<snip/>
> Actually thinking about it, you can't just XInclude another
> WSDL infoset
> verbatim, 'cause the resulting Infoset wouldn't match the WSDL schema.
> In theory you could XInclude all the [children] of
> wsdl:definitions. But
> in practice, in the face of wsdl:import and wsdl:types you'd have the
> same problem. So I think something along the lines of
> xsd:include is the
> way to go, it's a component level include rather than an infoset level
> include.
>
Same rationale that schema used for doing xsd:include rather than xinclude.
The issue about including just children of a top node is easily resolved
with a form or attribute of xinclude that says "just children". This is one
of the reasons why XML External Entities don't have a root node. It's also
been a regular source of debate whether or not XML should have a form
without a root node. At any rate, we are where we are.
This doesn't solve the issue that there are many different children nodes
that have to be included in different orders. Which raises an interesting
question. If the structure of wsdl is not optimized for syntactic
modularity, what is the structure optimized for? Recreating Yet Another XML
Inclusion proposal probably shouldn't be done without some thought for
alternatives.
Cheers,
Dave