Before we vote: I agree with these definitions below - but I think we do not need to add any additional on-error wrapping functions as part of the core standard. Error is an error - it should not happen, and if happened should not be hidden or silently substituted with a default value. If such functionality is truly needed - a second rule and an appropriate rule-combining algorithm can take care of that without blowing up the default list. Or it can be an extension. Daniel.