Saxon

Legacy ID: #3489832 Legacy Poster: Michael Kay (mhkay)

This is legitimate according to the rules in XPath section 2.3.4: as soon as a processor can determine that the result of an expression X is either a value V or an error, it may return the value V without doing further work to look for an error. Section 3.1.5 Function Calls reinforces this: "The order of argument evaluation is implementation-dependent and a function need not evaluate an argument if the function can evaluate its body without evaluating that argument." If the first argument to normalize-unicode is an all-ASCII string, then the result doesn't depend on the second argument, so it's not required that the second argument should be evaluated. Michael Kay

Legacy ID: #3490469 Legacy Poster: marcvc (marcvc)

Thanks for the clarification. And of course, fully agree. Thanks, Marc