Re: SV: SV: What's a valid instance...James Clark

Bryan Rasmussen wrote:
> Thus what one ends up with is XML like this
>
> <a:ProblemDomainDocumentElement xmlns:a="http://example.org/a"
> xmlns:b="http://example.org/b">
> <b:ProblemDomainClass>something or other here</b:ProblemDomainClass>
> </a:ProblemDomainDocumentElement>
>
> now I have seen a lot of this kind of structure over the last few years and I
> really can't remember seeing much of it before XML Schema became the de facto
> standard for doing validation in XML. And I think it is a response to this
> lack of being able to define a validation root. But you may have another
> opinion. 
>   
In some ways yes, the root validation is cause for this type of design.  
Personally, I like the global element concept not just for the use of 
multiple schemas that can reuse various modules, but because used 
correctly and dilligently it makes defining semantics much easier and 
resuable across various schemas.

The issue you bring up is compounded by the over zealous use of 
namespaces, where people try to come up with the perfect design.   There 
comes a point where real word, and theory just don't mix.   An example, 
in some NDR specs for representing a model in schema, it has been stated 
that every code list must have it's own namespace.   This in itself 
could lead to an application or a schema having to support litterally 
hundreds of namespaces.   When in practical reality, code lists are 
typically stored in one common namespace and reused.   It's the 
difference between what looks like a perfect design, and what is good 
enough to work to get the job done.

Namespaces themselves aren't bad, it's how they get implemented that is bad.

Dave

Received on Friday, 23 March 2007 20:00:30 UTC