[vorbis-dev] Problems compiling Mac Libs

Kevin Marks kmarks at apple.com
Mon Apr 30 10:53:17 PDT 2001



At 10:40 pm -0500 29/4/01, Steve Nicolai wrote:
>I've got this one tracked down, it's a CodeWarrior optimizer bug.
>You can't compile seed_chase() in psy.c with the global optimizer
>at level 3 or 4.  The bug causes the function to smash the stack.
>The bug is in both CodeWarrior 5.3 and 6.0.

Ouch! Do tell MetroWerks about it - I've found them very responsive 
in the past (before I worked for Apple they once gave us a dev build 
that fixed a compiler bug to help us ship our app).

>  > There isn't currently a way to make a VBR sound compression component
>>  in QuickTime. Existing ones (such as PureVoice) work by padding to
>>  the maximum length for a block, and then have the extra bytes
>>  stripped off by the packetizer before streaming or by the exporter
>>  before writing the output file. I'm not sure how amenable Vorbis is
>>  to this approach.
>
>Is this stripping done in a generic way, or does the packetizer and
>exporter have a list of codecs that it needs to do this for?

To do this the packetizer need to know the format, so the QT 
packetizer can't do it. Only the vorbis packetizer could.

>  I.e.
>how does the exporter recognize this and calculate the correct size?
>Is it just returning an ExtendedSoundComponentData with the buffer
>size filled in?  I'm just trying to save myself some time reverse
>engineering how PureVoice does it.

PureVoice only do it in the packetizer, as they don't have an 
exporter. As the exporter writes the file to the dataref directly, it 
can decide how to do it.

>  > If you put the codebooks in a SampleDescriptionExtension, you should
>>  have a sensible upper bound on the packet length. There is brief
>>  discussion of truncating packets to fit a bitrate on the Vorbis
>>  website, but I don't know if this has been tried.
>
>I'm planning on putting the codebook in a SampleDescriptionExtension,
>so this mechanism is attractive to me.
>
>The general concensus on peeling seems to be wait for chained
>codebooks, it will be much easier.

--- >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