RE: Fallback mechanism for the datatype

James Clark wrote:
>I don't think we can dictate behaviour in these 
>circumstances given the braod range of scenarios in
>which RELAX NG might be applied.

Hands off and non-vacuous. A good way to be. -Mike