Re: [webrtc-pc] Simulcast: Implementations do not fail (and that seems good) (#2762)

> > Regarding 2 and 3, erroring out when a remote SDP tries to narrow the envelope would violate the base IETF spec, IMO. Base spec requires us to allow this. Whether this breaks webrtc-pc is another question, but my position is that webrtc-pc should not be contradicting the IETF spec that it is based on.
> 
> I don't think it breaks anything and browsers are already following the IETF spec here, so this spec should probably yield.
> 
> > Regarding 6, this seems wrong. If offer/answer removes a rid by narrowing the envelope (which I think needs to be allowed, see previous line), it should probably not show up in JS anymore.
> 
> To clarify: 6 is about this line: _"If the number of [RTCRtpEncodingParameters](https://w3c.github.io/webrtc-pc/#dom-rtcrtpencodingparameters) now stored in sendEncodings is 1, then remove any [rid](https://w3c.github.io/webrtc-pc/#dom-rtcrtpcodingparameters-rid) member from the lone entry."_
> 
> Browsers _do_ trim the encodings array, they just don't get rid (no pun intended) of the rid member holding the name of the last remaining encoding when the array length is down to 1.
> 
> It looks like the a=simulcast line does go away a few renegotiations after this, so removing the rid member like the spec says seems consistent with the SDP browsers produce, so perhaps implementations should fix this.

Ah, I see. I am not aware of any IETF requirement that forbids a simulcast with only one rid, even though one could argue unicast-with-a-rid is silly. I think it probably makes sense to make the JS parameters and the SDP agree, whatever we end up doing.

-- 
GitHub Notification of comment by docfaraday
Please view or discuss this issue at https://github.com/w3c/webrtc-pc/issues/2762#issuecomment-1205772306 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Thursday, 4 August 2022 21:12:06 UTC