> -----Original Message-----
> From: Sean Russell [mailto:ser / germane-software.com]

<snip />

> Also, I wouldn't prefer to have contained classes be referenced like this.  
> I'd still prefer:
> 
>   <class name="A" id="1">
>     <class name="B" id="2"/>
>   </class>      
>   <class name="C">
>     <extends xlink:href="#1"/>
>   </class>
> 
> would be a mapping of:
> 
>   class A
>     class B
>     end
>   end
>   class C < A
>   end
> 

Except, where did #1 and #2 come from?  The arbitrary id attribute
values introduces information not found in the actual source, and such
information may not travel well.

...

> I suppose I'm not very creative.  I'd just map the Ruby structure to XML and 
> be done with it.

Yes, and the structure of the source code should contain enough information
to define unique identification values:  

  <class name="C">
    <extends xlink:href="#xpointer(someDocRoot/class[@name='A'])" />
  </class>


> 
> > As if anyone in their right minds would want a job that involved using
> > XML :)
> 
> Heh, it's unavoidable.  All jobs evolve until they require the use of XML 
> :-)

All too true .


James

> 
> -- SER
> 
>