[regext] Fwd: New Version Notification for draft-gould-regext-rdap-redacted-00.txt

[Mario Loffredo <[e-mail]>]

Hi James,

please find my comments below prefixed by [ML].



-------- Messaggio Inoltrato --------
Oggetto: 	Re: [regext] New Version Notification for 
draft-gould-regext-rdap-redacted-00.txt
Data: 	Wed, 14 Jul 2021 14:06:47 +0000
Mittente: 	Gould, James <jgould=[e-mail]>
A: 	[e-mail] <[e-mail]>, 
[e-mail] <[e-mail]>
CC: 	[e-mail] <[e-mail]>



Mario,

Thank you for your review and feedback. I provide replies to your 
feedback embedded below.

-- 
  JG



James Gould
Fellow Engineer
[e-mail] <applewebdata://13890C55-AAE8-4BF3-A6CE-B4BA42740803/[e-mail]>

703-948-3271
12061 Bluemont Way
Reston, VA 20190

Verisign.com <http://verisigninc.com/>

On 7/14/21, 3:58 AM, "Mario Loffredo" <[e-mail]> wrote:

     Hi all,

     Il 12/07/2021 13:26, Gould, James ha scritto:
     > Marc,
     >
     > Thank you for the quick review and feedback.  Below are responses to your early comments:
     >
     >      - would be good to include specific text about jscontact, so when we switch to it, this document does not need rev.
     >
     > Agreed, that was thought about while drafting, but we initially left it out.  I'm confident that the extension will work with JSContact, but some text would help for clarity.

     I guess it will work even better than with jCard! JSContact is more     object-oriented than jCard and the json path expressions are simpler     since maps are mostly used to represent collections (e.g.

     "$.entities[?(@.roles[0]=='registrant')].jscard.phones.voice.phone"     instead of     "$.entities[?(@.roles[0]=='registrant')].vcardArray[1][?(@[1].type=='voice')]")

JG - Yes, JSContact is much easier to express via JSONPath, since there is no dependency on the use of JSON arrays.

     Here in the following a first feedback from my side about the document:

     - if, as it seems, the "name" property is unique in the "redacted"     array, I would prefer to define "redacted" as a map where the "name"     value is the map key. In theory, every RDAP information can be redacted     but, in practice, the number of RDAP properties usually redacted is     limited. If the possible "name" values were standardized, il would be     easier for RDAP clients implementers to check and pick a property inside     the collection of redacted properties.

JG - Agreed, it would be ideal for the "name" property to unique and easy to key off of by the client.  The JSONPath expression provides the technical reference, while the "name" property can represent the logical reference.  We didn't want to go down the IANA registry route in draft-gould-regext-rdap-redacted, but if we add a RDAP JSON Values type for the redaction reason, we could also add an RDAP JSON Values type of the redaction name.  The question would be whether all "name" property values would need to be registered in the RDAP JSON Values registry.  If that is not the case, then we would need change the "name" member to be a JSON object with the "description" and "type" members used for registered and unregistered values in RFC 9083.

     - another possible pre-defined value for "pathLang" might be     "jsonpointer" as an alternative syntax to indentify a value within a     JSON doc.

JG - I don't believe JSON Pointer will work for RDAP based on the need for the conditional expressions (e.g., "$.entities[?(@.roles[0]=='registrant')]").  Do you see a way that JSON Pointer could be used for RDAP?  I would only add it if it will work for RDAP.
     

[ML] - I think it depends on whether the "jsonpath" is meant as a way to 
identify a generic RDAP response field or a specific field in the 
current RDAP response. Surely, unlike JSONPointer, JSONPath allows to 
select a field nested in an array regardless of the position. Therefore, 
it is able to represent multiple cases through the same 
respresentations. However, an RDAP profile might use conventions (e.g. 
the registrant is always the first entity in the array of entities 
associated to a domain) making the use of JSONPath filters redundant. In 
additon, there are cases where JSONPointer can be used 
straightforwardly: for example, a entity lookup response using JSContact 
:-) . Anyway, I agree with you that JSONPath fits better than JSONPointer.


