Liam,
Again, I claim only a top-level understanding of SGML. If it can fully
express the container abstraction then that's great. It would be great to
see all sorts of implementations.
As for your second point, I think we have a different focus, and I try to
express this in my yet-to-be-finished report. Yes, I completely agree that
we need a solution to the metadata problem within the context (tools,
protocols, languages) of the current WWW. Quick deployment is a high
priority, if we are going to make a dent in the problem. However, I think
that it is VERY important that we look beyond the web (as it exists now)
and propose a open metadata framework for a less constrained (crippled!?)
infrastructure. What is so important about this workshop and its results
is the mixture of .edu and .com types and the mixed focus on what works now
and what we really might want for the future (both are a high priority and
both groups play an important role in plotting that future course).
Finally, I'm a little uncomfortable with your last point "people will
interpret the data in the way that's most useful for their indexing
software". My feeling is that the architecture should aim towards allowing
the originator embed as much explicit meaning in data as possible, and not
leave it "up to the reader" to figure out what its all about.
Regards,
Carl
----------
From: [log in to unmask][SMTP:[log in to unmask]]
Sent: Wednesday, April 24, 1996 8:54 PM
To: [log in to unmask]
Subject: RE: Syntax for Dublin Core: paper available
Carl Lagoze <[log in to unmask]> wrote:
> The concept of recursive containers of arbitrarily
> complex, typed objects (possibly referenced indirectly) is a powerful
> abstraction that might be implemented in a variety of ways.
Agreed. This is, in fact, what SGML is for :-)
> - Now I will violate my first point. While I don't pretend to be an
expert
> about SGML or MIME, my intuition is that both of these technologies are
not
> sufficiently powerful to fully express the abstraction. I think for some
> relatively simple examples, SGML and MIME but be entirely appropriate.
My
> prejudice, however, is to model this using CORBA or ILU and rely on the
> strong typing provided by the distributed object model.
The trouble here is that you're likely to end up with something that the
grass-roots barefoot-programmer software on the Web can't deal with.
Unfortunately, if that happens, we've failed.
E.g. Windows 95 doesn't come with CORBA, but it does come with HTML
software.
I actually think that packaging objects up in any way at all is a little
risky, and that we should certainly allow a one-top-level-object-per-file
granularity for those people for whom it makes sense. It's the simple
cases that we have to solve. We're not reinventing a distributed version
or MARC here :-)
SGML does, however, have object types. It's not too hot on methods, but
it's not clear that many explicit methods are needed -- people will
interpret the data in the way that's most useful for their indexing
software.
Lee
--
Liam Quin, SoftQuad Inc +1 416 239 4801 [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|