RE: NMTOKEN elements?

>We have an pending issue to consider what aspects, if any, of
>attribute value normalisation to import into XML Schema processing,
>and when we do we need to consider whether whatever we do for
>attribute values we will also do for element content.

Which will bring up an interesting issue: if a given piece of information
can be designated as being of type NMTOKEN and treated the same by the
parser regardless of whether it's an attribute or element, then one of the
key distinctions between attributes and elements (and hence one of the key
criteria in deciding whether to make something an attribute or a subelement)
is no longer relevant. 

Bob