Re: [idn] nameprep failures

["Adam M. Costello" <amc@cs.berkeley.edu>]

"D. J. Bernstein" <djb@cr.yp.to> wrote:

> I see a domain name on a business card.  I type it in.  I'll be
> unhappy if I don't get the right name.  Isn't that the whole point of
> nameprep?

In my opinion (and this is just my opinion, not an interpretation of
the nameprep spec or the IDNA spec) the point of nameprep is to make
sure that users are able to type what they intend to type, and to make
sure that names are not corrupted by transcoders.  I think the visual
ambiguity problem is too much for nameprep to solve.  What do you do
about l and 1?  What about katakana ka and the Han character for power?
What about the katakana long vowel mark and the Han character for one
(and the em dash, and the minus sign...)?  I think Mark Davis is right
that this is a morass.

> If I see WWW.AOL.COM, where the first dot is actually U+3002, then
> I'm sure I'll type it incorrectly.  Isn't this why nameprep prohibits
> U+3002?

No, the nameprep spec is explicit about the reason for prohibiting
U+3002:

    U+3002 is used as if it were U+002E in many input mechanisms,
    particularly in Asia.  This prohibition allows input mechanisms to
    safely map U+3002 to U+002E before doing nameprep without worrying
    about preventing users from accessing legitimate host name parts.

> If I see WWW.AOL.COM, where the A is actually a capital Alpha, then
> I'm sure I'll type it incorrectly.  So why doesn't nameprep prohibit
> Alpha?

[Nitpick:  I wouldn't say you typed it incorrectly, I'd say you read it
incorrectly, then typed exactly what you intended to type.]

No one is going to put <Alpha>OL.COM on a business card and seriously
expect people to read it correctly.

AMC