[Jason Lango <jal@xxxxxxxxxx>]

On Fri, Nov 03, 2000 at 11:59:01PM -0800, Guy Harris wrote:
> > The dissector was tested against two Cisco 7960 IP phones and
> > Columbia University's sipd which was acting as a proxy between the
> > phones. A sample capture showing the three parties in action is at
> > http://www.cs.tut.fi/~hessu/sip2.cap
> 
> Hmm.  Jason, should SDP, like RTSP, have its dissector create
> conversations so that RTP streams and the like can be identified? In
> that capture, the first frame has "audio 16676 RTP/AVP 0" as the media
> description, and, later, there's a bunch of UDP traffic between ports
> 16676 and 18800, which I infer is RTP/AVP traffic, the 16676 coming from
> the media description.

That sounds right.  If the port number is zero, just ignore it and
don't create a conversation.

Interestingly Real sends 'audio 0 RTP/AVP 101' in its SDP even though
they have no intent of using RTP as the transport.  Fortunately for us
they don't slap in a bogus port number.  :-)

QT also sends 0 for the port number, since the client/server ports are
negotiated in the RTSP session.