Re: [Gen-art] Gen-art LC (and telechat) review for draft-ietf-regext-epp-rdap-status-mapping-01

[Robert Sparks <[e-mail]>]

Responses inline -


On 10/10/16 10:28 AM, Gould, James wrote:
> Robert,
>
> Thank you for your review and feedback.  I provide responses to your 
> feedback below.
>
> —
>
>
> JG
>
>
>
>
> *James Gould
> *Distinguished Engineer
> [e-mail]
>
> 703-948-3271
> 12061 Bluemont Way
> Reston, VA 20190
>
> VerisignInc.com <http://VerisignInc.com>
>
>> On Oct 5, 2016, at 4:58 PM, Robert Sparks <[e-mail] 
>> <mailto:[e-mail]>> wrote:
>>
>> I am the assigned Gen-ART reviewer for this draft. The General Area
>> Review Team (Gen-ART) reviews all IETF documents being processed
>> by the IESG for the IETF Chair.  Please treat these comments just
>> like any other last call comments.
>>
>> For more information, please see the FAQ at
>>
>> <http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.
>>
>> Document: draft-ietf-regext-epp-rdap-status-mapping-01
>> Reviewer: Robert Sparks
>> Review Date: 5 Oct 2016
>> IETF LC End Date: 10 Oct 2016
>> IESG Telechat date: 13 Oct 2016
>>
>> Summary: This draft is on the right track but has open issues, 
>> described in the review.
>>
>> Major Issue:
>>
>> Many of the descriptions describe only side-effects of the status 
>> instead of the status itself.
>>
>> All of the descriptions for the new rdap status codes start with "For 
>> DNR that indicates". This implies that there is a "For not DNR" case 
>> that's not discussed. I don't think the phrase is necessary and each 
>> description should look more like the other descriptions already 
>> registered at 
>> http://www.iana.org/assignments/rdap-json-values/rdap-json-values.xhtml.
>>
>> For instance, at 'auto renew period' the document currently says:
>>
>> "For DNR that indicates if the object is deleted by the registrar 
>> during this period, the registry provides a credit to the registrar 
>> for the cost of the auto renewal"
>>
>> That discusses something (and not the only thing) that can happen 
>> while the object is in that state. It does not describe the state.
>>
>> I suggest it should instead say (based on the text in 3915 and the 
>> current registry entry style):
>>
>> "The object instance is in a grace period provided between when its 
>> registration period expires and when its registration is 
>> automatically renewed by the registry."
>>
>> I don't think it's important to include the commentary about 
>> providing a credit if the entity is deleted by the registrar during 
>> this period, but since that commentary exists in 3915, you can 
>> include it if you want. The _important_ part to convey is the actual 
>> status.
>
>
> The “For DNR that indicates” can be removed from the descriptions. 
>  For example, the "addPeriod = add period; For DRN that indicates if 
> the object is …”  mapping could be "addPeriod = add period; If the 
> object is …”.  The purpose of this draft is to map the statuses 
> defined in EPP and RDAP, so the status descriptions included in the 
> draft where taken from the EPP RFC’s.  There is no intent to redefine 
> the statuses included in the EPP RFC’s in anyway.
But you are not including the entire EPP definition for most of these - 
you are only copying in _part_ of it, and it's not the important part.
Looking at -02 of the draft, you currently have this:

    addPeriod = add period;  If the object is deleted by the client
        during this period, the server provides a credit to the client
        for the cost of the registration.

Where did you take the definition out of the EPP suite though?
On a fast skim, I assumed you took it from this statement in RFC3915:

    addPeriod: This grace period is provided after the initial
       registration of a domain name.  If the domain name is deleted by
       the registrar during this period, the registry provides a credit
       to the registrar for the cost of the registration.


You left out "The grace period is provided after the initial 
registration of a domain name" which is what the the status _is_. That's 
what the status code is conveying. The extra words about credit after 
deletion are commentary about things that can happen while the object is 
in that state.

(And you're already changing words by using "the client" instead of "the 
registrar".)

Maybe you took the state definition from some other place?

Many of the other definitions in this document have that same problem.

>
>>
>> All of the descriptions will need similar attention. Some of them 
>> (such as clientUpdateProhibited) currently have 2119 words in the 
>> description. That doesn't make sense - this is a status, not an 
>> protocol instruction, and trying to put normative language in a 
>> registry will lead to confusion about where the behavior you are 
>> trying to describe is actually defined. (To be fair, 5731 has this 
>> same problem). Again, I suggest following the style that's already in 
>> the registry and say something like "The client has requested that 
>> any requests to update this object instance be rejected."
>>
>>
>
> The clientUpdateProhibited status is defined as:
>
> clientUpdateProhibited = client update prohibited;  For DNR that
>         indicates the client requested that requests to update the object
>         (other than to remove this status) MUST be rejected.
>
> Where do you see 2119 words in the clientUpdateProhibited description? 
>  The status descriptions were taken from the EPP RFC’s with no intent 
> on changing their meaning.
You copied it - above - it's the MUST.
This is 5731's issue - that MUST should have been in text about what 
servers do with requests received while the object is in that state, 
instead of being part of the state definition, and the state description 
in a registry.
I understand not wanting to risk introducting confusion by restating a 
definition since you are simply wanting to take the EPP definitions 
completely, so it's probably the better trade-off to propagate that 
problem rather than fix it in this document.

>> Minor Issues:
>>
>> You're setting up a minor maintenance headache for any future work 
>> that might update this document by having the descriptions listed in 
>> two places. I don't think it's necessary to list the descriptions in 
>> section 2 (currently the bulk of page 4 and the beginning of page 5). 
>> Instead, stop after the paragraph that ends at the top of page 4, and 
>> note that the descriptions of each new status code are provided in 
>> section 3.
>
> The desire was for section 2 to stand on its own to define the 
> statuses and the mapping, and for section 3 to be used to register the 
> statuses in registry.  I believe it would be cleaner to duplicate the 
> descriptions in this instance.
As I note, this is a minor issue, but I disagree. Cleaner for _who_? 
It's certainly not cleaner for the anyone who has to revise this 
document (and it's not cleaner for you as the editor of this document or 
the RFC editor since you have to make any change in two places, risking 
having the document become internally inconsistent). I don't see how 
it's cleaner for the implementer of the specification either.


>>
>> Nits:
>>
>> Near the end of page 3, the document says "In the DNR, the client and 
>> server prohibited statuses are separate an RDAP MUST support the same 
>> separation." There are several nits to address with this. That MUST 
>> is not a good use of 2119. DNR hasn't been expanded (and "the DNR" is 
>> not particularly clear).
>>
>> I suggest you replace that sentence, and the one immediately before 
>> it with:
>>
>> "EPP provides status codes that allow distinguishing the case that an 
>> action is prohibited because of server policy from the case that an 
>> action is prohibited because of a client request. The ability to make 
>> this distinction needs to be preserved in RDAP.”
>>
>
> This change will be made.
>
>>
>