From: Erland Sommarskog (sommar-usefor@algonet.se)
Date: Sat Aug 10 2002 - 16:11:22 CDT
=?ISO-8859-1?Q?Claus_F=E4rber?= (list-ietf-wg-apps-usefor@faerber.muc.de) writes:
> Erland Sommarskog <sommar-usefor@algonet.se> schrieb/wrote:
> > Nah. There is one obvious problem, the Turkish i and I. There might
> > be a few more. But it should be possible to define sensible rules
> > for a general case-insensiivty for the Latin/Cyrillic/Greek scripts.
>
> "Turkish" is a language, not a script, and that's the whole problem:
> Most of the special casing rules apply to languages, not scripts.
>
> So you need language information to handle casing correctly. This can't
> be done with domain, mailbox and newsgroup names as the language tags
> would result in different names for the same character sequence (and
> there's no way to look up metadata that's not part of the name).
>
> The best solution is to ignore the problem and to accept that a small
> set of character pairs that some people view as upper and lower case
> counterparts won't be seen by the computer as such.
Yes, that was I had in mind. You would need to define some general rules,
and probably point out the places where it breaks for a specific language.
For the Turkish case, the best may be to define all four varities of "i"
as equivalent. The german ß is more devilish, but possibly the result
would be that a mail to king.haßan@court.mo would reach the Morrocan
king.
Obviously, there are likely to be some confusing cases, but what are
the alternatives:
1) Make all mail address case sensitive and break a lot of things.
2) Only make all non-ASCII characters case-sensitive.
3) Define general language-dependent case-insensitiveness as best-possible.
Certainly 2) would be a lot more confusing than 3). For an innocent user
it's not clear why erland.sommarskog@bolag.se works equally well as
Erland.Sommarskog@bolag.se, whereas to reach Östen Åkesson, he must
know precisly whether it is östen.åkesson@bolag.se or Östen.Åkesson@bolag.se.
For the local part, one could let the admin configure his server for
rules fitting the local language, but I'm not sure this is a good
idea.
-- Erland Sommarskog, Stockholm, sommar@algonet.se