Patrik F�ltstr�m wrote:
> I.e. I see this problem being orthogonal to the encoding of the > Unicode codepoints, and have not much to do with the DNS protocol, > which already can handle 0x00-0xFF. The root problem case is that the STD13 legacy label type is used to provide multiple encodings; this is an overloaded data-type problem more than anything else [OT: it is the same root problem case that required EDNS extended labels to be used for UTF-8]. The readily-visible problem case is that the canonical UCS character code will have two values when encoded, and the appropriate encoding cannot always be determined (see above). -- Eric A. Hall http://www.ehsco.com/ Internet Core Protocols http://www.oreilly.com/catalog/coreprot/
