Re: RTX support (RFC 4588)

On 21/10/2014 19:51, Peter Thatcher wrote:
>
>
>     So, at least that I am aware of, it is not possible to now in
>     advance which of 345259865  or 2693756249 <tel:2693756249>
>     corresponds to the rtx stream. In this case, the ortc
>     implementation may still behave correctly as only one of the
>     streams in the bundle has the rtx codec, but not sure how would it
>     be able to choose what stream an rtx packets belong, if for
>     example the audio stream has also a rtx codec, or there is more
>     than one video stream with rtx in the rtp session.
>
>
> ​ ​In Chrome, that SDP is interpreted to mean that the RTX ssrc is ​ 
> 2693756249, which makes it unambiguous and possible to know in 
> advance.  ORTC follows the same model: it's unambiguous and possible 
> to know in advance.

Well, that is because you know Chrome internals, not because the order 
of the ssrc means anything in the specs. In fact, other webrtc 
implementation could reverse the order and still be perfectly compatible 
with the current specs, or even worst, add a third one and shuffle them.

So, I understand that this is an issue of WebRTC and the way the SDP is 
used there, but that would prevent and ORTC endpoint to talk safely to 
an webrtc endpoint.

Best regards
Sergio