Re: trang and xsd-friendly schemas

We have been arguing this point on an internal list of the TEI - should 
we take
advantage of all Relax NG's power, or should be only use the portable 
subset,
because we know people will want DTD and XSD versions? Do you use Relax 
NG as
a schema language, or a notation of an interchange schema language?

James' idea is elegant, but I don't think it solves the problem of 
interleave, because
the outer schema will be more restrictive. ie if it says
  a = x, y, z
whereas the inner one says 
  a = x  & y & z
when what is the advantage of the inner one if it lets documents past 
which the
outer one rejects? where the inner schema is *more* restrictive (we had 
an example
where an attribute was only present if certain children existed), it works.

I am not sure you can solve this without having a clear plan of why you 
are writing
schemas, who uses them, when, what problems they solve etc.

Sebastian