Re: [regext] Benjamin Kaduk's No Objection on draft-ietf-regext-org-11: (with COMMENT)

(Also inline)

On Wed, Oct 24, 2018 at 05:09:07PM +0800, Linlin Zhou wrote:
> Dear Benjamin,
> Thanks for your review. Please find my feedbacks below with [Linlin].
> 
> Regards,
> Linlin
> 
> 
> Linlin Zhou
>  
> From: Benjamin Kaduk
> Date: 2018-10-24 01:38
> To: The IESG
> CC: regext-chairs; pieter.vandepitte; draft-ietf-regext-org; regext
> Subject: [regext] Benjamin Kaduk's No Objection on draft-ietf-regext-org-11: (with COMMENT)
> Benjamin Kaduk has entered the following ballot position for
> draft-ietf-regext-org-11: No Objection
>  
> When responding, please keep the subject line intact and reply to all
> email addresses included in the To and CC lines. (Feel free to cut this
> introductory paragraph, however.)
>  
>  
> Please refer to https://www.ietf.org/iesg/statement/discuss-criteria.html
> for more information about IESG DISCUSS and COMMENT positions.
>  
>  
> The document, along with other ballot positions, can be found here:
> https://datatracker.ietf.org/doc/draft-ietf-regext-org/
>  
>  
>  
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
>  
> Some of the element descriptions (e.g., <org:postalInfo>) appear to be
> duplicated in several places and are also rather long in prose form.  Is
> there value in attempting to consolidate the structural definition to a
> single place in the document and just refer to that structure from the
> places where it can appear?
> [Linlin] This document borrowed the text structure from RFC5731, RFC5732 and RFC5733 of EPP. I think the intension of having some duplicated descriptions is for users' easy reading. When seeing the examples, they do not have to scroll up and down to find the elements definitions. Some descriptions are a little different, although <org:postalInfo> elements appear to be duplicated. Such as, "One or more <org:status> elements" in <info> response and  "Zero or more <org:status> elements" in <create> command. Of course putting all the elements definitions in a section is a concise way for the document structure.

It sounds like it is not worth making this change for this document, then.
Thanks for the background explanation!

>  
> Section 1
>  
>    There are many entities, such as registrars, resellers, DNS service
>    operators, or privacy proxies involved in the domain registration
>    business.  These kind of entities have not been formally defined as
>    an object in EPP which will be specified as "organization" in this
>    document.
>  
> nit: run-on sentence.  I suggest:
>    These kind of entities have not been formally defined as having
>    an object in EPP. This document provides a way to specify them as
>    "organization" entities.
> [Linlin] OK.
>  
> Section 2
>  
>    The XML namespace prefix "org" is used, but implementations MUST NOT
>    depend on it and instead employ a proper namespace-aware XML parser
>    and serializer to interpret and output the XML documents.
>  
> I suggest mentioning more explicitly that "org" is used in the examples as
> shorthand for the full namespace "urn:ietf:params:xml:ns:epp:org-1.0";
> draft-ietf-regext-allocation-token would be a fine example to look at.
> [Linlin] Yes. "The XML namespace prefix "org" is used for the namespace "urn:ietf:params:xml:ns:epp:org-1.0".
>  
> Section 3.4
>  
>    Status values that can be added or removed by a client are prefixed
>    with "client".  Corresponding status values that can be added or
>    removed by a server are prefixed with "server".  The "hold" and
>    "terminated" status values are server-managed when the organization
>    has no parent identifier [Section 3.6] and otherwise MAY be client-
>    managed based on server policy.
>  
> The list/descriptions that follows shows several that don't start with
> "client"/"server", including some not mentioned here.  Are we supposed to
> assume that these "unprefixed" values are also server-managed?
> [Linlin] Yes, other status values that do not begin with either "client" or "server" are server-managed. I'll add this setence to clarify.
>  
>    o  ok: This is the normal status value for an object that has no
>       pending operations or prohibitions.  This value is set and removed
>       by the server as other status values are added or removed.
>  
> I guess this is intended to be parsed as "(pending operations) or
> (prohibitions)", but could also be parsed as "pending (operations or
> prohibitions)".  Perhaps "operations pending or active prohibitions" is
> less prone to misreading.
> [Linlin] OK.
>  
> In general, the sort of "all combinations are permitted, except for these
> restrictions" approach taken here can lead to some non-sensical
> combinations, if insufficient care is taken by the document authors.  I did
> not attempt to validate all possible combinations, but do note that (e.g.)
> we make statements about "linked" in combination with "ok" and
> "client/serverLinkProhibited", but not about "linked" in combination with
> "terminated" or several other status values.  The first of those cases
> serves as a limitation on "ok", and the second would seem to be intended to
> clarify that an apparent conflict of status is permissible, and so it may
> well be okay to leave as the default ("everything goes") for other
> combinations, but I hope that the WG has done a careful analysis here.
> It may also be useful to list what considerations were used in this
> analysis, in case there is ever a need to add a new status value (in which
> case the analyses would need to be performed anew for the added value(s)).
> [Linlin] The combination statuses of "ok", "linked" and "client/serverlinkProhibited" are allowed in the first example. But "The organization which has been terminated MUST NOT be linked." which is mentioned in the "terminated" definition. So the server will provide service to update the "linked" status for those terminated organizations.
> We listed the conflicted combinations of the statuses. Other status combinations not expressly prohibited may be used.
>  
> Section 3.4
>  
> (Same comment as above re "pending operations or prohibitions")
> [Linlin] I think you mean the "ok" definition in section 3.5. This will be modified as "operations pending or active prohibitions" as well.

