Re: The subject line is irrelevant these days

On Wed, Oct 23, 2013 at 8:14 AM, Henri Sivonen <hsivonen@hsivonen.fi> wrote:

> On Wed, Oct 23, 2013 at 5:22 PM, Mark Watson <watsonm@netflix.com> wrote:
> > That's always going to be a business decision based on the costs and
> > benefits of supporting an additional platform.
>
> That this is the case is part of the mismatch between EME and how W3C
> specs normally are supposed to work and a source of a lot of grief.
>

I understand. But it's the same with <object> and plugins (for all
applications of those, not just DRM). I assume you're excluding those from
"normally supposed to work" and that's probably where there's a divergence
of opinion. One could argue that those windows on proprietary functionality
are important parts of the platform that have played an important role in
enabling services on the web but which have some disadvantages (notably
security and interoperability). And we should address these disadvantages
by moving *as much functionality as possible* from that realm into the
standardized interoperable realm.

It's just that there remains some widely-used functionality which cannot be
completely migrated. The fact that we've migrated everything else doesn't
change the status of that remaining piece as something which existed before
on the web and will continue to exist.

If independent interoperable implementation of the entire stack (not just
the W3C specification) is an absolute requirement, not just a goal, then
first <object> should not just be deprecated but outright removed from the
specification - as should the ability to use non-RF codecs - and second,
yes, EME doesn't work either. In that case we should just not have APIs at
all to any unspecified or non RF functionality. The fact that W3C has not
done these things suggests that there isn't consensus that this is a hard
requirement.

On the other hand, we can be more practical, note that goals and hard
requirements are different things, note that formally the goal applies to
specifications not the whole stack, note that there is precedent in
<object> and therefore continue the project of deprecating <object> by
providing constrained support for that last remaining functionality.

...Mark


>
> (Unfortunately, even sites that don't use DRM don't live up to the
> ideal of sites not having to do anything in particular to support a
> new browser implementation or a port to a new platform and break
> things through UA sniffing, for example.)
>
> --
> Henri Sivonen
> hsivonen@hsivonen.fi
> http://hsivonen.fi/
>