W3C home > Mailing lists > Public > public-webrtc@w3.org > July 2014

Re: ice connection state change to "connected" vs data channels

From: Harald Alvestrand <harald@alvestrand.no>
Date: Tue, 01 Jul 2014 08:02:10 +0200
Message-ID: <53B24EE2.4090307@alvestrand.no>
To: public-webrtc@w3.org
On 06/30/2014 09:01 PM, Philipp Hancke wrote:
> The spec says the following about the ice connection state:
>   2. If the ICE Agent has found one or more candidate pairs for each
>   MediaStreamTrack that forms a valid connection, the ICE
>   connection state is changed to "connected".
>
> I ran into a bug where I accidentially dropped all ice candidates
> related to the m=application m-line and was not using BUNDLE.
>
> I did not expect the iceConnectionState to change to connected,
> however it did.
>
> I think the spec doesn't consider this case. Nor does the text
> describe what to do on a data channel only connection.

I think the text should say "for each Transport that forms a valid
connection", with a definition somewhere of Transport that makes it
clear that a Transport is something you can map MediaStreamTracks or
data connections onto.

We already have the Transport concept exposed in stats, so it's not
completely new.

>
>
> http://code.google.com/p/webrtc/issues/detail?id=3526  shows a way to
> reproduce this. I _think_ Firefox behaves the same way, but it's
> harder to drop all ice candidates because the local ones are included
> in the initial offer/answer.
>


-- 
Surveillance is pervasive. Go Dark.
Received on Tuesday, 1 July 2014 06:02:45 UTC

This archive was generated by hypermail 2.3.1 : Monday, 23 October 2017 15:19:41 UTC