Two comments on your abstract of MHEG:
(1) As a standard that wants to define an object model, I think it
would be much more helpful to have the definition in terms of
languages designed for that purpose -- for example, CORBA's IDL or
ILU's ISL -- rather than ASN.1 or SGML. These object meta-languages
have the benefit of being easily compilable into a variety of popular
programming languages, using existing tools.
I realize that the focus of your effort is on the ability to
interchange persistent multimedia object repositories -- something
not properly covered by current object metasystems like CORBA (as far
as I know...). But why invent something specifically for multimedia?
Shouldn't there be a general standard for persistent object
interchange, derived directly from object meta-language
specifications?
(2) I'm not so sure that we want HTML/SGML to turn into the master
meta-language for describing the complete dynamic relationships
between multimedia document components. I'd rather think of HTML as
one of the many specialized formats for individual media components
(in this case static, hierarchically-structured text), and let a
dedicated, secure, object programming language take over the
management of dynamic and non-hierarchical interrelationships between
the various document components.
In other words, what would seem more practical to me is a set of
generic object models for linking various formats to each other,
without designating a single "master format". (This is already being
done in an ad hoc way for many formats.)
For example, the model would define, in generic terms, how an HTML
document could contain links to a Java script, which could then in
turn act back on elements of the HTML -- all without Java having to
have specific knowledge about HTML (just that it's a hierarchical
text medium), and without HTML having to have specific knowledge
about Java (just that it's a scripting language which is willing to
be given control of various components of the document).
This would be implemented through format-specific bindings of the
appropriate generic object model for each type of component. I don't
see that as a portability disadvantage compared to (what I presume to
be) the MHEG model, since MHEG must surely defer to format-specific
decoders for individual media anyway. I'm just adding the job of
"object model binding" to the responsibilties of the decoders.
--------------------------------------------------------------------
Paul Burchard <burchard@math.utah.edu>
``I'm still learning how to count backwards from infinity...''
--------------------------------------------------------------------