- From: Amit Hilbuch via GitHub <sysbot+gh@w3.org>
- Date: Thu, 28 Feb 2019 18:34:21 +0000
- To: public-webrtc-logs@w3.org
@murillo128 Regarding RIDs being ignored in chrome. This is currently by design, as there is no need for them: 1. It has been made clear in the past (in the last interim as well), that the simulcast scenario requires an SFU, and that PC does not support receiving simulcast. Two peers do not need to use simulcast because they can directly negotiate the best configuration for their session without sending multiple layers. In group call scenario, not having an SFU means that we have a quadratic number of connections and that leads to a terrible experience that would be worsened by trying to send and receive multiple layers. As such no support has been added to PC to enable receiving simulcast. The [spec](https://rawgit.com/w3c/webrtc-pc/master/webrtc.html#simulcast-functionality) also indicates that there is only support for issuing a=simulcast with the send direction, consistent with this decision. To be fair it also mentions the scenario of receiving but does not define how this should be done. 2. Without simulcast, the use of RID is redundant as MIDs are sufficient identifiers. Assuming unified-plan semantics, MID is sufficient to identify streams and route them correctly so RIDs are redundant in this sense. Moreover, the [spec](https://rawgit.com/w3c/webrtc-pc/master/webrtc.html#dom-rtcpeerconnection-addtransceiver) indicates that a single RID should be removed and is not negotiated in the SDP: ` If the number of RTCRtpEncodingParameters now stored in sendEncodings is 1, then remove any rid member from the lone entry. ` -- GitHub Notification of comment by amithilbuch Please view or discuss this issue at https://github.com/w3c/webrtc-pc/issues/1174#issuecomment-468385559 using your GitHub account
Received on Thursday, 28 February 2019 18:34:24 UTC