Re: Encrypted Media proposal (was RE: ISSUE-179: av_param - Chairs Solicit Alternate Proposals or Counter-Proposals)

On Fri, Mar 2, 2012 at 11:19 AM, Tab Atkins Jr. <jackalmage@gmail.com>wrote:

> On Fri, Mar 2, 2012 at 10:14 AM, Glenn Adams <glenn@skynav.com> wrote:
> > On Fri, Mar 2, 2012 at 4:07 AM, Silvia Pfeiffer <
> silviapfeiffer1@gmail.com>
> > wrote:
> >> On Fri, Mar 2, 2012 at 8:41 PM, Henri Sivonen <hsivonen@iki.fi> wrote:
> >> > On Thu, Mar 1, 2012 at 11:57 PM, Mark Watson <watsonm@netflix.com>
> >> > wrote:
> >> >> The underlying content protection systems are things like PlayReady
> >> >> (from
> >> >> Microsoft), Widevine (from Google) and Marlin. Adobe have something,
> >> >> but I
> >> >> don't know what they call it.
> >> >
> >> > As a co-proposer, does Microsoft plan to integrate PlayReady into IE?
> >> > As a co-proposer, does Google plan to integrate Widevine into Chrome?
> >> > Do the co-proposers plan to make their CDMs available to other
> >> > browsers? Do the co-proposers plan to provide APIs that'd allow adding
> >> > other CDMs to their browsers?
> >>
> >> I'm confused. I thought the whole idea of the proposal was to just
> >> provide an API for adding CDMs into browsers such that when you have
> >> the library installed on your computer, any browser is able to make
> >> use of it, no matter if it's Google's Widevine library or Microsoft's
> >> PlayReady - e.g. Firefox would be able to make use of these and any
> >> other CDM library. There would be no need to implement something
> >> additional into browsers. If this is false, somebody better clarify
> >> how else it is supposed to work.
> >
> >
> > Yes, that is indeed the idea. But Henri keeps wanting to drill down into
> the
> > CDMs themselves, which is really out of scope for the proposal.
>
> Given that the CDM is a necessary component of the mechanisms
> described in the spec, the details of the CDM are extremely relevant.
> There's no way to implement the spec without involving a CDM.


Of course, at the black box level for the purpose of defining the API
behavior of the CDM, it is necessary to define semantics. However, a single
instance of a no-op CDM (that translates plaintext to plaintext) would be
sufficient to verify that behavior and test the API. Any other details of
the CDM implementation, including licensing, trade secrets, patent
encumbrance, etc., should be out of scope. At least that's my opinion and I
would guess the opinion of the proposers.

I understand, however, that a browser implementer may have interests in
what goes into a particular CDM if they have to put it in their product,
but I would suggest that is out of scope of discussing the utility or
functional usage of a CDM as accessed through the proposed interface.