Yes; thank you.

> Section 3.6
>  
>    Take a reseller organization, for example, the parent identifier is
>    not defined for the top level reseller, namely the registrar of the
>    registry.  [...]
>  
> nit: this also looks like a run-on sentence; I'd suggest something like
> "The case of reseller organizations provides an example.  The parent
> identifier is not defined [...]"
> [Linlin] OK. "The case of reseller organizations provides an example. The parent identifier is not defined for the top level reseller, namely the registrar of the registry"
>  
>    Loops MUST be prohibited.  For example: if organization A has B as
>    its parent identifier, organization B should not have organization A
>    as its parent identifier.  The same is true for larger loops
>    involving three or more organizations.
>  
> I'd suggest s/should not/cannot/
> [Linlin] Yes.
>  
> Section 4.1.1
>  
>    In addition to the standard EPP command elements, the <check> command
>    MUST contain an <org:check> element.  This element or its ancestor
>    element MUST identify the organization namespace.  [...]
>  
> "the organization namespace" is perhaps ambiguous; am I correct in
> inferring that this refers to the full "urn:ietf:params:xml:ns:epp:org-1.0"
> namespace value as assigned to the "org" short name?  (I'll refrain from
> repeating this comment every time it applies.)
> [Linlin] Yes, I'll add the full namespace. 
> "This element or its ancestor element MUST identify the organization namespace "urn:ietf:params:xml:ns:epp-1.0"."
>  
> Section 4.1.2
>  
> The <org:addr> restrictions seem somewhat contrived/artificially
> restricted; for example, there are postal addresses in the US with no
> associated city.  Whether an organization would want to use such an address
> as its contact location is another question, but I don't have a clear model
> of what sort of constraints are intended to be applied by this element.
> [Linlin] Sorry that I am not familiar with the US address. If no city is available, how about municipal or county information?

The type of place I'm thinking of would usually be referred to as
"unincorporated DeKalb County" or occasionally "unincorporated Chicago"
with some associated municipal or city information.  (As something of a
side note, I currently reside in Saint Louis city, which is not part of any
county; there is a distinct Saint Louis County that covers the surrounding
areas.  Addresses can be hard; just like (personal) names!)

> This <postalInfo> element have the same structure with the <postalInfo> defined in RFC5733. This is an existing XML schema in the running EPP system. Domain registries and registrars have already implemented it. Acturally this is an optional info here. You can have no <postalInfo>, one <postalInfo> or two <postalInfo> elements.

I definitely do not want to suggest diverging the definition of
<postalInfo> from RFC5733, so please treat this entire conversation as a
side note with no actions to take.

> Section 4.2.1
>  
> Just to check my understanding, the <org:creData> contains only a short
> list of fields because the server is required to either respect the various
> <org:role>, <org:postalInfo>, etc. in the <org:create> request or to return
> an error?  That is, the client would not need to immediately perform an
> <info> query to confirm the status of the organization object at the
> server.
> [Linlin] The result codes which are defined in RFC5730 will return in the response. Client do not need to perform <info> immediately. If the command has been processed successfully, client will recieve "1000". If fails, client may recieve such as  "2001" means "Command syntax error" to give a hint.

