[
Lists Home |
Date Index |
Thread Index
]
From: "John Cowan" <jcowan@reutershealth.com>
> What is new (except to Lispers) is the idea of a standardized intermediate
> syntax into which the surface syntaxes can be translated. We do not need
> one parser per vocabulary *per implementation*; it suffices to have a
> single standard translator from the specialized syntax to the general one.
Don't forget SGML short-references. You can normalize a short-refed
document into a fully tagged document.
Where SGML short tags go wrong in the XML world is that they disrupt
well-formedness: as well as the nice
<formula>a^2</formula>
which expands to
<formula>a<sup>2</sup></formula>
it also allows
<formula>a^2</sup></formula>
So any reintroduction of a short-ref-like mechanism into XML must
be as a post-process (or Schema process) with WF-XML in and
WF-XML out (augmented with other infotypes if you like).
WXS has its lists and regular expressions, and Simon's Regular
Fragmentations goes a little further. I don't think these go far
enough in their current forms.
Cheers
Rick Jelliffe
|