On Jul 4, 2017, at 9:53 AM, Randy Bush <randy@xxxxxxx> wrote: > i would offer to put my keyboard where my mouth is. but i fear that, at > the bottom, i would have the unreasonable desire for dns classes to > support these kinds of things. i.e. i don't think we have a clean fix. > but it would be nice to document the good with the bad. That sounds like a solution, not a motivation. That is, you care about the problem hypothetically, and have a hypothetical solution. In practice when we’ve talked about using dns classes to solve problems that have motivated rfc6761 allocations, it hasn’t really helped, because the infrastructure required to use them this way is not present, and this isn’t how they were originally intended to be used. For example, is ICANN.org with a different class not a subdomain of the .org TLD? Would ICANN not object to us designating it for use by someone else? I suspect yes, and I wouldn’t blame them.