[vlc-devel] Re: vlc: svn commit r10012 (fenrir)

Laurent Aimar fenrir at via.ecp.fr
Mon Feb 28 11:53:08 CET 2005

On Fri, Feb 25, 2005, Ross Finlayson wrote:
> >> One comment, though: As you note in your comments, your new 
> >"RollOverTCP()"
> >> function duplicates a *lot* of code from "Open()".  It would be better to
> >> remove this duplication, by having just a single 'RTSP open' routine that
> >> requests either UDP or TCP streaming, based on a parameter.
> > I've done it this way to be able to test changes faster (and to avoid
> >remove/add of all ES), but yes it should be reworked :)
> >
> > I thought that I just need to issue a setup with new parameters and
> >that's all.
> No - no RTSP server will let you switch to streaming over TCP just by 
> sending another "SETUP".  Instead, you need to "TEARDOWN" the existing 
> connection, then do "DESCRIBE", "SETUP", "PLAY" all over again.
 Oh ok.

> As far as I can tell, your new "RollOverTCP()" function is correct.  Thanks 
> again for doing this.  My only comment was that because it duplicates a lot 
> of code from "Open()", the code from these two functions should - at some 
> point - be folded together into a single, common routine.
 Yep, I agree (I will try to work on it).

> > (the server was openRtsp from live.com),
> FYI, "openRTSP" is a RTSP client, not a server.
 Oops, my bad, I used testOnDemandRTSPServer  (the one you give test.* files
and 'VOD')

> >Oh btw, is there any way we can know if the streaming is using multicast
> >or unicast ?
> I think the following should work:
>         IsMulticastAddress(sub->connectionEndpointAddress())
> (This should be called *after* the RTSP "SETUP" has been done.)
 Perfect thanks. (I failed to find this one).

> In fact, for multicast streams, it's unlikely that - if no multicast data 
> is received - that a new attempt to stream from the same URL using 
> RTP-over-TCP will work (because TCP streams are, by definition, 
> unicast).  I suppose there's no harm in trying, though...
 Yep, you're right, so I will just use a high timeout and fail if no data.


This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html

More information about the vlc-devel mailing list