aviad rozenhek wrote:

On Tue, Aug 23, 2011 at 13:33, Howard Chu <hyc@highlandsun.com <mailto:hyc@highlandsun.com>> wrote:

aviad rozenhek wrote:

I'm testing RTMPDump with a live stream rtmp://cp102272.live.edgefcs.__net/live/tbn_hd_2@29502 <http://cp102272.live.edgefcs.net/live/tbn_hd_2@29502> <http://cp102272.live.edgefcs.__net/live/tbn_hd_2@29502 <http://cp102272.live.edgefcs.net/live/tbn_hd_2@29502>> --live, and when I look

at the result file, I see that all the A/V frames received during the first few seconds of streaming have dts=0. I first saw this problem when using ffmpeg/libav, therefore I opened a bug #12 for it in libav bugtracker <https://bugzilla.libav.org/__show_bug.cgi?id=12 <https://bugzilla.libav.org/show_bug.cgi?id=12>>

for example, I can either use RTMPDump to download the stream to an flv file, and then probe it using ffprobe to look at timestamp, or just use ffprobe [which is RTMP-enabled] to look directly at the timestamps. in any case I see that there is a significant amount of audio and video frames that all have non-increasing dts=0 timestamp

i'm attaching a dump of the packets as packet.txt

What timestamps are in the actual RTMP stream?

I'm sorry but I don't know. its a live stream, not a VOD, and I'm not the one running the server. RTMPDump and ffmpeg/libav are the tools I know to analyze the stream

If you haven't looked at a hex dump of the packets on the wire then you have no evidence there is any bug here. If you can show that the timestamps in the raw stream are actually some other value and that rtmpdump's output doesn't match, then we'll have something to investigate.