[MPlayer-dev-eng] Re: [Live-devel] About adding "Real" RDT protocol support to Live....

[Ross Finlayson]

>I noticed that there is good Audio/Video syncronisation support in the
>LIVE.COM Streaming Media libraries and it's also used by the interface
>layer libmpdemux/demux_rtp.cpp.

Yes, but the actual audio/video synchronization is done by the 
(higher-level) MPlayer code.  The a/v sync support in the LIVE.COM code 
uses information in RTCP  "Sender Report" packets to set accurate 
presentation times on audio and video frames, but it's the higher-level 
MPlayer code that does the actual synchronization.

However, I don't think RealNetwork's proprietary  'RDT' protocol uses RTCP 
(at least, not in the standard way), so the existing LIVE.COM code wouldn't 
help at all here.

>My guess was that adding the code from mplayer/libmpdemux/realrtsp/real.c
>(it contains the proprietary RDT protocol and RealPlayer handshaking)
>could give the LIVE.COM Streaming Media library the possiblity to retrieve
>RealAudio/RealVideo "rtsp://" streams.
>
>Would this be possible?

Yes, it would be 'possible'.  Right now, though, I'm not particularly 
inclined to do this, unless there is commercial demand for it.  (Instead, 
I'd rather see RealNetworks' RDT hack continue to fade away into 
extinction, as standard RTSP/RTP/RTCP gets increasingly used instead.)

Instead, I suggest that you look at hacking MPlayer's RealAudio/RealVideo 
"rtsp://" code (i.e., not the LIVE.COM code) so that it properly sets the 
"pts" (presentation timestamp) field in each "demux_packet_t" (using 
whatever mechanism RDT uses to specify this.

>PS: For compiling the live.2003.07.29 on my gcc-3.3+ system, I needed
>the attached patch.

Note that I already explain this in 
<http://www.live.com/liveMedia/#config-unix>.  (See "If you're using "gcc" 
version 3.0 or greater").


	Ross Finlayson
	LIVE.COM
	<http://www.live.com/>