CNXML Flavors and the Spec
Document Actions
Submitted by
cbearden.
on 2005-11-15 15:11.
CNMXL 0.6 (like 0.5) will have four variants, depending on whether or not MathML and QML are used. How to reflect this in the specification?
Then there's the question of which schema flavor to use as the container for the spec text. The CNXML+MathML+QML flavor seems to be the best candidate, since it incorporates all the language features we use. However, when equation is redefined in the top-level schema file, it is redefined in a way that excludes the prior definition. Thus
<define name="equation.content">
<group>
<optional>
<ref name="name"/>
</optional>
<oneOrMore>
<choice>
<text/>
<ref name="media"/>
</choice>
</oneOrMore>
</group>
</define>
becomes
<define name="equation.content">
<group>
<optional>
<ref name="name"/>
</optional>
<ref name="mathml_rng"/>
</group>
</define>
and the first definition is discarded. Therefore, if the content model for an element in the spec is supposed to be autogenerated from the schema, then the spec won't know anything about the fact that in some flavors equation can take a mixture of text and media tags as its content, instead of a single MathML math element. So how do we get the auto-generated content model to reflect the fact that, under some circumstances equation? Do we modify the top-level schema to permit the mixed text/media as an alternative to m:math, so that the correct content model can be auto-generated for the spec? Indeed, how do we handle the question of the four flavors of CNXML 0.6 in the writing of the spec? The CNXML 0.5 recommends using MathML to encode the content of equation elements, but in the context of the exercise element it ignores QML. Should we explicitly refer to the different flavors of CNXML in the new spec, or should we just indicate that MathML can be used in equation elements, e.g.? Since we want to do away with the different schema flavors in the next version and validate based on the namespaces present in a given document, I'm somewhat inclined to think that we should treat the question of the different flavors in the new spec as we did in the 0.5 spec. It's kind of a fudge, but it reflects the direction in which we want to take CNXML. In any case, the specification should have a section or subsection displaying and explaning the four CNXML document type declarations.
Re: CNXML Flavors and the Spec
