Comments on the Unicode Codepoints and IDNA Internet-Draft

[Kenneth Whistler]

> > This kind of restriction should be a matter of registry
> > policy, but not be something built into the protocol
> > and the table it uses.
> 
> Just in case, I would not know what "old Hangul" is when
> it bites me.  The registry in Korea and maybe others will
> know it, they can forbid it.

Correct.

>  Other registries could say
> "it is PVALID, here is your domain, have fun". 

Also correct. And even meaningful. And if we don't want
the protocol (and table) definition to get in the business
of prohibiting, say, Old Irish, as opposed to modern
Irish, or Old High German, as opposed to modern German,
or Classical Chinese, as opposed to modern Mandarin,
then we shouldn't start editing the table to try to
screen out historic characters from it. Even more so,
since the consensus of this group was not to omit
purely historic scripts, on the chance that *somebody*
might want to use them in IDNs.

So yes, I think the restriction (if any) should be
up to the registry in this case. If the Korean NIC
wants to restrict itself to the 11162 modern Hangul
syllables at U+AC00..U+D7A3, that's fine.
 
> 
> If you see potential harm in conjunction with a clueless
> registry please say so.

No, I don't. Certainly no more so that a clueless registry
that lets somebody register a Sumero-Akkadian Cuneiform
domain name or a Runic domain name.

>  Similar a Korean registry might
> not know how good, bad, or ugly Latf is; and a proposal
> to disallow Latf generally could make sense from my POV.

Except that Latf is *not* separately encoded in Unicode,
but is treated as a style of Latin letters. You might
recommend that people not display URLs in Fraktur,
but that is different from "disallowing Latf", since
there is no distinct Latf script in Unicode to
disallow -- it is all just Latn.

--Ken