Re: [xml-dev] Blog post on limitations of import/include in W3C XML Sche

[noah_mendelsohn@us.ibm.com]

Anthony B. Coates writes:

> Hi Noah.  You seem to be talking about Schema *file* 
> repositories, I was certainly talking about Schema *definition*
> repositories, which are not quite the same (just as the old IBM
> "Visual Age for Java" product was different in the way it did 
> method-level versioning instead of file-level versioning).

I don't think I missed that distinction.  For XSD, we have schema document 
files for those who want a standard means of setting out schema 
definitions in a file, and we have schema components, which are the 
abstractions for element declarations, type definitions etc.  I think the 
analogies to Java are pretty good:

Java Source file <==>  Schema Document
Abstract class or method definition <===> schema component

In fact, the schema Recommendation provides great flexibility as to 
whether Schema Documents are used.  It allows for others to create 
specifications for repositories of various sorts, and those repositories 
can work at the level of schema documents, schema components, or some 
mixture.   Now, what is true, is that components tend to live in a larger 
context.  If I declare in a schema document:

        <simpleType name="t">
         <xsd:restriction base="ns:b">
                  <xsd:length value="100"/>
         </xsd:restriction>
      </xsd:simpleType>

that's not enough to uniquely determine a component.  We also need to know 
the component that ns:b resolves to.  So, while you can also make 
repositories that manage schema document fragments like the one above, 
they are not what we mean by a "component".  Once you know that bs:b is a 
subtype of, say, string, then you'll know that this is a legal derivation, 
you'll know the base type chain, etc.  This is quite analagous to defining 
a derived Java class;  you don't know until you find the base class what 
the real signature of the derived class is, or whether in fact the 
derivation is legal.

Noah

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------