afinney> consider a model with separate namespaces...:
afinney>
afinney> <parameter id="p"/>
afinney> ...
afinney> <species id="p"/>
afinney> ...
afinney> <kineticLaw>
afinney> <math>
afinney> <apply>
afinney> <times/>
afinney> <ci> p </ci>
afinney> <ci> p </ci>
afinney> </apply>
afinney> </math>
afinney> </kineticLaw>
afinney>
afinney> which p are the ci elements refering to?
Actually, there is a way to make this work. It is something
that is closer to what Nicolas is advocating (viz., tagging
the elements with the namespace they're coming from).
MathML allows 'ci' elements to have a definitionURL
attribute. (See Section 4.4.1.1 of the MathML 2.0 spec.)
In principle, we could have something along the lines of
This would be more in line with XML and MathML than Nicolas'
suggested approach of using "par:p", etc. (which I know
Nicolas was only proposing as a starting point for
discussion and not a final implementation).
I'm not advocating this -- in particular, I haven't though
through the other possible implications of assuming all the
symbols are in separate namespaces -- but, technically, it
does appear possible to use MathML constructs to track the
different classes of objects referenced in MathML
expressions.