At 8:34 PM -0000 6/8/99, Richard Light wrote:
>I was just pondering over the potential role of TEI in the XML world.
>One thought is that the semantics of TEI elements are generalized, wide-
>ranging and well-documented.  Couldn't this be used to help with the
>rendering of arbitrary XML documents?

I would think, not so much the rendering as the structuring itself -- TEI
of course doesn't standardize rendering, and one can think of lots of ways
to sensibly render many of the TEI constructs. But I agree the semantics
ought to be more widely adopted, they cover an awful lot of what people
want to do.

>First, invent (!) the idea of a "TEI architectural form" which
>application designers can apply to their DTDs.  This simply consists of
>a convention that there is a reserved attribute (say TEIeqv) which
>specifies which TEI element type this element type corresponds to.  This
>can be declared as #FIXED in the DTD, so it never has to be entered in
>actual documents:

This has been discussed at some length, and seems to me to have value.
There is a subtle interaction to be managed, though, between HyTime's
notion of architectural forms, and W3C's notion of namespaces. Of course
one can now prefix any TEI tag with "tei:" or some such, and then use it
unambiguously in any other DTD (we should probably define a normative URI
for the TEI namesapce). AFs have the big advantage of allowing renaming
(where namespaces only let you *qualify* the name with a named context --
not at all the same thing); on the other hand, namespaces being from W3C,
will probably have widespread support much faster. Any thoughts on the


[log in to unmask];
Chief Scientist, Scholarly Technology Group, and
   Adjunct Associate Professor, Brown University