[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [idn] stringprep comment 1
> At 10:51 PM -0800 1/31/02, Yves Arrouye wrote:
> >As for case folding, as soon as one adds a new case mapping in the
> >profile, one needs to upgrade clients at the same time as servers. Oops.
> Could you give an example of this? Why would we ever want to add a
> new casefolding rule for a current character? Adding a casefolding
> rule for a *new* character should not make a difference because the
> new character is not yet stored in domain names.
Sure. I'll use the same example I just used and used before.
Server S runs Nameprep-08. It *does* have names that use this new character,
and it has been case folded since Nameprep-08 has a case mapping for it.
Client A has an email client using Nameprep-07. At the same time, this email
client runs on the latest and greatest <your favorite OS here> which has
support for the version of Unicode that is used in Nameprep-08 (and has this
new character in both upper and lower case--I am speaking of the Unicode
version here). Now, this character (the uppercase one) is typed by the user
(since the OS supports it). The email client sees it as being unassigned as
far as its Nameprep-07 tables go, and thus pass it unchanged to the server.
On the server, there is no match, because the registered string was case
folded (since the server uses Nameprep-08).
But since you said that DNS makes no guarantee of case mapping (and I guess
it's fair to say that if IDNA does not make any guarantee, it doesn't matter
what DNS for ASCII did) I guess it's okay. It's just going to be fun when
the user registers his IDN with uppercase but cannot use it from his own
email client :) It's all about user education... (And fortunately, most
people type these things all-lowercase anyway.)