Re: [EAI] Proposal for Alternate ABNF (was: Consensus Issue #6: current metalanguage model acceptable?)

[Shawn Steele <[e-mail]>]

Yes, it's legal for the existing RFCs, however the existing RFCs don't allow the use of MAIL FROM UTF8SMTP.  Presumably it could be a requirement that if you're using the EAI RFC, then punycode could be illegal.

At the very least u-labels should be a SHOULD.

-Shawn

 
http://blogs.msdn.com/shawnste


________________________________________
From: John C Klensin [[e-mail]]
Sent: Sunday, January 23, 2011 2:58 AM
To: Shawn Steele; [e-mail]
Subject: Re: [EAI] Proposal for Alternate ABNF (was: Consensus Issue #6: current metalanguage model acceptable?)

--On Saturday, January 22, 2011 22:54 +0000 Shawn Steele
<[e-mail]> wrote:

>> I can live with this approach.  If it is adopted, I'll help
>> sort out the domain name part.
>
> As mentioned, IMO, the domain part should "just" say UTF-8 as
> well, though it's fair to point out the dependencies on
> existing RFCs that define DNS.  DNS is a bit trickier than the
> other headers because we explicitly want to disallow the
> punycode I think, and require UTF-8 U-Labels instead.

Shawn,

I know that you want to "disallow" punycode.  In a perfect
world, I would too, and that is part of what
draft-iab-idn-encoding is about.  But, realistically, while we
can try to discourage its use, banning it is just impractical
given that A-labels conform to the RFC 1035 "preferred syntax",
the syntax requirements of RFCs 5321 and 5322, and even the
rather careful rules in RFC 5892 about what does and does not
need to be checked on lookup.

    john