[vorbis-dev] More spec notes
Monty
xiphmont at xiph.org
Thu Aug 8 17:14:02 PDT 2002
On Fri, Aug 02, 2002 at 05:36:37PM -0400, Kenneth C. Arnold wrote:
> A few more notes on the specs -- I just spent a few minutes and did up
> floor 1 header decode. Short and sweet, but still important.
>
> Monty, please read these and my several earliers -- I'm trying to
> improve and validate the spec for your codec, and though I understand
> you were taking a break before, a reply would be appreciated. Note
> (again, if you read my other emails) that if you help me get CVS
> access restored, I can make some of the changes myself instead of
> bugging you for everything.
>
> Another person has already expressed interest in undertaking a similar
> project to mine. For that and other reasons, I'd also like to commit
> what I have so far to CVS.
>
> Here goes for this round:
>
> vorbis-spec-floor1.html
> * header decode
> - step 5 - so maximum_class + 1 iterations?
correct
> - step 9/10 - what to do if subclasses == 0 (or does it matter? I don't think so)
Ah, what do you know. Spec bug, but not where you thought:
There's always at least one class if there's at least one partition.
If there's not at least one partition, there are no classes. So, step
2 should be:
2) [maximum_class] = -1
> - step 12 - really subtract 1? why not store it minus 1 in the stream?
"Because it isn't" :-)
> - step 22 - the first assignment writes to X_list element j+values = 0+1 = 1,
> overwriting the previously stored element there -- how big is X_list anyway?
You're right, floor1_values should start at 2 in step 17.
The X_list is calculated from partition classes and order. If you
have three partitions of class 0 1 and 1, then it would be 2 + size of
class 0 + 2 * size of class 2. The total size of the list isn;t read
anywhere, it has to be counted.
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