This issue was raised at the May 2003 f2f. See: 
(http://www.w3.org/XML/Group/2003/05/xml-schema-ftf-minutes#d0e750 

XML Schema Part 2 suggests that the XMLSchema-datatypes URI is an alias for the 
xmlschema namespace URI: 

"To facilitate usage in specifications other than the XML Schema definition 
language, such as those that do not want to know anything about aspects of the 
XML Schema definition language other than the datatypes, each built-in datatype 
is also defined in the namespace whose URI is: http://www.w3.org/2001/XMLSchema-
datatypes " 

while the documentation in the schema for the XMLSchema-datatypes suggests a 
different relationship, which is consistent with the notion of one-name-per-
component: 

<documentation>Note this schema is NOT a normative schema - - It contains types 
derived from all the builtin simple type definitions with the same local name 
but in a distinct namespace, for use by applications which do no wish to import 
the full XMLSchema schema. Since derivation is not symmetric, unexpected 
results may follow from mixing references to these definitions with references 
to the definitions in the XMLSchema namespace. For example, although 
dt:positiveInteger is derived from xs:integer, the converse does not 
hold.<documentation>

See bug 2214 for resolution of this issue in schema 1.1.