[Bug 4353] [XSLT+XQuery] Identity constraints at element level

http://www.w3.org/Bugs/Public/show_bug.cgi?id=4353

           Summary: [XSLT+XQuery] Identity constraints at element level
           Product: XPath / XQuery / XSLT
           Version: Recommendation
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: XPath
        AssignedTo: chamberl@almaden.ibm.com
        ReportedBy: mike@saxonica.com
         QAContact: public-qt-comments@w3.org


Both XSLT and XQuery state that when validating an element node (as distinct
from a document node), identity constraints specified in the element
declaration are not taken into account.

XQuery says: If the operand node is an element node, the validation rules
named "Validation Root Valid (ID/IDREF)" and "Identity-constraint Satisfied"
are not applied. This means that document-level constraints relating to
uniqueness and referential integrity are not enforced.

XSLT says: Validation of an element or attribute node only takes into
account constraints on the content of the element or attribute. Validation
rules affecting the document as a whole are not applied. Specifically, this
means... The validation rule "Validation Rule: Identity-constraint
Satisfied" is not applied.

In both cases, the reasoning given in the spec is spurious. Identity
constraints (xs:unique, xs:key, and xs:keyref) defined on an element
declaration are not document-level constraints, their effect is local to the
element on which they are defined. It would be perfectly possible to apply
such constraints when validating at element level, and it seems odd that we
should diverge from the XML Schema rules in this regard. There might be some
rationale if we ignored everything on the element declaration other than its
type, but we don't: we take into account properties of the element
declaration such as abstract="true", nillability, and fixed values.

The effect of the omission is that it's very easy to output a document which
appears to have been validated, but which will then fail validation if put
through a free-standing schema validator. This is also true of ID/IDREF
constraints, of course, but in this case there is a solid justification, in
that validation of a non-document element would otherwise give spurious
IDREF errors.

In the XSLT case, it's not entirely clear when this rule applies. Does it
apply, for example where validation is initiated using <lre xsl:type="xxx">?

Michael Kay

Received on Tuesday, 27 February 2007 16:23:00 UTC