This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
Tes cbcl-avg-010 is: avg((xs:double("NaN"), current-date() - xs:date("1997-01-01") )) I believe that under "Errors and Optimization", NaN is a legitimate result. Once an xs:double NaN value has been found in the input sequence, the only possible outcomes are NaN or an error, so the processor is entitled to take an early bath and return NaN.
Agreed.
Fixed as suggested.