Re: [ietf-smtp] Are A-label and U-label addresses supposed to be equivalent ?

[John C Klensin <[e-mail]>]

--On Sunday, July 12, 2020 00:13 -0400 John R Levine
<[e-mail]> wrote:

> On Sat, 11 Jul 2020, John C Klensin wrote:
>> [ Are corresponding A-label and U-label addresses different
>> mailboxes? ]
> 
>> As distinct mailboxes, I think it would be within its rights
>> to do so.  I also think that, like the "user" and "UseR"
>> local part distinction, doing so would be fairly dumb and, if
>> I were implementing the MTA that <whatever> was going to use,
>> I don't think I'd go out of my way to make separating the two
>> easy.
> 
> That's what I was thinking.  But I was also thinking it might
> be different from upper/lower case mailboxes since, there is a
> defined mapping betwen A-labels and U-labels, and we already
> have at least one place, DNS lookups, where the MTA has to
> turn one into the other.

But, while, again, I'd think that taking advantage of the
distinction would be dumb, an MTA also has to do DNS lookups to
get the MX records but substituting the MX-target domain into
either the RCPT command or header "To:" would not work and would
be catastrophically stupid as well as clearly non-conforming.
So, as far as the MTA is concerned, that is a one-way
transformation with no need that I can thinks of to ever map tan
A-label to a U-label except _maybe_ in error reporting.

> I'm doing a project for the UASG testing the EAI conformance
> of a bunch of mail software.  It appears that Postfix treats
> A-label and U-label addresses differently, and Coremail may
> only accept the U-label and reject the A-label, although it's
> hard to tell because at the moment, its mail server for the
> 互联网.中国 domain doesn't accept anything.

The last I looked at Coremail, my impression was that it had
that MUA, MSA, and MTA functions a little confused.  Whether the
above would be plausible might depend on sorting that out.  In
addition, if the local-part is all-ASCII, a conversion to
A-labels is allowed by IDNA, especially in the MSA (just another
application) and SMTPUTF8 need not be involved.  I'm not going
to go back and look, but I think the SMTPUTF8 specs are fairly
clear that A-labels anywhere prior to the MTA's interface to the
DNS are strongly discouraged but, because of how IDNA works, not
prohibited.

> MUA support leaves a lot to be desired, e.g. Roundcube webmail
> always turns the domain into A-labels, even if the mailbox is
> non-ASCII.

That is clearly non-conforming to the intent of the specs;
whether it is clearly non-conforming is another question.  But I
suggest that USAG does no one any favors by hair-splitting about
the difference between intent (which was intended to promote
both operability and a decent user experience) and conformance
(which is closer to the former only).

> PS:
> 
> Coremail isn't accepting any mail at all for the
> 互联网.中国 domain. Telnet to port 25 and it says:
> 
>   554 IP<10.12.1.170> is rejected: 0
> 
> Huh? What?

If it sees your Telnet client as 10.12.1.170 and absent any
other information, I'd be inclined to wonder whether that is a
"we aren't accepting connections or mail from RFC 1918
addresses" rather than anything having to do with the address.

best,
   john