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 wording of the definition for [node table] at [1] can lead to a scenario [2], where keyrefs resolve perfectly, even if the relevant key-sequences of the referenced key/unique have duplicates in the subtree. Is this intended? [1] http://www.w3.org/TR/xmlschema-1/#d0e14060 [2] http://lists.w3.org/Archives/Public/xmlschema-dev/2005Sep/0060.html
The Schema WG considered this issue at its 2006-12-22 telecon, and decided to close this issue with no further action. The resolution is somewhere between "invalid" and "wontfix". The analysis from the email exchange is correct: "the success of keyref-resolution is dependant on a specific constellation of the keys in the tree". This can be viewed both as a feature and a bug. One can certainly argue that key values specified on the parent element *overrides* those specified on the children, which would support rules in the current spec. This was a deliberate decision in the schema spec; it will only affect very few cases (recursive element references with key/keyref); changing it may dis-stablize the spec. The WG decided not to change the spec based on these factors.