> But no rationale is given for imposing the restriction
> except that the a:documentation element is intended to contain "text",
> yet "text" cannot be uncritically identified with mere sequences of
> Unicode characters.
I think the rationale is that this is a DTD Compatibility spec; the
justification for including a:documentation in the spec is that
a:documentation corresponds to a comment in a DTD and comments in a DTD
contain mere sequences of Unicode characters.
There's certainly scope for a future spec to standardize a richer
documentation facility. However, I don't think it's as simple as saying
the a:documentation element can contain arbitrary sublements. That doesn't
seem sufficient to me from an interoperability. Imagine writing a GUI for
editing RELAX NG schemas: how is it going to be able to display the
comments in a useful way unless it knows what the subelements mean? Maybe
a subset of XHTML is the way to go for a future spec.
James
