I'm trying to figure out why the http streams go bad and I'm finding
that with a perfect connection the packet size is around 400 bytes.

I found this by printing out the value of iBytesUsed in:

PullBuffer::EndRead(size_t iBytesUsed)

When streams go bad, however, the skips correspond to shorter reads,
they are as small as 1 byte, but usually around 100 bytes.  When these
packets make their way through the stream, they are responsible for
the problems.  It doesn't appear to be related to the size of the
buffer.

This problem is exacerbated if I use another icecast to relay the
stream.  If you want to test this, try pointing freeamp at
http://penguin.censoft.com:8000/mix

The server in this case is icecast, and it happens mainly with low
bandwidth connections; in this case I am streaming a 128kbps stream
to my 512kbps DSL line.  The buffer size appears to be staying at the
size I'm specifying, so it's not like the server can't keep up.

Is there a problem on the server end that is causing this?  Perhaps
there are some socket options I need to research, or maybe there is
a kernel TCP/IP setting I need to set?  I'm kind of shooting in the
dark here; but eventually I'll find the problem.  Any assistance or
thoughts would be greatly appreciated.

Thanks,
Mark


_______________________________________________
[EMAIL PROTECTED]
http://www.freeamp.org/mailman/listinfo/freeamp-dev

Reply via email to