Re: [weirds] RDAP registrar entities

[Brian Mountford <[e-mail]>]

I see. So an array of child entities is legal inside an entity or
nameserver, not just a domain, correct?

On Fri, Oct 30, 2015 at 1:09 PM, Andy Newton <[e-mail]> wrote:

> Hi Brian,
>
>
> On Oct 30, 2015, at 12:01 PM, Brian Mountford <[e-mail]>
> wrote:
>
> First of all, is there a reason why registrars and contacts were lumped
> together into one type of thing? Just curious.
>
>
> The reason is quite simple. All sorts of organizations end up being
> modeled the same way, with only a relationship qualifier differentiating
> them.
>
> The example in RFC 7483 for a registrar included several addresses and
> phone numbers, as if the sample registrar also had contacts associated with
> it. But none of the addresses and phone numbers were marked with anything
> that would distinguish between the contacts (e.g. a type of "admin",
> "tech", etc.). Should we just take all the data we have for a registrar,
> including all addresses of associated contacts, and throw them into the
> entity JSON object one after another? Can we mark them with the contact
> role as a type? Or does type always need to be "home" or "work”?
>
>
> This is similar to how some RIRs model ISPs. The ISP is an entity, that
> then has child entities for the contacts.
> See http://rdap.arin.net/registry/entity/ARINOPS as an example.
>
> This type of thing may not be so obvious in the RFC because of the way the
> JSON is described. I have an RDAP JCR draft that might make it more
> readable.
> https://tools.ietf.org/html/draft-newton-rdap-jcr-00
>
> RFC 7482 describes searching for an entity by name. In the case of our
> registrars, we have two things that could be a "name": the name of the
> registrar itself, and the name of any associated contact. Am I correct that
> the search is confined to the registrar name? Or are we supposed to look
> through to the contact names, and pass back any registrar with a matching
> contact?
>
>
> How you limit or broaden the search is your choice. RDAP just specifies
> how the query is formulated by the clients. You may have requirements from
> a policy body about how that is to work, but the protocol is not that
> proscriptive.
>
> I hope that helps.
>
> -andy
>
>