[Cryptography] canonicalizing unicode strings.

[John Levine]

In article <51df1b45-fca4-7b29-0231-126f0060257c at echeque.com>,
 <jamesd at echeque.com> wrote:
>Attempts to restrict people to only using one script in an identifier 
>are not going to fly, but if someone uses more than one script, we need 
>to check against all potentially conflicting identifiers for homoglyphs.

It doesn't sound like you have much experience with practical i18n
usage.  While I agree that it would not be a great idea to limit all
your passwords to, say only Cyrillic or only Han, saying that any
particular password has to be in a single script or single language is
fine.  Keyboards are configured for a language and switching languages
in the middle of a word is hard since you have to tell your keyboard
to switch its configuration.

>And no official list of homoglyphs, or official software to canonicalize 
>them.

Given that people have been working on i18n for many decades, you
might consider how likely it is that you are the first person in the
history of the Internet to run into this issue, and whether there
might be some reason that a giant list of canonicalized homoglyphs
would not be the best way to solve it.

R's,
John