[
Lists Home |
Date Index |
Thread Index
]
Jonathan Robie wrote:
> A map is not the same thing as the terrain that it represents. Certainly, any
> real world object may be given one or more representations in a particular
> model, but the real world is not part of the model that represents it. And
> these representations can be contradictory, which is why both a physics based
> on Euclidean geometry and one based on non-Euclidean geometry are in
> widespread use for practical problems.
Hi Jonathan.
Clearly this is the day for deja vu:
http://lists.xml.org/archives/xml-dev/200108/msg00705.html
The point I wanted to make last year was that the marked-up syntax is processed
in order to elaborate specific semantics. A map is an infoset understanding of
the territory--a tightly constrained set of chosen semantics presented as a
base instance in its own right. For a great many subsequent operations that map
is an entirely suitable base instance, just as an infoset is for processes
designed to operate against it. Yet you understand quite clearly that for the
most undeniable hard physical realities the map is not a suitable base instance
because you cannot serialize or otherwise realize the physical terrain back out
of it. For memories of childhood or the smell of onions the desired mimesis
can, in fact, be processed from a map-like infoset base instance. That is a
principal function of art. For Alpha Centauri, a black hole, or even your
grandfather's axe, attempting to recreate the physical from the abstract plan
quickly indicates why it was counterproductive to abstract away in the first
place what appeared to be purely concrete quirks. So too, I would suggest, with
marked up text. In simplest terms, abstracting away anything from the syntactic
instance implies, as a first premise, some 'intended' use of the document, some
'expected' processing of it, or some other abstract view which, when carried
out in infoset creation, will necessarily curtail some semantic possibilities
of the base text. Those intentions and expectations form the semantics of the
Infoset, though John would claim there are none such. The real question is how
certain are we that what we expect will be done with a document is sufficiently
comprehensive that we will ab initio curtain any other possibility by asserting
that there is nothing of value to in the base syntax which is not provided for
in a well designed infoset.
Respectfully,
Walter Perry
|
