[whatwg] on codecs in a 'video' tag.

[Dave Singer]

At 23:29  +0300 2/04/07, Henri Sivonen wrote:
>On Apr 2, 2007, at 23:13, Dave Singer wrote:
>
>>At 23:07  +0300 2/04/07, Henri Sivonen wrote:
>>>Some implementations only support AVC level up to a magic level 
>>>that you have to know.
>>
>>are you telling us that all implementations of Ogg and Theora can 
>>play audio and video up to any bitrate, screensize, channel count 
>>etc., without dropping frames, getting behind, decoding badly, or 
>>other limits?   That would be quite an achievement...more 
>>impressive than getting a quart out of a pint pot...
>
>Of course I am not. But hitting CPU limits is different from hitting 
>magic profile parameters (even if in theory the parameters in theory 
>protect the hapless user from hitting the CPU limits).

Well, I rather suspect that there are verions of the open-source 
codecs, no?  Theora is suggesting it might get to do interlace 
eventually...what then happens to deployments of the old version, 
when given interlaced content?  Chances are, it won't work;  there 
are new 'tools' exercised by interlaced content.

You can call this versioning, you can call it profiling, but unless 
you issue a perfect final product first off, you have to cope with 
it.  If you are arguing that MPEG makes *too much* use of profiles, 
then maybe that's an argument to have, but criticizing the system 
that formalizes 'you need to implement this package of features to 
play this bitstream' is, again, merely prudent and *improves* 
interoperability (or at least determining whether things will 
interoperate).  Does Ogg/Theora have a 'required features' or 
'required version' in the bitstream?
-- 
David Singer
Apple Computer/QuickTime