i think we can't use something in a standard that doesn't have a canonical format. if there is a F(name) that produces a minimal I18N result which is the same as any other F(name) that would look the same or mean the same if written, then we should outlaw on-the-wire strings that are not in that canonical format. if there is no such F(name) then we should not use I18N in DNS, punycode or no.
masataka ohta pointed this out 15 years ago and he was entirely correct. vixie _______________________________________________ DNSOP mailing list [email protected] https://www.ietf.org/mailman/listinfo/dnsop
