CSS, XSL and MathML. Some questions.

I have been monitoring these discussions for a
couple of weeks now primarily from a point of view
of determining how the mechanisms and proposed mechanisms
interact with MathML 1.0  In particular, I am
interested in what the right approach is to addressing
two problems outlined below.  I am particularly 
struck by the similarity of the problems that are
being discussed.  

I am curious how people see the following issues fitting
into the XSL discussions.

For the benefit of those of you unfamiliar with MathML,
the most relevant parts of it are as follows.
MathML is described in terms of XML.  It introduces
two families of elements - one "display elements" primarily focussing on
display issues and another "content elements" primarily focussing on
semantic issues.

Problem 1.
----------

Each of the XML elements have attributes.  Default
attribute values are specified, but these can
be over-ridden.

The attributes of the "display elements" affect the
visual or aural rendering.  The attributes
of the "content elements affect its "semantic" 
rendering.

There is a need for a style sheet like
mechanism that systematically sets the default
attribute values of the XML elements.  

Note 1.  This appears to be exactly the problem 
addressed by CSS, but CSS apparently only addresses
this issue for flow objects.

Note 2.  The "content elements" could be regarded as the
flow objects of a "semantic" rendering.  The actual
semantics are affected by the attribute values.

Note 3.  Just as for CSS, there is a need to reset
these attributes on a document wide basis.

Problem 2.
----------
For each content element there is a need to specify
the visual/aural presentation of a "content element.".

The solution goes beyond just the setting of attributes.  
In fact, it involves a possibly sophisticated "transformation" from one
expression tree to another.  The first expression tree is a "MathML
content element" based tree. The second is an "MathML display element"
based tree.  The trees need not be (and generally are not) isomorphic 

Note 1.  The MathML presentation elements are in some sense, flow
objects so the transformation is analogous to a translation from HTML
elements to flow objects.

Note 2.  The actual transformation of the trees requires some sort of
non-trivial scripting / programming capability.


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