

["Hutton, Andrew" <[e-mail]>]

On: 25 September 2013 14:25 Sergio Garcia Murillo: Wrote:

> Anyway, the argument is still the same, turn over websockets will work
> on every scenario where turn over tls works, and in addition it may
> work
> in some others. Also, given that most people are using websockets for
> signaling anyway, we will ensure that in 100% of the cases if signaling
> over websockets works, media over websockets will work (except if
> explicitly forbidden by enterprises policies).
> 
> We can argue later if the percentage of cases additionally covered by
> turn over ws is significant enough so it is worthy to add a new
> protocol
> and change both clients and server (while I see that other initiatives
> are requiring it already). But I think that at least having an
> alternative proposal is a good thing.
> 

This debate is definitely a good thing and we seem to be agreed that there is a problem to solve so we are making progress. I was trying to have the debate now on whether the websockets overhead is worthwhile or not I don't see any reason to delay that discussion.

I am still to see any concrete examples of scenarios where the websockets approach provides better connectivity which is what I think we need to see before we can say that it is worthwhile.

Regards
Andy