Thank you

> Section 4.2.2
>  
> Is there value in an example of the 2305-error response?
> [Linlin] I think the example of 2305 error would like follows,
> S:<?xml version="1.0" encoding="UTF-8" standalone="no"?> 
> S:<epp xmlns="urn:ietf:params:xml:ns:epp-1.0"> 
> S: <response> 
> S: <result code="2305"> 
> S: <msg lang="en">Object association prohibits operation</msg> 
> S: </result> 
> S: <trID> 
> S: <clTRID>ABC-12345</clTRID> 
> S: <svTRID>54321-XYZ</svTRID> 
> S: </trID> 
> S: </response> 
> S:</epp>
>  
> This example is much the similar with the existing one except for the "code" and "msg".

This is true, which suggests that there is not much value in having the
additional example.  Thank you for showing me what it would look like,
though.

> Section 4.2.5
>  
> The elements in <org:add>/<org:rem> vs. <org:chg> seem to be disjoint sets;
> what factors went into deciding to split them this way?
> [Linlin] From my understanding, <org:add>/<org:rem> are used for the one-multiple relations. And <org:chg> is used for the one-one replacement.
>  
> Section 4.3
>  
>              The status of the corresponding object MUST clearly reflect
>    processing of the pending action.  [...]
>  
> It's not entirely clear how this sentence is to be interpreted.  From
> context, I assume that the intent is that <info> queries and similar must
> report that the appropriate pendingFoo status values, but a literal reading
> would seem to have this sentence be a requirement that the server change
> what it reports for the object status, once the action is actually taken.
> (Which is also something desired, but arguably already required by other
> text.)
>  [Linlin] The "transform command" is mentioned in the previous text. So the status means "status in the response", no need to perform the <info> query. So how about changin like "The status in the response of the corresponding object...".

I think that's an improvement, thanks.  (There might be further
improvements possible, but we probably don't need to look for them.)

>    The status of the organization object after returning this response
>    MUST include "pendingCreate".  The server operator reviews the
>    request offline, and informs the client of the outcome of the review
>    either by queuing a service message for retrieval via the <poll>
>    command or by using an out-of-band mechanism to inform the client of
>    the request.
>  
> I don't think the "either" is appropriate; the earlier text *requires* the
> service message, and allows for additional optional notification
> mechanisms.
>  [Linlin] This <poll>mechanism is supported in section 2.9.2.3 of RFC5730. So this is a easy and convinient way to inform the clients.

I'm saying that the choice for the server is not "do X or do Y", it's: "do
X, then either do Y or don't do Y".  The word "either" here implies that it
is sometimes acceptable to not do X (where X is the queuing of the service
message).

> (side question: what's the mnemonic for "pan" in "panData"?  "pending
> action"?  Ah, the full schema suggests "pending action notification".
> Also, why is the top-level a "pan" prefix but the children just "pa"?)
> [Linlin] This section was adopted from RFC5731. 
> I think "pan" is used to tell clients the following messages are "pending action notification". But for a particular element, this is just a "pending action" tag, not the notification message.

Ah, that makes sense -- thanks

-Benjamin

> Section 7.3.1
>  
>    Registrant Name: For Standards Track RFCs, state "IESG".  For others,
>    give the name of the responsible party.
>  
> Just to clarify, is the intended behavior for non-standards-track
> IETF-stream RFCs that the registrant is one of the RFC authors?  I could
> see a case that "IESG" would work for all IETF-stream documents, not just
> standards-track ones.
> [Linlin] OK.
>  
>    Registrant Contact Information: an email address, postal address, or
>    some other information to be used to contact the registrant.
>  
> Perhaps a side note, but postal address in particular has come up
> frequently in GDPR discussions, with the question of whether it is either
> needed or useful.
>  [Linlin] This element can be "zero" if you do not like to provide.
> Section 9
>  
> This document is pretty boring from the security perspective (to be clear:
> that is a good thing!).  The only thing that came to mind is that in one of
> the examples, we show the client asking for <org:id>s of res1523, re1523,
> and just 1523.  Only "re1523" was in use, indicating that the other two
> would be free for new allcations.  In some contexts this kind of "very
> similar looking" identifier can be problematic, especially when a human is
> called upon to verify or compare the value(s).  From what I understand of
> EPP usage, that doesn't seem likely to be a concern here, but I mention it
> in case my understanding is incorrect or incomplete.
> [Linlin] Thank you.
>  
>  
> _______________________________________________
> regext mailing list
> regext@ietf.org
> https://www.ietf.org/mailman/listinfo/regext