xml-dev - Re: XML Schemas: ref'ing versus in-lining

[Curt Arnold <CurtA@techie.com>]

Henry Thompson wrote:

> A misunderstanding on your part: this is specifically allowed, because
> there is in fact only one tag-type binding in this local context.  In
> other words, the value of type _does_ matter.
> 

You are the master, however I didn't see it in a cursory look at part
1.  Is there a specific place in the document that the acceptance of
duplicate declarations is mentioned?

If it is, what is the criteria for an acceptible duplicate.  Do they
have to just match on type and name, or do they have be exact duplicates
down to the whitespace in the annotation elements?  They'd at least need
to match on any <appinfo> element since that might be used for
additional processing.

Having to detect when multiple declarations are exactly (or
sufficiently) identical seems to be unnecessary complexity.

Roger Costello wrote:

>what about this:
>
>    <element name="foo"> 
>       <complexType>
>          <element ref="a:bar"/>
>          <element ref="a:bar"/>
>       </complexType>
>    </element>
>
>Would you consider this to be legal or illegal?  It's not really
>defining an element twice.  Rather, it's referring to the single
>declaration twice.

Legal.  It is not making any element declarations in the local namespace
of the complex type, hence no duplicate names in the symbol space.


>On the other hand, Henry says that the 'ref' form is equivalent to the
>in-lined form,

They have some similarities, but as I mentioned before they aren't
precisely equivalent.  The "ref" example declares an <bar> element in
the global scope that can be used in other contexts, the inline form
defines a <bar> element that can only be used in the context of the
<foo> element.  I don't think Henry meant that they were
indistinguishable.