Re: [idn] Thoughts on nameprep

["D. J. Bernstein" <djb@cr.yp.to>]

Adam M. Costello writes:
> prevent names that look the same but are actually different

We're going to have registrars prevent this. We're specifying a set of
good names. Freely available registration software will recognize good
names and reject bad names. The whole point is that two different good
names won't look the same. End of problem.

The big question is whether we need to do more than this. Do we need to
allow, in some contexts, bad names that look the same as good names? Do
we need to convert those bad names to the corresponding good names? Why
is this necessary? How will the bad names appear?

Half-width Japanese kana, for example, are bad. Registrars will reject
names that contain half-width kana. But do we need to convert half-width
kana to full-width kana?

Patrik and Martin have repeatedly told us that the answer is yes. Users
will see Japanese domain names in print and type half-width kana into
the computer, Patrik says. They'll expect to connect to the same site.

But Bruce, writing from Japan, says that the answer is no. Users type
full-width kana. Redeploying network software to convert half-width kana
to full-width kana, as Patrik is insisting that we do, would be idiotic.

> It's actually pretty simple; it just uses large tables.

Tables that aren't in current software. Tables that may change. We're
talking about software upgrades on a massive scale, and perhaps further
upgrades on the same scale when Unicode is modified. I'm astounded that
some people are taking such a careless approach to such a large project.

> We can't forbid the combining acute accent, because some languages need it,

Registrars will forbid strings that have an e followed by a combining
acute accent. They'll allow the precomposed e-with-acute-accent.

Now, are you claiming that the usual keyboard interface in some locale
will produce an e followed by a combining acute accent? Please provide
verifiable details so that we can avoid making a Patrik-style mistake.

If there is a problem, why not fix it in the keyboard interface for that
locale? This is fast nameprep: bad names are converted to good names
before they escape the keyboard interface.

---Dan