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 specification of fn:idref() (in 15.5.3) says: A node $N has an IDREF value equal to V if both of the following conditions are true: * The is-idrefs property (See Section 5.6 is-idrefs AccessorDM.)of $N is true * The sequence fn:tokenize(fn:normalize-space($N), ' ') contains a string that is equal to V ... Now if $N is a node that has acquired the is-idrefs property by virtue of having the schema type xs:IDREFS, then the call to normalize-space() will implicitly atomize $N to produce a sequence of atomic values of type xs:IDREF; and if there is more than one such atomic value, the call will fail with a type error, because its signature expects an argument of type xs:string? (that is, it does not accept a sequence of more than one string). I think that what is intended is fn:tokenize(fn:normalize-space(fn:string($N)), ' ')
The proposed change was accepted by the WG on 16 Dec 2008. Erratum E29 has been drafted to the 1.0/2.0 spec, and the change has been applied to the editor's master of the 1.1/2.1 spec. Note that the erratum supersedes E3.