RE: XML Schema: DOs and DON'Ts

["Arnold, Curt" <Curt.Arnold@hyprotech.com>]

> Kohsuke KAWAGUCHI wrote:
> So if you have the general concept only and doesn't have any specific
> details of the derivation-by-restriction, then you would 
> probably think
> this is a valid restriction, which was my intention.
> 
> My reasoning was
> 
> - when it comes to the derivation-by-restriction, the general
>   understanding is not enough. You need to learn many specific details
>   of the derivation by restriction.
> 
> - so compared to its merit, it doesn't worth learning.

My feeling was that other than trivial cases, it was so complex
that interoperability of implementations should not be expected.
I guess we/they will find that out in the development of 
the Schema conformance suite.  I would suggest avoiding it for
general use and use named content and attribute groups.

Substitution groups are also problematic for complex schemas.
I recently wrote a fairly complex schema for defining DOM
conformance tests and my original approach was to use a
"statement" substitution group to represent all the statements
in the language, trying to define all the complex types in 
use as a derivation of the base "statement" complex type was
extremely cumbersome and difficult.  

I eventually bailed and used an named group though I 
would have liked to have had the implicit membership
 of substitution groups. I was also having serious 
difficulty defining a substitution
group that contained both elements with simple content 
and elements with complex content.  Maybe it is possible,
but I couldn't make it work and I'm not a novice.