I could see where this becomes a question in those cases where schema is assigned to an XML document in a manner that's beyond the control of the transformation writer (or in those cases where type compatiblity is needed but the XML itself is potentially dirty with regard to type). Admittedly its an edge case that probably wouldn't appear if you know more about XSLT 2 and XPath 2 programming, but at least for a while I suspect that you'll see a lot of these cases in XSLT that's being upgraded from 1.0.