[JDEV] request for ideas: RFC822 to JID mapping

[Matthias Wimmer]

Hi Hiroaki!

Hiroaki Nakamura wrote:

>RFC822 is obsoleted by RFC2822.
>http://www.ietf.org/rfc/rfc2822.txt?number=2822
>
Yeah ... but this hasn't changed what addresses are valid. RFC822 for me 
is just one type of a mail transfer concept that is based on many other 
RFCs too (e.g. all the MIME RFCs).

>In Japan, as far as I can see these days, we don't use non-latin letters
>in mail addresses. First, we never use non-latin characters in addr-spec.
>Second, A display-name is either an ascii only phrase or an phrase
>with some or all words encoded by RFC1522 using iso-2022-jp.
>http://www.ietf.org/rfc/rfc1522.txt?number=1522
>
Yes ... but do you like to use latin characters in your mail address? I 
don't think so and because Jabber allowes users to register with nearly 
every unicode string as their username they do it (I know of even two 
users at my (German) server that registered with asian characters).
And even (or just because) you don't use non-ASCII characters in RFC822 
mail there is the problem what to do with mails these users write. What 
will be the sender of these mails after they have been gated to the 
RFC822 world?

>
>For example, my mail address is one of the following:
>hnakamur at v003.vaio.ne.jp
>"Hiroaki Nakamura" <hnakamur at v003.vaio.ne.jp>
>=?iso-2022-jp?B?XCIbJEJDZkI8GyhCIBskQjkwNTEbKEJcIg==?= <hnakamur at v003.vaio.ne.jp>
>
>Actually I have never seen only some words are encoded like an example
>in RFC1522:
>CC: =?ISO-8859-1?Q?Andr=E9_?= Pirard <PIRARD at vm1.ulg.ac.be>
>When we use RFC1522 encoding, usually all words are encoded as a
>whole.
>
Yeah, that's because you use BASE64 encoding (that's the "?B?" in it) 
and "we" use quoted printable ("?Q?"). But you are only allowed to use 
MIME in the real name part ... not in the mail address.

I think yesterday evening when I went to bed I had an idea that solves 
the problem partially:

There are protocols in development that support Unicode for domain names 
(see http://www.i-d-n.net/) it's even thought about using this for top 
level domains (see 
http://www.icann.org/committees/idn/registry-selection-paper-13jun02.htm) 
and I hope that as soon as internationalized domain names are used that 
there will be demand on internationalized user names too.
If all that has been introduced to RFC822 mail we will have a straight 
forward mapping. But what can we do until then?

Maybe (because I beleave that internationalized domains will come before 
internationalized user parts) until then we can map non-ASCII-JIDs to 
domains. There are already drafts for internationalized domains that 
could be used for this. Sure as long as a mail user agent isn't able to 
display internationalized domains this isn't better then displaying MIME 
encodings in clear, but I think that user agents will implement IDNs 
very soon after the final standard exist.

One example, think of the user 崰渰뤰퐰ﰰ줰朰@amessage.de he could be 
translated to the RFC822 address jabber@崰渰뤰퐰ﰰ줰朰.amessage.de and 
with the current DUDE coding this would be 
jabber at dq--vsvpvd7hypuivf4q.amessage.de on the wire.

(Note: to see the above example your mail reader has to support UTF8.)


Any comments about this?


Tot kijk
    Matthias

-- 
Fon: +49-700 77007770		http://matthias-wimmer.de/
Fax: +49-89 312 88654		jabber://mawis@charente.de