RE: draft-hoffman-utf16-01.txt available

Appendix F of the XML specification

(http://www.w3.org/TR/1998/REC-xml-19980210#sec-guessing)

makes it clear that BOM isn't actually necessary for auto-guessing
of charset between UTF-16LE and UTF-16BE, because of the differing
encodings of "<" and "?" in the initial "<?".

I might suggest that the XML recommendation be amended to disallow
charsets that cannot be recognized using the mechanisms included
in Appendix F; however, I'm not sure there are any already registered.

Larry