This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The definition of conformance in the Datatypes spec seems to need reworking. The definition of 'minimally conforming' processors seems to suggest that there is only one constraint on schemas and only one validation rule. (In reality, Datatypes defines no constraints on schemas, but a number of validation rules.) The definition of 'conformance to the XML representation of schemas' is out of synch with the corresponding wording in Structures, which has been revised to make the terminology less clunky. At a minimum we should align the text. But we should probably also ask ourselves whether the two classes of 'minimal conformance' and 'schema-document-aware conformance' are useful, necessary, and sufficient. Is it meaningful to imagine a schema-document-aware datatypes processor that is not also XSD-aware (i.e. that does not also support Structures)? And is there no need for any constraints on host languages for Datatypes, no occasion for specifying what they must do to use Datatypes meaningfully? It will be difficult to specify clearly how support for additional primitives (bug 3251) affects conformance, with the conformance story in its current shape. N.B. some aspects of this question apply to 1.0 as well as to 1.1, but I am raising this bug only against 1.1; I don't believe a re-thinking of the conformance story is suitable for an erratum.
On 25 April 2008 the working group accepted a proposal to resolve this issue by making the wording changes outlined at http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b5585.html (member-only link)
As the originator of the comment, I record my belief that the WG decision recorded in the previous comment satisfactorily resolves the issue I raised.