Re: [dd] DBOUND scope

[Ben Schwartz <[e-mail]>]

> Where would this appear in http://www.example.com?  Or widget.sales.example.com? The way you use the PSL is to walk up the
> tree until you find the boundary. Or are you saying your cache will do
> this tree walk for you and put the results in the EDNS?  Yuck.

In the simplest implementation, the cache would simply include all the DELEG records appearing within the QNAME.  A DELEG-aware resolver needs to acquire these records anyway during resolution, so this is not asking the resolver to perform any additional resolution work.

>>It's true that further delegations down the tree could set the "registry" flag disingenuously.  That's not a problem: the client is only interested
>>in the "topmost registrable domain", and would ignore any indications from further down the tree.

> If you look at the PSL, that doesn't work. There are real cases like
> uk.com where there is one registry underneath another. It is not a bug
> that the entries in the PSL are manually reviewed to see that they're
> plausible.

I'm not saying we can get rid of manual curation processes.  I do think we can construct a few mechanical processes that cover the most common PSL use cases most of the time, reducing manual curation work to exceptions and edge cases.  There may even be some clients whose needs are met by purely mechanical metadata checks, with little or no manual intervention.

> This is hard.  Let's not get DELEG stuck in this swamp of a side show.

I agree!  We should ship DELEG before attempting this.  But also, we should try to structure DELEG so that it can be extended to support this later.  Right now, I think we're on track for that anyway.