[Ffmpeg-devel] Native ogg demuxer

[Måns Rullgård]

matthieu castet <castet.matthieu at free.fr> writes:

> Hi M?ns,
>
> M?ns Rullg?rd wrote:
>> matthieu castet <castet.matthieu at free.fr> writes:
>>
>>>What's the point of doing external sub-plugin that use global
>>>variable instead of static variable and if conditions ?
>> Cleaner design, IMHO.
>>
> Yes but you could have all the sub-plugin in the same file and make it
> static. This is more or less what do liboggz

I fail to see a significant difference.

>>>Also now the parsing of the extrada (comments, codec init, ...) is
>>>done twice : once in the demuxer, once in the decoder.
>> The codec does not, and should not, touch the comment section.
>>
> Actually if you look theora implementation it read and skip and the comment.
> So you sugest to not put the comment in the codec extradata (2nd header
> in vorbis and theora) ?

Libvorbis requires the comment section, but doesn't actually use it
for anything.  (Yes, we all know it's stupid.)

>>>There are usefull only if you want remuxer ogg in another format. But
>>>actualy there no standart way to store the extradata in other muxer (I
>>>know only vorbis in ogg in avi, and there still the ogg layer), and
>>>these informations could be still overwritten by the decoder.
>> There are lots of matroska files with vorbis audio, and I've heard
>> mention of storing vorbis in nut.
>>
> So what's the standard way of storing vorbis extradata in matroska ?

http://www.matroska.org/technical/specs/codecid/index.html

-- 
M?ns Rullg?rd
mru at inprovide.com