While I agree with the overall content of Nick's message, I don't like
the way this implies you *can't* get good visual display with
In a structured document, the document is a tree of nodes. To display
the document, one walks down the tree, processing each node, and there
is no inherent limitation on the amount of processing performed on
each node, so one if theoretically unlimited in what one can do.
Particularly if one uses the boxes and glue model, you can build very
nice looking documents indeed.
That said, I do agree that while you can get good results from such
systems, at the finest level, it is usually hard to match the output
of a composition engine with runtime formatting like that in HTML (or
even DynaText(tm)), so yes, there is certainly a place for various
different viewing technologies.