- it seems to me that the document assumes that the role used to     identify the redacted entity within the array of entities should be in     the first position of "roles" array. Maybe such assumption should be     added somewhere in the document.

JG - That is the convention, but if there are multiple roles, the server could choose to use a different role value reference.  There is the option for a contact to serve multiple roles ("registrant", " administrative", "technical", "billing"), but it's unclear whether that is actually used.  If the multi-role contact is redacted by role, then it will add complexity from a redaction perspective, but it can be done.  In the end, I believe this will come down to a implementation consideration item.

[ML] - Generally, an entity is redacted without regard to the role but based on other conditions: whether the entity represents an individual or not, whether the entity have expressed the explicit consent for publishing or not. At least for ccTLDs, it is normal that an entity is both the registrant and the amministrative contact linked to a domain.


  - there is a typo in the json path expression of  "Registrant Street"

JG - What typo do you see in the JSON Path expression of "Registrant Street".  I tested the "Registrant Street" JSONPath expression "$.entities[?(@.roles[0]=='registrant')].vcardArray[1][?(@[0]=='adr')][3][:3]" against the unredacted RDAP response, with the result being:

[

   "",

   "Suite 1235",

   "4321 Rue Somewhere"

]

[ML] - Understood. You considered both extended address and post office box as street components. I have always considered the street information made up only by the third ADR component beacuse RFC6350 recommends not to use the first two components to increase interoperability (see below). That being said, I admit that your JSONPath is able to cover any street representation.
        Unlike the use in RFC8977, a JSONPath expression is allowed to select an array of values in order to communicate that all the values are redacted.

        "Experience with vCard 3 has shown that the first two components
       (post office box and extended address) are plagued with many
       interoperability issues.  To ensure maximal interoperability,
       their values SHOULD be empty." (RFC6350 section 6.3.1)




[ML] - Another feedback about implementing this extension in search responses. Provided that this is not the .it case where searches are allowed only to users legitimated to see the full contact information, I wonder how RDAP servers permiiting searches more or less publicly might implement it.
        Returning the list of all the redacted JSON values included in the response seems to me inefficient. On the other hand, relying on the JSONPath capability to select a result set, it would be wrong to use a single generic JSONPath expression
       (e.g. use "$.domainSearchResults[*].entities[?(@.roles[0]=='registrant')].vcardArray[1][?(@[0]=='email')]" to communicate that the emails of all domains' registrants are redacted) because, in general, the same property in different results could be either redacted or unredacted.
        Do you think the document should address this topic and how?



     Best,

     Mario


 >
 > - I really would like to have an IANA registry for the « reason » 
property. Because this would be potentially displayed to the user, and 
to avoid having all variations of the same reason in different words. A 
registry of reserved words would also facilitate translation in multiple 
languages.
 >
 > We had a desire to stay away from needing to setup an IANA registry 
for redaction. I view the reason property as not a good candidate for an 
IANA registry, since it's meant to be a human readable informational 
value that includes normative language not to be used as a client 
processing dependency. This is a good discussion item.
 >
 > Thanks,
 >
-- Dr. Mario Loffredo
Technological Unit “Digital Innovation”
Institute of Informatics and Telematics (IIT)
National Research Council (CNR)
via G. Moruzzi 1, I-56124 PISA, Italy
Phone: [phone]
Web: 
http://secure-web.cisco.com/1F-PFooJyrWLg51lyW_I49Zvwy3moYXof6OwGqaZON9_0DcwS-s4HC3KttHlW4-GTU-3j7eTkRexCFzAuK9qR5TFk1Va8rwSoh15OdINxFIShnToELNciLpmFWsMNghBtzhTGEfz81tSFv-8jpnVkIF56EAcX0m47l_U0xLdYxjWxvkBykLE6XGyZkw-IHz_8LfdHOV6Ja-dDltW2jsD7CdSMhuy3B4g2ihEL5ekGFFizXUM5jM8rQrdTICQAhGxBqTHe6CwiveYGEUunTsQJxA/http%3A%2F%2Fwww.iit.cnr.it%2Fmario.loffredo


_______________________________________________
regext mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/regext