On Wed, Mar 14, 2012 at 10:45:58AM -0400, Ram Mohan wrote:

Kim,

I am not certain registries would want to use an automated/machine-readable mechanism for importing tables from other registry IDN implementations.

I strongly agree, and would go farther: I think it would be a bad idea for registries to blindly import tables from other registries. But registries are not the only people who may be interested in the variant relationships in a registry. For instance, the anti-abuse community might reasonably want to use those tables as the basis for constructing policies for what domains ought to be expected to be delegated. The anti-abuse community might want to use those tables to detect opportunities for phishing, and to try to produce mitigation strategies. The corollary to using registry policy as the source of sane IDN deployment is that clients need to be able to know what those policies are, and to test actual network behaviour against the policy. In the absence of such tests, the registry policies aren't enough. The only way that applications are going to be able to integrate the expected policies in their applications is if the policies can be expressed in a machine-readable way. If they can't be, then I predict the policies will be "boutique" ones: some applications will work, but others won't. (This is written from the ICANN Universal Acceptance of TLDs session, where I am keenly hoping this application perspective will be expressed.) Best, A -- Andrew Sullivan ajs@anvilwalrusden.com