>to make sure that the result is a singleton. That is inforced by the rule >logic, and not by ERROR at the time of evaluation. I would have to disagree that this is important enough to intriduce such additional complexity. We agreed that we have to handle errors in evaluation anyway (not just invalid sequence length, but invalid syntax, division by zero, or whatever else an extension function may require to be communicated as "unable to compute" state) While we can enforce such limited type safety for standard functions - in the way you described - it may not be possible to enforce for extension functions. As we have this mechanism in place - I do not think there is anything wrong with using this for checking for correct sequence length - it will not affect speed or security in any way. XACML function can be interpreted INTERNALLY, within implementation, to have such length enforcement logic - PAP can deal with that if needed. Daniel..