Vorbis dev - Feb 2000 - Decode questions

Sorry, to interrupt the license flame war...

I downloaded the source from CVS as of 11am this morning and there are
a couple of issues that I ran into:

1) I built the package and then in the examples dir I executed the
following commands:

# cdparanoia -B 1-1  
# ./encoder-example < track01.cdda.wav > test.vor
# ./decoder-example < test.vor > out.wav

The rip and the encode both work fine. test.vor is 22280990 bytes (!)
and track01.cdda.wav is 57998012 bytes long. decoder-example sez:

  Track encoded by encoder_example.c

  Bitstream is 0 channel, 0Hz
  Encoded by: Xiphophorus libVorbis I 20000121

  Floating point exception 

I haven't bothered to dig into the source to find out what is going on
-- before I do that I wanted to inquire about the current status of the
codebase in CVS. Should this work? Or is this a case of operator error?

Next, I dug into the source for decoder_example.c and I saw that in the
decode cycle at the end the file does the double -> 16 bit unsigned int
conversion as the last step, outside the decoder. Are there any plans
to have the Vorbis synthesis layer take care of this step? I ask this,
since this decoder/convert step moves the WAV data across the system
bus a time more than is necessary. This could be avoided by simply
carrying out the integer conversion before the decoded data is written
to the output buffer.

I also get the feeling that this might be jumping the gun a little bit.
:-)

--ruaok         Freezerburn! All else is only icing. -- Soul Coughing

Robert Kaye -- robert@moon.eorbit.net  http://moon.eorbit.net/~robert

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/

> # cdparanoia -B 1-1  
> # ./encoder-example < track01.cdda.wav > test.vor
> # ./decoder-example < test.vor > out.wav
> 
> The rip and the encode both work fine. test.vor is 22280990 bytes (!)
I'm only adding the big residue compression now.  I expect to have (at a
minimum) 128kbps streams this week.
> and track01.cdda.wav is 57998012 bytes long. decoder-example sez:
> 
>   Track encoded by encoder_example.c
> 
>   Bitstream is 0 channel, 0Hz
>   Encoded by: Xiphophorus libVorbis I 20000121
Oop, sorry, I though I committed the fix for this... the residue header
unpacking is out of sync with the packing and the decoder example doesn't
check
the return value.
> I haven't bothered to dig into the source to find out what is going on
> -- before I do that I wanted to inquire about the current status of the
> codebase in CVS. Should this work? Or is this a case of operator error?
No, CVS captian error.  I committed a mildly broken version.
> Next, I dug into the source for decoder_example.c and I saw that in the
> decode cycle at the end the file does the double -> 16 bit unsigned int
> conversion as the last step, outside the decoder. Are there any plans
> to have the Vorbis synthesis layer take care of this step? I ask this,
> since this decoder/convert step moves the WAV data across the system
> bus a time more than is necessary. This could be avoided by simply
> carrying out the integer conversion before the decoded data is written
> to the output buffer.
That assumes that you always want the output in a specific, predetermined 
format (like 16 bit interleaved stereo, particular endianness).  OTOH, it's 
pretty clear that most people will, in fact, desire 16 bit PCM in host 
bytesex... so I guess I should add this, yeah.
> I also get the feeling that this might be jumping the gun a little bit.
> :-)
Nope, you're right on schedule.  I'm just a bit late as always.

Monty

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/