Valid RTSP/1.0 ? yeah right, even the IETF admits that RTSP version 1.0 is a broken protocol (read the RTSP/1.1 summary).
We could try to be compatible, and use RTCP for RTSP unicast streaming keep-alive...
But would this not break some clients? OTOH, the current RTCP-less stuff is broken anyhow.
In fact, RTCP is optional on the client side: the client can use RTSP and/or RTCP to refresh the timeout. However, on the RTSP server side it's mandatory, since some clients might be using to notify the server they are still up.
Currently, we probably do not really need RTCP in the RTP access, which is biased toward multicast or manually setup UDP streams anyway. And we do send RTCP from the RTP access output - we don't really care what we receive in that case for similar reasons.
However, most of the work would be in the RTSP server where there is currently nothing, meaning RTSP timeouts are currently broken.
Rémi Denis-Courmontchanged title from RTCP support to RTCP support for RTSP stream output
changed title from RTCP support to RTCP support for RTSP stream output