If we had 'Character Set Override' at the attribute level,
parsing could become impossible (depending on the new
character set). For example, if you shifted from UTF-8
(required as default charset in all new Internet protocols,
per RFC 2277 "IETF Policy on Charsets and Languages")
to UTF-16, byte ordering problems would be insoluble.
Also, clients are permitted to send in text in languages
which are NOT supported (for error/status response
messages) by servers (IPP Printer objects), but the
charset MUST be mutually understood, for parsing.
Except for some markup languages, I don't know of any
protocol that lets you change the charset on an
individual attribute value basis.
- Ira McDonald (outside consultant at Xerox)
High North Inc