[Live-devel] Add an audio sub-session makes the video stop

[Fabrice Triboix]

Hello everyone,

I hope somebody could point me to some directions where to investigate.

I am trying to modify an existing RTSP server based on live555. It 
streams live video without problems, and I have to add live audio 
sub-streams for each video streams.

As a first step, I wanted to stream an MP3 file, so I created an audio 
source class based on the "DeviceSource" template. Every time the 
doGetNextFrame() function is called, I read 10000 bytes from the file 
and update the data members of the class accordingly, and then call the 
static FramedSource::afterGetting() method.
I normally connected the output of my class to an 
"MPEG1or2AudioStreamFramer" by calling
   CMySource* src = CMySource::createNew(...)
   MPEG1or2AudioStreamFramer::createNew(envir(), src)

In the existing code, there is also a class based on 
OnDemandServerMediaSubsession, let's call it CMySubsession. It implements:
  - createNewStreamSource(): which returns a MPEG1or2AudioStreamFramer* 
created as above
  - createNewRTPSink(): which returns an MPEG1or2AudioRTPSink

Now, if I don't add the audio sub-session, the video plays fine with 
VLC. If I add the audio substream, the destination address in the 
"groupsock" for the video stream stays at 0.0.0.0, and thus nothing is 
sent. I also noticed on the VLC side that no SDP is sent and the server 
closes the RTSP TCP connection after about 10 seconds.

The MP3 file looks OK, I can stream it using testMP3Streamer.

Thanks a lot for any help!

Best regards,

   Fabrice