[vorbis] Rgenerating headers

[Beni Cherniavsky]

On 2002-11-24, Michael Smith wrote:

> On 23 Nov 2002, Peter Maersk-Moller wrote:
> >
> >If the encoder andthe decoder can ask the vorbis library for a major and
> >minor version number, then if the version number of the encoder is
> >passed out of band to the decoder, then the decoder can look in a sort
> >of header cache for the appropriate header and only download (or generate new)
> >headers if they are not present in the cache.
>
> Version number is insufficient, since different encoders can exist,
> etc. A better way to do this would be to transmit a hash (MD5, or
> whatever else you prefer) of the codebook header. THIS can easily
> be looked up in your cache (generating new ones isn't possible with
> this approach, but that's probably a good things, since it's horribly
> failure-prone to do so).
>
> With this, the player only needs to obtain the large header once, then
> cache it locally. You can do the same with the primary header, but it's
> small enough that it may not be neccesary.
>
See also the thread on vorbis-dev starting with this message:

http://www.xiph.org/archives/vorbis-dev/200106/0057.html

It discusses more or less the same ideas.  A hash-based cache for
codebooks is definately possible, especially if its use is constrained to
on-the-wire and full headers are expanded if the downloaded file is stored
locally (long-term storing codebook-less files is not very good since
availiability of the refered codebooks can't be unsured).  Nothing like
this is implemented however...


-- 
Beni Cherniavsky <cben at tx.technion.ac.il>

What's lower level than machine code?  A spreadsheet: not only addresses
are numeric and hand-allocated but also all loops are hand-unrolled and
all calls hand-inlined... (and macros are unheard of, of course).

<p><p><p><p>--- >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-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.