[FFmpeg-devel] RTCP packets issue

[sateesh babu]

[...]
>>* Sorrry again, I misread the code. I now think you are right; ntp_time*>>* is not currently set in the correct way. What you do is equivalent to*>>* "ntp_time= ff_ntp_time();", but I am not sure it is correct... In this*>>* way, you compare a time received in the SR packet (last_rtcp_ntp_time)*>>* with the local time... I am going to check what should be done, here.*
>The attached patch should fix this issue (at least, wireshark now shows
>reasonable times... But I did not try to use them for round-trip
>propagation delay computations).

>Sorry again for the misunderstanding.


Luca,

    Thanks for the patch. The 'delay_since_last' values are more
sensible now. I think I have got a funny encoder with me which is not
recognised as either RTSP_SERVER_REAL or RTSP_SERVER_WMS but it
expects a 'GET_PARAMETER' to keep connection alive. So I have added a
RTSP_SERVER_GENERIC to the RTSPServerType enumeration and handling the
connection alive messages in 'rtsp_read_packet(...)' function call as
described above. A more cleaner implementation might be possible.

Another thing I have noticed is that both VLC and Quicktime RTSP
implementations are writing their own SSRC before writing the
hostname(computer name?) in a RR packet. But according to RR packet
format in http://webee.technion.ac.il/labs/comnet/netcourse/CIE/RFC/1889/20.htm,
this is incorrect. Any ideas why they might be doing this?


Sateesh