[Live-devel] RTSP Redirect

[xcsmith at rockwellcollins.com]

The way to implement a proxy is to make it both a RTSP server (accepting
commands from the original client) and a RTSP client (connecting to the
back-end server, transparently to the original client) - without any change
in the way that the RTSP protocol is used.
-------------------------------


I have a similar RTSP complexity.  My application RTSP client connects to
an RTSP Server on an embedded device.  The device can receive a multicast
TS and decode MPEG video/audio for analog output to monitor and speakers.
The device RTSP server is directed to SETUP and PLAY the stream.  The
device does not have its own storage, so it must connect to a second RTSP
server and request the stream.  The device does not know where the server
with the data is, but the original application RTSP client does.  So how
should my messages look?  Should the original RTSP client perhaps use
something like the "content-base" field to tell the device the location of
the storage server?  (If not, what is the purpose of "content-base"?)  Or
should the original RTSP client direct the storage server to play the
stream and ANNOUNCE the stream to the embedded device?

Because we did not know how to solve this problem before, we decided that
the RTSP message sent by the original client will contain the RTSP URL of
the storage server, even though the message is sent to the embedded device.
The device learns the location of the storage server from the RTSP URL in
the original message.  Not really ideal, but better than anything else we
came up with. What do you think of this?  When we read the standard, we
thought that ANNOUNCE would be deprecated, so we did not try to use
ANNOUNCE.

Thanks!
~Medra