Side-effect of sending duplicate packets to clients

[penguinol]

In #906, we are considering not filtering duplicate rtp packets(received after rtx packets) and instead sending them to client.
If we enable transport-cc, a transport-wide-cc-id will be set to each video packet.
On the client side, duplicate packets will be dropped by libsrtp when it attempts to decrypt them, and these packets will be marked as NR in tcc feedback.
Below is a simple example, i just motified the simulcast consumer, and sent each rtp packet twice.

Below is the pcap file:
tcc.zip

[aankur]

Hi,
One suggestion can we calculate transport-wide-cc-id, before sending the packet, the packet will have the same twcc-id even if resent, even if srtp fails the packet, it would not matter, as the twcc id would already be seen

[jmillan]

One suggestion can we calculate transport-wide-cc-id, before sending the packet, the packet will have the same twcc-id even if resent, even if srtp fails the packet, it would not matter, as the twcc id would already be seen

There is already a fix for this issue here #1683

[aankur]

Thanks