Any Chance of Giving XSLT the Ability to Parse Attribute Val

One day, I had a crazy idea of wanting to transform XML to XSL using XSLT.
And then depending on the actual output device, use XSLT to transform XSL to
HTML, PDF, etc. When I looked at the XSL draft, I realized that quite a bit
of information is in the attributes.

Perhaps it is still possible to use XSLT to transform XSL to HTML (so that
most browsers can still show XSL document without a client-side renderer, or
even a client-side renderer that is an XSLT engine with a fixed XSLT
document for transforming XSL documents only). I am just thinking about the
general idea of recognizing the values of the attributes.

I can imagine the basic idea now: An XML formulation of BNF or EBNF with the
template-style constructs for declaring what to do when a syntax construct
is matched, just like most of the compiler-compilers we have right now.
However, to avoid needing an XSLT compiler, we can use Early's algorithm for
dynamic recognition for the grammar. Since that algorithm has a big-O of n
cubed, it is not as fast as the linear time algorithms we can have when we
use a compiler, but it is not too bad for small grammars that I imagine are
all we need in most cases.

I guess I dream too much. I dream that XSLT can be used for transformation
of XPointers and XLinks, and later on XML Schema, RDF, etc. If we have the
Early algorithm built-in, I guess there is no stopping us from doing
transformation on text as well. Ok, I do dream too much.

Khun Yee


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list