Theoretical ruminations on SXML and XML

The SXML version of next XML fragment

<em>important</em>

is

(em "important")

and this generates a mapping between both representations.

Now take the SEXPR (root 5)

Using above mapping, this would be translated to XML (ignoring tokenization)

<root>5</root>

This was the way taken by the w3c in the original HTML math draft.
However, the current MathML 2.0 specification uses (again ignoring
tokenization by brevity)

<apply><root/>5</apply>

MathML authors claim several advantages using this last content model.
Then the (more or less exoteric) question is, would the SXML

(em "important")

be encoded as

<apply><em/>important</apply>

rather than traditional

<em>important</em>?

That is, are there advantages on copying the MathML 2.0 content model for
example in some future XHTML version?


Juan R.

Center for CANONICAL |SCIENCE)