Re: [codec] draft-ietf-codec-oggopus and "album" gain

[Gregory Maxwell <[e-mail]>]

On Wed, Aug 27, 2014 at 4:23 PM, Timothy B. Terriberry
<[e-mail]> wrote:
> Ian Nartowicz wrote:
>>
>> This seems like a step backwards to me.  That MUST is a requirement
>> that wasn't present before.  An earlier statement is "Virtually all
>> players
>> and media frameworks should apply it by default.", which I think is the
>> appropriate guidance.
>
>
> I agree with Ian. The point was to clean up usage of an un-capitalized RFC
> 2119 keyword that we all agree was not meant to be normative. We shouldn't
> be adding NEW normative requirements in its place.
>
> We have not to this point had any problem getting players to apply the
> output gain by default with the current spec language (and Greg Maxwell's
> test files that tell people their player is broken if they don't implement
> this). We've had to cajole a few, but once we have, I believe it has been
> fixed in every case.
>
> -With my individual hat on

Part of the reason for this is that the libopus as of 1.1 (IIRC) makes
it very easy to handle this— you just make a ctl and pass the content
of the tag, and libopus handles the gain internally... so no DSP for
implementers. (And it handles it in a way that expands the dynamic
range for fixed point).

Alas, not even RFC6919 has a "MUST unless you actually know better".