Re: Fw: [idn] IDNs in email message bodies

["Eric A. Hall" <ehall@ehsco.com>]

> > The point is still the same, ACE-encoding/decoding can occur at the
> > same  time as 2047-encoding/decoding, when the display data is
> > converted into protocol data.
> 
> there are two problems with this statement.
> 
> the first is that address processing is done at many different places,
> not all of which are concerned with presentation.  an SMTP server might
> need to do ACE decoding on addresses independently of any user agent.

Why is that a problem?

> the second is that in many current implementations rfc 2047 is done
> at the wrong time - a message is RFC 2047 headers is often "converted"
> to use raw characters at some point prior to presentation, and those
> raw characters often leak out in replies, forwarded messages, etc.

I expect that ACE leakage will be a problem forever, just as 2047 leakage
is going to be a problem forever. Not a major problem though if the
addresses are verified when the envelope is created. If an implementation
isn't following the old rules there's not much we can do about it.

> I'd agree that if a user agent were implemented properly, it would
> decode rfc 2047 and ace domain names *for the purpose of presentation*
> at the same time - but this might not be the same time that rfc 2047
> is decoded by a current implementation of that user agent.

Agreed. For the purpose of message composition, address books and the
like, the IDN representaionn should be the visible form (just like 2047)
while the underlying mechs and structures will use ACE (just like 2047).

-- 
Eric A. Hall                                        http://www.ehsco.com/
Internet Core Protocols          http://www.oreilly.com/catalog/coreprot/