Re: [regext] Second WG LAST CALL: draft-ietf-regext-rdap-reverse-search

[Mario Loffredo <[e-mail]>]

Hi Patrick,

thanks for your review.

Please find my comments inline.

Il 25/09/2022 18:21, Patrick Mevzek ha scritto:
> On Mon, Sep 12, 2022, at 08:54, Antoin Verschuren wrote:
>> Please review this document and indicate your support (a simple “+1” is
>> sufficient) or concerns with the publication of this document by
>> replying to this message on the list.
> I should probably have said something earlier, sorry about this.
Better in late than never ;-)
> But I have a concern about §6 Implementation Considerations
> as I think it glances over far too quickly on very important points.
>
> I think it can be easy to expect reverse queries to generate "lots" of results,
> but then all examples given ("restricting the
>     search functionality, limiting the rate of search requests according
>     to the user's authorization, truncating and paging the results, and
>     returning partial responses.") are not given details, which means there will be left
> to implementors and hence multiple incompatible solutions will emerge which will make writing
> a client more complex, for any case where it has to span multiple RDAP servers
> (and then you are exactly in same territory as EPP extensions, too many of them and too incompatible between them to easily write one client working with all servers).
>
> There is RFC 8977 "Registration Data Access Protocol (RDAP) Query Parameters for Result
>                             Sorting and Paging" but it is not even referenced from this draft.
> Same for RFC 8982 "Registration Data Access Protocol (RDAP) Partial Response", shouldn't
> be cited at least as a non-normative reference?

Absolutely. Thank you for this remark. Agree that they could be added as 
informative references.

>
> - "restricting the search functionality" does that mean by things related to the protocol like constraints on `{searchable-resource-type}` or on `{related-resource-type}` or on `<search-condition>` or by things external to it, like rate-limit? How will a client discover that it got limited for any of those reasons?

Do believe that such note can be applied to the RDAP searches in general.

That said, each provider can decide to restrict the usage of the query 
capabilities as he sees fit.

One restriction on generic searches could consist in allowing only those 
partial matching queries including a minimum number of characters before 
the wildcard.

Another one, specific for reverse search, could be to mandate the use of 
the "role" parameter.

The way for servers to signal clients about having issued a search 
request that cannot be processed is defined in section 4 of RFC 9082, 
that is, by returning an error.

For each of the implemented aforementioned restrictions, the RDAP server 
can return an error response including information about the reason of 
the request failure.

> - "truncating and paging the results": maybe mention RFC 8977 and 8982
> - "returning partial responses.": RFC 8982?
Yes, see my previous comment.
>
> But how RFC 8982 would apply here since it is not necessarily the client asking for limited
> data in return but the server deciding to prune them in content or length?
>
> Same question in fact for RFC 8977, that starts with client requesting specific subsets and order.

Don't see any difference in an RDAP server supporting the operators 
defined in both RFC8982 and RFC8977 in this specific search rather than 
in other searches.

The benefits for clients from using such operators are common to all of 
the searches as their implementation supports clients in issuing 
requests that are less likely to be pruned by the server and obtaining 
more manageable responses. Hence, they can achieve relevant information 
in shorter time.

Could you please clarify why those operators would be useless 
specifically here ?

>
> I also dislike the mention of indexes here because this is specific terminology
> of specific technologies and as such I don't believe an RFC describing a protocol
> should lay any assumption or give constraints on how implementers decide to implement it.

Seems to me that the sentence in question works quite well since the 
words "indexes and similar functionalities" are used in their common 
meaning of techniques to speed up the data retrieval.  They don't hint 
at a specific technology.

In addition, the sentence is set as a recommendation in order to guide 
implementers to choose the reverse search properties appropriately.

The proposed reverse search properties are generally "indexed" but the 
document allows the RDAP providers to define additional ones.

Anyway, I would have no problem to change that sentence if there was a 
better way to express the same concept.

For example, would it be fine for you If I changed the sentence to the 
following?

    To limit the impact of processing the search predicates, servers are
    RECOMMENDED to make use of techniques to speed up the data retrieval in their
    underlying data store such as indexes or similar.


Best,

Mario

-- 
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://www.iit.cnr.it/mario.loffredo