[vorbis-dev] Oggenc ideas / source and request

Monty xiphmont at xiph.org
Sun Dec 10 17:13:57 PST 2000



> Hello everybody,

> what's new or modified:

Umm... first off, is there a reason you collapsed every source module
into a single file, or did you just not understand how to use it the
way it was?  This change alone makes the source impossible to compare
to any other revision.  That's not mentioning the scattered ad-hoc
changes in your newly-monolithic encoder that for which I don't see a
coherent rationale.

If you'd like to submit changes, please do so in patches.  Leave the
structure of the source alone.  If you don't know how, please ask.

> - wav-header and wav-data-len handling

What about the preexisting support was inadequate?

> - multichannel encoding

Also exists in oggenc.

> - header update (after encoding) with an average bitrate value (into the
>    vi.bitrate_nominal variable)

This breaks streaming, and the average bitrate value will be incorrect
if the file is edited (which potentially renders it even less useful
than 'absent'.)

> - 64bit floating point precision (maybe you say that it is not important,
>    but I'd like to compare with the 32bit verision)

Doubles are totally superfluous.  Play with it and you'll see it makes
zero difference and seriously degrades computational performance (not
because doubles are slower than floats, because they're bigger and
suddenly you're thrashing the L1 cache on virtually any processor).

> And I have a small request too:
> If somebody has a little time, please make an executable from this source
> with the same compiler witch was used to the beta3 encoder (for Win9x),
> because it seems my Watcom C compiler doesn't like some parts of the
> encoder (but the decoder works properly), and the compiled program
> makes  a bad quality ogg file (sounds like a bad mp3 file (underwater
> effect)

Finding and fixing the problem is the correct answer here, not
sticking to a single compiler.  Either the original source has a
compiler dependant bug (which I'd like to find) or you've introduced a
compiler dependant bug.  Until we establish which, I'd not consider
the changes for inclusion (aside for the other problems I've
mentioned).

> ps. I hope the program doesn't contain too many bugs

Not entirely reassuring...

Monty

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.



More information about the Vorbis-dev mailing list