[ogg-dev] final changes to mimetypes rfc

ogg.k.ogg.k at googlemail.com ogg.k.ogg.k at googlemail.com
Fri Aug 15 04:03:25 PDT 2008

>> Just one small question, 10.3: "In particular, .ogg is used for Ogg
>> files that contain only a Vorbis bitstream, while .spx is used for Ogg
>> files that contain only a Speex bitstream."
>> Should that be a "Vorbis I" bitstream, or is it intentionally left open?
> Technically it is a Vorbis I bitstream, but I think we should leave
> that flexible. Any other opinions?

I read the "Vorbis I" definition to be a (physical) stream composed of a single
(logical) Vorbis stream, so I'd say the wording implies it is Vorbis
I. If it is to
be left open, the wording might need to be changed to remove the implication
(eg, ".ogg is used for Ogg files that do not contain non Vorbis bitstreams", or,
to avoid the double negative, ".ogg is used for Ogg files that contain only
Vorbis bitstreams", though that last one could be read as directly allowing
several Vorbis streams, rather than hinting).

Actually, since libvorbisfile will happily play a Vorbis stream alonside a non
Vorbis stream, but won't (AFAIK) play the second Vorbis stream within a
stream composed of (Vorbis+Vorbis), the former is probably closer to the
".ogg" idea than (Vorbis+nonVorbis). This would depend on the number of
other implementations that may be better the other way (I'm not counting any
shipped-on-silicon implementations, since these would likely fail on both in
the first place).

More information about the ogg-dev mailing list