Re: [EAI] Unicode / ASCII fallback mappings?

[Yangwoo Ko <[e-mail]>]

On Sat, Jun 27, 2009 at 8:31 AM, Shawn Steele<[e-mail]> wrote:
> I realize that a UTF-8 email address and the ASCII fallback for that address
> are left to the server to figure out, but I was wondering if anyone had any
> preconceptions?
>
> I can think of 3 basic ways to solve generating ASCII addresses: 1) Some
> sort of encoding mechanism that is algorithmic and likely makes an
> unreadable ASCII address, 2) aliasing, where aliasing happens automagically,
> like perhaps Cyrillic to ASCII transliteration, and 3) aliasing where the
> ASCII alias doesn't necessarily correlate with the native-script form.

First and second options are unacceptable in many ways including
technically and culturally.

> My preconception is that the last is the most useful, unless the native
> script has well-defined transiliteration to Latin rules.
>
> Furthermore I'd expect that when I created a new account on something like
> live mail/gmail/yahoo mail, that it'd be nice to get prompted for both the
> Unicode and ASCII aliases, possibly with a generate ASCII alias button for
> users that aren't as familier with that form.

Well it seems to depend on cases. For instance, if such a mail service
provider only provided (or were only known for its) ASCII domain name,
it would be far less useful to support registration of unicode local part
resulting in [e-mail].

>
> Lastly, I'd assume that "enabling" EAI on such a system, such as with a
> Chinese ISP, would mean that when I log in with an existing ASCII address, I
> could be prompted with a "Get your chinese alias now!" button, so that I
> wouldn't have to make a new account or lose service/migrate messages,
> etc. from my old account.

As a marketing term, "alias" can be used. But, technically, we should not
call them as alias because we are not providing a technical binding between
them except delivering the fact that user of those mail addresses bound them
together.

>
> It occurred to me that what I think is obvious may not be what
> everyone/anyone else is thinking.  Does anyone care to share any other
> options or whether they agree?  I know that earlier Mark had suggested a
> punycode form, but I'd assumed that most end-users would prefer a more
> friendly alias.  One that would make sense on a double-sided business card
> for example.

I agree with the double sided business card analogy in a sense that it requires
human & manual intervention to bind/match two different names and that we
usually don't need to look at both sides.

>
> -Shawn
>
> _______________________________________________
> IMA mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/ima
>
>



-- 
a human / twitter as yangwooko / msn as [e-mail] / nate
as [e-mail]