Re: rough sketch of a potential solution

At 09:55 03/11/17 +0000, Adam M. Costello wrote:

[taking your conclusions first]

>I don't know if this problem will be big enough to justify building a
>special new infrastructure to solve it, but I can see that there is
>indeed a problem.

That's the point I'm trying to make: There are of course cases
where this might be handy, but they are rather special, and
people are used to deal with these kinds of issues in a natural
way anyway, which reduces the need to get this handled completely
by machines.

That suggests that it is not really worth the effort of creating
an additional protocol and an infrastructure, with all it's issues.


>Martin Duerst <[email protected]> wrote:
>
> > The sender should not send an address that the receiver doesn't know
> > how to read.
>
>Here's one example that motivates Keith's problem statement (the one
>I called problem2):  Let's say you and I recently met, and we speak
>language X as a first language, and language Y as a second language.
>I know a third guy who speaks language Y, but not X.  I want to send
>a message to both of you (in language Y, obviously).  I know his
>Y-address, but not yours, because when we met you only told me your
>X-address.  So the other guy is going be stuck looking at your X-address
>or your ACE address, both of which are gibberish to him.
>
>If there existed an address mapping service, then I could type in your
>X-address, and my MUA could discover your Y-address, and put both in
>the header, so that the other guy's MUA could select the Y-address for
>display.

It looks like this issue would be solved almost equally well by
having a separate (directory-like) lookup structure, independent
of the mailing address issues. I.e., you would check whether
I had a Y-address. Then you would put that Y-address in the 'To:'
field, and send the address.

Of course, this could be somewhat automated, so that it would
look like a single step to you. The difference to Keith's
proposal would be that in every single mail, only a single
address (or potentially a fixed number, such as one i18n address
and one ASCII fallback,...) would be sent.


Regards,    Martin.