[UA-discuss] Review of UASG Charter

[Mark Svancarek]

This is the crux of the problem.
		Since every A-label has exactly one U-label and conversely, why isn't the application catching the equivalence

The app doesn't catch the equivalence simply because it never had to do so in the past, and because the mail store/db, the transport, and the search/sort function are likely in different teams working on code bases of different provenance and different update cadence.  Just plain old software development snafus of the sort routinely encountered in UA.


-----Original Message-----
From: Andrew Sullivan [mailto:ajs at anvilwalrusden.com] 
Sent: Thursday, February 18, 2016 12:16 PM
To: Mark Svancarek
Cc: ua-discuss at icann.org
Subject: Re: [UA-discuss] Review of UASG Charter

On Thu, Feb 18, 2016 at 08:00:51PM +0000, Mark Svancarek wrote:
> I refer to ASCII at xn--something

Got it.

> 1.       I send an email to a pair of recipients:  A-label at U-label and  ASCII at ASCII.
> 
> a.       (The former is EAI and the latter is non-EAI).

Sorry to be a pain, but one of the things that has hurt us in i18n discussions is terminology.  So, I want to say that differently.
Let's lay this out carefully.

Your MUA supports EAI.  Your MTA supports EAI.

You compose mail:

    To: ascii-local-part-1 at IDN-as-U-labels, ascii-local-part-2 at LDH-only-labels
 
> 4.       The iPhone email app does not reply to A-label at U-label – it replies to A-label at xn--something.  That just happens to be how that app is currently designed.
> 

This, of course, is because the app doesn't support i18n in the headers at all, right?

> a.       Now my mailbox contains 2 replies.  The reply from the EAI user is to the same addresses that I originally used.  The reply from the iPhone user contains a different address than I used.
>

Is that true?  Your system, under hypothesis, supports EAI.  Didn't your system convert the A-label in the server-part to a U-label?  Why not?  (There may be a gap in the EAI protocol here; I'm not sure.)

> 
> a.       AFAIK there is no requirement that these addresses be treated as equivalent.  Email is not DNS.
> 

But the server-part in an EAI address is an IDNA-conformant IDN, or else a non-IDN.  Since every A-label has exactly one U-label and conversely, why isn't the application catching the equivalence?  It already does IDNA.

A

--
Andrew Sullivan
ajs at anvilwalrusden.com