The Future of IDNA
Erik van der Poel
erikv at google.com
Thu Mar 19 20:53:52 CET 2009
On Thu, Mar 19, 2009 at 12:08 PM, Andrew Sullivan <ajs at shinkuro.com> wrote:
> On Thu, Mar 19, 2009 at 11:12:39AM -0700, Erik van der Poel wrote:
>> HTML processing. The point is that IF the client does not map Final
>> Sigma to Normal Small Sigma and Characters with Tonos to Characters
>> without Tonos, THEN the server must bundle all of the permutations
>> (final/normal and with/without tonos).
> I'm not sure I accept your premise. I think that the IDNAbis work
> started with the premise that what you describe above is a policy
> matter for the zone operator. If you want to argue it's _not_ (and
> we've had some such arguments, most strongly from Mark Davis, I think,
> on security grounds), that's ok.
I'm not arguing this on security grounds. I'm arguing that the .gr
folks have made reasonable requests to push the policy from the zone
operator to the client, by IDNA protocol. And, since modern Greek is
used in such a way that the .gr registry felt the need to bundle on
the server side when the IDNA2003 protocol did not require mapping
tonos on the client side, we should pay attention to their requests
and try to fulfill them if at all possible. There may be requests from
mathematicians or whoever to distinguish Final from Normal Sigma, but
I don't accept that such requests are more important than those coming
from the .gr folks. (Mathematicians or whoever are a "small"
community, in my view.)
> Disagreement about this premise is,
> in my opinion, the crux of the controversy over whether IDNA2008 or
> IDNAv2 is the right path to try to follow.
I think neither IDNA2008 nor IDNAv2 is the right path to follow.
IDNA2008 does not distinguish between the two character sets
(lookup/registration vs display) and IDNAv2 does not support Eszett,
ZWJ nor ZWNJ.
> But I don't think you can
> assume that everyone will come up with the same answer to the question
> of what one does in the case the client does not map these characters.
The .gr folks have already given us their answer, and that is that
server side bundling is very painful, and that they would prefer to
have the client map Characters with Tonos to Characters without Tonos.
And when they said that, we did not hear from any community, large or
small, that they wanted to distinguish Characters with Tonos from
Characters without Tonos.
More information about the Idna-update