Keith Moore wrote: > it's a false dichotomy. > > first, there are probably some purposes for which identifiers > (including DNS names) should stay ASCII (and not even 10646 > encoded as ASCII) for the forseeable purpose. Agreed. Well-known data-types such as Message-ID are actually harmed if they exist in multiple forms (this also includes the transliterated form that results from IDNA). This is why implementation is a per-WG effort, not a global mandate. However, the infrastructure (in the broadest sense) should support native IDNs in order to facilitate the most effective scoping for any WG. > a UTF-8 everywhere approache forces him to wait for everybody > else to adopt IDNs before he can use them. lol, speaking of false dichotomies -- Eric A. Hall http://www.ehsco.com/ Internet Core Protocols http://www.oreilly.com/catalog/coreprot/