[vorbis-dev] Return of the Son of MIME type

[Maciej Stachowiak]

Ralph Giles <giles at ashlu.bc.ca> writes:

> 
> Monty and I discussed this a bit today; I thought I'd post a progress
> summary.
> 
> We've both found the arguments you (Maciej and Rakholh) have made
> convincing, but it remains to develop a technical solution.

Excellent! I'm glad we could provide useful input on this issue.

> 
> What's not clear is how to arrange for this to be efficiently testable. As
> I said before, the only way I can see mime-magic working is to key on the
> filename extension, or to guarantee that the first substream will be a
> special-purpose, binary table effectively adding a header to the
> headerless ogg. We're not really willing to do either of those.

Why is the second approach you mention a problem? You can always make
some default assumptions if this initial stream is missing. While
being headerless is a laudable goal, I am not sure how the metadata
header you describe below is different. But I am not up on all the
issues.

> That's about as far as we got. I've definitely added efficient
> type-determination the the requirements for the metadata header, but
> unless someone has a brilliant idea, a limited substring search will
> probably be required. :-/

Substring search in a range, or unlimited substring search in the
whole file? I think with a proper design for the metadata header will
make it really easy to do efficient sniffing. A fixed offset is really
best for overall efficiency.

> We still wish you wouldn't fork the mimetype, if only because we don't
> know how we're going to do this yet. The only ogg files we're generating
> now are "degenerate" in that they contain only a single vorbis substream.
> However, since "efficient type determination" almost certainly means
> putting the stream description metadata substrem (whew!) first, looking
> for 'vorbis' at byte 29 like I initially suggested is probably not
> the best idea. Associating application/x-ogg with the initial 'OggS', and
> handing that to ogg123 by default strikes me as the most future-proof.

I'd agree that we should stick with "application/x-ogg" and "OggS" as
the magic string for now. I will advocate this approach for the time
being.

 - Maciej

--- >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.