Re: interleave and strings

[James Clark <jjc@jclark.com>]

Kohsuke KAWAGUCHI wrote:
 
> I think the spec should describe constraints as currently it does, too.
> But I feel it should be based on a mathematical model, otherwise it will
> be unnecessarily ad hoc.

I don't think the restriction on TREX is ad hoc.  It is based on a very
simple principle: typed data should be delimited as an element or
attribute.  In other words if you have a string representing an integer
or a decimal or some other typed information, that string should be
explicitly marked up.  Something like

<element name="foo">
  <data type="xsd:int"/>
  <element name="bar">
    <empty/>
  </element>
  <data type="xsd:int"/>
</element>

is bad because the two int's are not marked up explicitly. I think this
restriction is one that is natural and inoffensive to markup users, and
simple to grasp intuitively.  XML Schema has the same restriction.

James