Re: [idn] IDNs in email message bodies

[Keith Moore]

> as far as I'm concerned, RFC 2047 encoding is a non-starter for domain
> > names in protocol elements.  2047 was specifically designed to encode 
> > text whose only purpose was to be presented to a user; and it therefore 
> > completely skirts all issues related to normalization, canonicalization, 
> > etc.
> 
> While I agree to a certain degree (within the protocol and
> transportation/routing), I would like to say that in the context of this
> discussion where we are disucssing the "message headers" and "content body"
> of an email, the information is in fact for display purpose.

no.  not only is this not "in fact", it's simply incorrect.
the addresses in the message header are intended both for display
and for use in forming replies.

also, in practice, RFC 2047 encoded-words are routinely decoded, 
translated, re-encoded, and/or corrupted by mail transports and
delivery systems.  

Keith