That's a proper subset of UTF-8, so not really necessary. Can a particular XML file be represented in, say, 8859-6 or JIS-X, and still be standard? I don't like this because it means that a file can't be read unless the parser knows that character set.