I'm arguing that it does matter what's in the spec, insofar that it
should
match what implementations do.
Can we agree to disagree?
We've narrowed codecs down to two. The spec could say that UA which
supports <video> MUST implement at least one of Theora or H.264. All
vendors can comply with that, and that's better than not specifying
any codecs at all (e.g. doesn't allow browsers to support WMV only).
Similarly, authors publishing <video> MUST put at least one source in
Theora or H.264, SHOULD publish both. That's probably what authors
will have to do to achieve interoperability in current situation.
These requirements won't ensure full interoperability - that is not
possible with the impasse we have - but will match implementations,
make situation clearer for authors and disallow even less
interoperable implementations.
--
regards, Kornel
