RE: NMTOKEN elements?

>> This means that you are free to use NMTOKEN as an element value but
>> your instance documents will not conform to the XML 1.0 spec.
>> 
>Just a bit of clarification here.  Your instance document will still
>"conform" to the XML 1.0 spec, in the sense that it is still a well-formed
>document.  The point of the "for compatibility" clause is that if you
define
>the text content of an element to be of type NMTOKEN (or ID, IDREF, etc.)
>then there will not be a DTD equivalent for your schema (i.e., it will not
>be possible to translate your schema into a DTD which validates the same
>instances).

I understand. I was just curious if the issue of a parser's required
behavior with NMTOKEN elements had been addressed--that is, if assigning an
element a type of NMTOKEN (as an example in section 4.3.3 of Part 1 of the
schema spec does) would actually buy you anything. 

thanks,

Bob