Inaki said:
"However, Bernard pointed out that such a design lacks the ability to
set different RTX/FEC options per layer, which I do not fully
understand since, at least in a similar thread in ORTC, he assumes
SRST. This is, a single RTP stream, meaning that just a single RTX/FEC
setting applies."
[BA] I believe that SDP cannot currently express the desire to provide differential protection for SVC codecs utilizing SRST encapsulation, only for MRST. So support for differential protection in WebRTC 1.0 does not seem easily achievable.
However, in ORTC it does seem possible to request a sender to differentially protect SVC layers. For example, FEC or RTX can be supported only for the base layer.
For RTX, this can be implemented on the sender by only retransmitting base layer packets and not retransmitting in response to NACKs for extension layer losses. For FEC, the sender would only apply forward error correction to base layer packets (Flexfec can support this, no?)
Assuming that this is correct (we have no implementations, so we don't know for sure), it seems like it is purely a sender decision and the receiver does not need to be configured for differential protection (or even for the layer configuration for VP8, VP9 or AV1). Also, it would appear that this can be made to work for SRST.