The only problem would come in would be if you were trying to read a CharML file that *itself* was encoded using a character set that your XML parser didn't know. That's one reason for encoding the CharML files themselves always in UTF-8 or ASCII. I'll post this to a broader mailing list, since some others may have similar concerns. Mark ___ [EMAIL PROTECTED] IBM, MS 50-2/B11, 5600 Cottle Rd, SJ CA 95193 (408) 256-3148 fax: (408) 256-0799 "Claude Tardif" <[EMAIL PROTECTED]> To: Mark Davis/Cupertino/[EMAIL PROTECTED] cc: <[EMAIL PROTECTED]> 2003.03.19 21:44 Subject: Unicode Technical Report #22 Your document referenced in the title of this message specifies an XML format for the interchange of mapping data for character encodings. Inversely, the Extensible Markup Language (XML) 1.0 (Second Edition) section 4.3.3 specifies an entity for changing the character encoding of XML formatted documents. If character encoding uses XML and XML uses character encoding, there is necessarily an interdependency loop. For example, what if a conversion library such as ICU parsed character encoding files using an XML parser which itself used ICU to convert character encoding in entities? Then, if the XML file defining the charset encoding for ISO-8859-1 contained the entity <?xml encoding='ISO-8859-1'?>, this would cause a loop as the character encoding could never parse itself. My question is: Is there a way for a conversion library and XML parser to make use of their services mutually without causing such an interdependency loop and, preferably, without having such requirements as character encoding files not containing character encoding in entities? Marc Tardif