PasTim;635123 Wrote: 
> I could use the commercial email address for feedback if that suits. 
> Let me know.Yes. That would be fine thanks.

PasTim;635123 Wrote: 
> I do notice that the content range in the last flac log looks very
> strange and there are errors "Requested Range Not Satisfiable" and
> "Range: bytes=0-1" looks weird, but then I don't know the protocol.It is my 
> theory that these HTTP Range requests are the root of the
problem with the Humax. They are certainly very unusual, and I have
encountered no other player that makes such.

As you notice, before starting a transaction, it does a "HTTP HEAD
Range: bytes=0-0" request (read the response headers corresponding to a
GET of one byte); and if that fails, it does a "HTTP GET Range:
bytes=0-1" (actual read of two bytes).

The release version of Whitebear accepts such range requests when
transcoding music because it CAN deliver these first one or two bytes.
However in the test version I sent you, I decided to see what would
happen if Whitebear would reject such range requests (416 Requested
Range Not Satisfiable) because, even if it can deliver the first 1 or 2
bytes, it would later be in a pickle if asked to deliver the NEXT one or
two bytes. This is because when transcoding via flac.exe and lame.exe
you can never be exactly sure of the time offset in the input file that
would be needed in order to deliver a coresponding byte offset in the
output stream. Not to mention that mp3 streams deliver data in samples
that are always larger than one or two bytes.

By the way, the reason it works with the Humax when playing native mp3
files is that, since the server has the actual mp3 file in its hands,
it CAN deliver exactly any range of bytes requested. Unlike the case
when double transcoding where the server has to make guesses on the
fly.

Frankly, I am not sure WHY the Humax is doing these fancy HEAD or GET
Range requests. I suppose it is trying to determine the full length of
the stream from the response Content-Range header before it actually
launches the full download of the music the stream using a proper full
blown GET. 

As an aside, for example Windows Media Player, also always does two
full blown GETs, of which the first one is immediately aborted; so the
Humax is not alone in behaving oddly...

Of course we still have the other theory about the UUUs in the mp3 file
too... 

So, in my next test build, I shall do two things: 1) revert to
accepting these odd Range requests, so that the response Content-Range
header is delivered properly, and 2) capture the delivered stream in
parallel to a disk file, so that you can then later test if the disk
file plays as a native file on the Humax (UUUs and all).


-- 
AndrewFG

Regards,
AndrewFG

Try out Whitebear. The middleware that joins the two worlds of:
1. UPnP/DLNA media clients and media players, and,
2. Squeezebox Server and Squeeze Players
Download it for free here: http://www.whitebear.ch/mediaserver
------------------------------------------------------------------------
AndrewFG's Profile: http://forums.slimdevices.com/member.php?userid=15838
View this thread: http://forums.slimdevices.com/showthread.php?t=87559

_______________________________________________
plugins mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/plugins

Reply via email to