mkosma wrote:
bump ... welcome back Richard.
Thanks :).
Any thoughts?
Yeh, this threads got way too long, but I've learnt some useful netstat
commands on windows ;)
It looks like you have had several problems here, lets take each one in
turn:
1) SSH connection problems
From your early logs (for example in post #5) it is clear that you were
having some network problems. In that log a couple of clues exist:
1063491 [AudioDecoder-5] DEBUG audiobuffer.verbose - buf R=6389181
W=6389181 #=7613
1063501 [AudioDecoder-5] DEBUG audiobuffer.verbose - avil R=6389181
W=6389181 A=0
1063501 [AudioDecoder-5] DEBUG audiobuffer.verbose - avil R=6389181
W=6389181 A=0
1063501 [AudioDecoder-5] DEBUG audiobuffer.verbose - audio buffer underrun
1063501 [AudioDecoder-5] DEBUG audiobuffer.verbose - avil R=6389181
W=6389181 A=0
1063501 [AudioDecoder-5] DEBUG audiobuffer.verbose - avil R=131840000
W=135343616 A=3503616
Firstly the A=0 and audio buffer underrun shows that no bytes are
available to the mp3 decoder. The buffer then fills quickly with 3503616
bytes available in the buffer.
1064633 [SlimTCP-1] ERROR player - strm IO error
java.net.ConnectException: Connection refused: connect
Secondly this error shows that the tcp stream could not be opened for
the next track (Squeezebox/Softsqueeze opens a stream per track). This
explains why the play back stops, as the next track has not been opened.
I guess this should be handled more gracefully - I am not sure what a
SB2 does in this situation?
You then tried using putty for SSH, and as I understand things got
better. I would be interested if you could do another comparison now;
was it a software or network issue you originally saw?
I did have some problems with the Java SSH library when I originally
integrated it. The SSH 'window' is not correctly implemented, this only
effects streaming applications like Softsqueeze. I had to make some
workarounds, fixing the library would be a big job and other
applications are less sensitive to the bug.
2) Buffer underruns (see post #20, post #44)
672972 [AudioMixer-0] WARN javasound - Java Sound Audio buffer underrun
bufAvail=3470848
672982 [AudioDecoder-4] WARN javasound - AudioDecoder could not write
buf to outputBuffer
The 'Java Sound Audio buffer underrun' warning is probably not a
problem, it happens sometimes. I added the warning thinking it would
indicate a playback problem but this does not seem in my experience.
The 'AudioDecoder could not write ...' error is a problem. This
indicates that decoded audio has been lost, and will lead to a glitch in
the playback. Does this always happen with the same tracks in the same
place? I will fix this in the release of Softsqueeze - it is caused by
lazy coding on my part.
3) Network health plugin
Nice work Triode :)
4) MP3 Plugins
Your description of the buffer behavior (post #52) is good, just a shame
about audio glitch. Again I suspect this is due to the 'AudioDecoder
could not write ...' error.
As for Java MP3 plugin verses jlayer read the following post for
pros/cons (ignore all the it don't work posts!)
http://weblogs.java.net/blog/jonathansimon/archive/2004/11/suns_mp3_plugin.html.
The Java MP3 plugin does use less CPU to decode the tracks, but it is
not perfect. For example I had to make sure all meta data is parsed out
of the stream before passing it to the decoder, otherwise it can spin in
a tight loop.
Hope this helps. I will post a fix for the AudioDecoder in a day or two,
let me know if this helps.
Richard
_______________________________________________
Discuss mailing list
[email protected]
http://lists.slimdevices.com/lists/listinfo/discuss
