RE: is that a fork in the road?

[Sean McGrath <sean@digitome.com>]

At 03:12 PM 3/2/01 -0600, Bullard, Claude L (Len) wrote:
>We aren't forking.  We've known forever this problem
>existed.  Again, the grove guys weren't noodling for
>their edification and Newcomb has repeatedly stressed
>in this forum that a simple DOM API wouldn't cut it
>in the long run if one wants to do all of these ambitious
>things.

The grove stuff could be the fork in the road. I think
groves/property sets/ abstract infosets accesible
only by APIs cos XML/SGML is just syntax is a
non sequiter. A towering intellectual achievement
to a *man made* problem.

We need to dare to speculate that the world can be
a simpler place and see where that gets us.

ISO 8879 was written in stone and all that
groves, architectures, hytime stuff had to
build on that stone rather than re-visit
its foundations.

XML 1.0 is written in stone but it is a much
smaller stone (and all syntax!) and its higher
layers are still on the drawing board.

Right now with XML, we have a chance - never to
present itself again - to learn from the SGML
experience and dare to ask what alternatives
are out there for the theoretical foundations
of all this.

Layered architectures have their limits - so do
intertwingled architectures. Thanks to Steve Newcomb,
Charles Goldfarb, Eliot Kimber etc. we can see
into the crystal ball where that road may lead.
It appeals to some. It does not appeal to
others. It certainly does not appeal to me.

The layered architecture approach has a much fuzzier
crystal ball. Henry Thomson was doing some great
stuff with it when I first me him at an SGML conference
in Munich, Germany about 5 years ago. Omnimark Technologies were
onto a great idea in my opinion with Micro Document
Architectures but then dropped the ball.
Then there was Metamorphosis, COST and more
recently fascinating stuff going on with TREX/RELAX.

There is tons of stuff here we can learn from. Tons
of stuff we would be foolish to ignore. Just because
XML picked up a lot of good stuff from SGML, it
does not follow that we should blindly pick up
all the other stuff and start tacking it onto
XML.