[FFmpeg-devel] Realmedia patch

[Luca Abeni]

Hi Ronald,

Ronald S. Bultje wrote:
> Hi Luca,
> 
> On Sat, Aug 30, 2008 at 6:30 AM, Luca Abeni <lucabe72 at email.it> wrote:
>> Ronald S. Bultje wrote:
>> [...]
>>> @@ -918,7 +920,9 @@
>>>              if (transport[0] != '\0')
>>>                  av_strlcat(transport, ",", sizeof(transport));
>>>              snprintf(transport + strlen(transport), sizeof(transport) - strlen(transport) - 1,
>>> -                     "RTP/AVP/UDP;unicast;client_port=%d-%d",
>>> +                     "%s;unicast;client_port=%d-%d",
>> Is "client_port=%d-%d" really needed here, or would "client_port=%d"
>> be enough? (RDT does not seem to have an RTCP-like companion protocol)
> 
> UDP has one port per "m" stream 

This is exactly my point... UDP needs a port per stream, and here you are
declaring two ports per stream (this makes sense for RTP, because it also
needs port+1 for RTCP).


> (i.e. in the case of audio+video, they
> stream over a separate UDP session with separate port).

I am not sure about this... If I understand well, you will see a single
"m=" line, and both audio and video streams will be muxed inside the RM
stream.


> So I think it
> is needed. (This is of course in the theoretical case that Realmedia
> RDT supports UDP - you and I both have seen only TCP-supporting
> servers...)

Note that there is a "SETUP" request per stream (one setup per "m=" line
in the SDP). So, the "%d-%d" looks wrong (this is why I asked).


I'm going out right now; I'll look at your new patches on Monday.



			Thanks,
				Luca