On Tue, Feb 12, 2008 at 07:55:49PM +0100, Michael Niedermayer wrote:
Iam listing all issues which have not been solved and all possble solutions. You just asked today or yesterday if mpeg had some document about the reasoning behind some of their decissions.
It would be unprofessional if id simply not list a possible solution because we maybe decided 2-4 years ago that its not a good choice.
OK.
And yes i agree that things just needed for raw dont belong in the stream header. And IIRC the consensus 2-4 years ago was to use the fourcc though i dont remember all the reasoning anymore but we had a pretty good argument, i just dont remember it :)
OK.
Now you seem you prefer extradata, after alban suggested it, this is as close to the consensus 3 years ago as using the stream header. Besides its one of thee things which would be a nightmare to implement in practice and would need tons of special case.
I agree it has some major practical issues, such as the fact that no existing PCM "decoder" is prepared to handle extradata. I'm totally happy with the fourcc approach too; I don't have much preference one way or the other here.
If(raw){ extract all the data out of extradata make extradata disapear somehow as its nut specific and not the codec private area }else{ }
Yes.... Rich