Best practices for using RtcpFeedbacks

I know that using RtcpFeedbacks, in general, would help to improve the quality of media transfer in bad network conditions. However I didn’t find any advice in the docs on when to use them and when not. Is enabling all the feedbacks for a specific codec (e.g. VP8) useful or they create overhead and there must be performance considerations when using them?

And why mediasoup demo does not benefit from them assuming that the demo code is kind of reference for developers?

RTCP feedbacks are defined in their corresponding RFCs and drafts. The mediasoup website/documentation is not a generic RTP/RTCP glossary.

The mediasoup-demo uses mediasoup and mediasoup-client. So of course those RTCP feedbacks are properly set and used. if you still think they are “not used” please be specific about what exactly you mean.

I see.

I just didn’t find any RTCP feedback usage in the codes, neither in the config.example.js nor in the server.js files.

Sorry I just confused. RTCP feedbacks are provided by the client using Device.rtpCapabilities so no need to be explicitly declared like what we do for PlainRtpTransport.

mediasoup-client includes per device supported RTCP feedbacks in each Producer rtpParameters it generates. In addition, device.rtpCapabilities defines the per codec RTCP feedback that the device supports for receiving.

That’s perfect. :ok_hand: