In <200204251916(_dot_)g3PJGJe10745(_at_)astro(_dot_)cs(_dot_)utk(_dot_)edu>
Keith Moore <moore(_at_)cs(_dot_)utk(_dot_)edu> writes:
a. encode domain names in IDNA
b. encode local-parts in an encoding that is to be defined, but which
hopefully is very similar, perhaps identical, to IDNA
The problem here is that nameprep is stricter than mailbox naming has
historically allowed for. Specifically, the use of mandatory lowercase
characters in nameprep will prohibit mixed-case local-parts.
Good point. OTOH, if we defined a special version of nameprep for
local-parts to allow mixed-case, then local-parts of internationalized
email addresses would be subject to case-sensitive comparison.
Of course, strictly speaking, interpretation of the local-part is up to
the domain, but in practice, case-insensitivity is often taken for
granted by both users and software.
I have not studied the Punycode specification in detail, but I have
experimented using the code provided in the appendix.
If I encode CHLchl, itc comes out as chlchl- (which then decoded to
chlchl).
However, if I construct CHLchl- and then decode _that_, it comes out as
CHLchl. So Punycode appears not to be reversible, which is a Bad Thing.
But there is a further problem. Presumably, two '-'s in succession are
illegal in a domain name, so if the DNS sees zz--tu-mnchen-t9a, it knows
that decoding is needed. But zz--jrgen-kva is already a valid local-part as
it stands. Perhaps the mail really was intended for that entity.
So I think some limitation of the RFC 2822 syntax is going to be required
whatever.
--
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133 Web: http://www.cs.man.ac.uk/~chl
Email: chl(_at_)clw(_dot_)cs(_dot_)man(_dot_)ac(_dot_)uk Snail: 5
Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9 Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5