[
Lists Home |
Date Index |
Thread Index
]
Henry S. Thompson" wrote:
>
>...
>
> That was certainly the goal. If you _do_ depend only required
> sub-parts/attributes, and _don't_ access sub-parts by working backwards
> from the end, you will always win regardless of xsi:type.
Is it that simple? What if I have a content model like
<!ELEMENT a (b,c)+> (but expressed in XML Schema!)
Can someone extend it:
<!ELEMENT a' ((b,c)+,c,b)>
If so, that could really confuse most element-triggered processing
specifications.
Paul Prescod
|