Re: [urn] corporate URN namespaces

[Peter Saint-Andre <[e-mail]>]

On 2/2/24 4:55 PM, Peter Saint-Andre wrote:
> On 1/31/24 12:03 PM, Peter Saint-Andre wrote:
>> On 1/29/24 1:14 PM, Dale R. Worley wrote:
>>> Peter Saint-Andre <[e-mail]> writes:
>>>> I'm curious how community participants and expert review team members
>>>> see the matter.
>>>
>>> Hmmmm...  I'm not so concerned by circumscribed organizations creating
>>> their own pseudo-URNs for their internal usage.  I am concerned, though,
>>> about a namespace that's meant to be used to interoperate between
>>> systems operated by many organizations, but the details of the namespace
>>> are not made public.  This happens particularly when a trade group
>>> standardizes a bundle of technology but keeps the details secret.
>>> "*This* part of the Internet is private to members of the club!"  Even
>>> in the situation of the Thread Group, where it seems anybody can obtain
>>> the details, but nobody is allowed to discuss them in public.
>>
>> To your first point about circumscribed organizations, when I worked 
>> at Cisco 10+ years ago, engineers on a project I was advising wanted 
>> to create a namespace "urn:cisco". That would have been convenient for 
>> them, but no one proposed reasonable governance for the namespace 
>> (e.g., Cisco has multiple CTO offices for different product lines, so 
>> it wouldn't have been easy to centralize issuance of URNs, although 
>> perhaps folks could have used "sub-namespaces"). In the end I believe 
>> the project decided to use HTTP URIs as XML namespace names, not URNs.
>>
>> But yes, that kind of thing would have been tucked away in a single 
>> company's product. I suppose the same is true of what LinkedIn and 
>> Amazon are doing in the examples I mentioned. We need to ask ourselves 
>> whether we really care about such things. Perhaps I'm making a 
>> mountain out of a molehill...
> 
> Thanks to a message sent to IANA on a related topic, I just now found 
> documentation for the urn:li namespace:
> 
> https://learn.microsoft.com/en-us/linkedin/shared/api-guide/concepts/urns?context=linkedin%2Fcontext
> 
> This usage is especially problematic because it violates Section 5.1 of 
> RFC 8141, which states the following constraint on NIDs:
> 
>     3.  It MUST be more than two characters long, and it MUST NOT start
>         with ALPHA ALPHA "-", i.e., any string consisting of two letters
>         followed by one hyphen; such strings are reserved for potential
>         use as NIDs based on ISO alpha-2 country codes [ISO.3166-1] for
>         eventual national registrations of URN namespaces (however, the
>         definition and scoping of rules for allocation of responsibility
>         for such country-code-based URN namespaces are beyond the scope
>         of this document).  As a consequence, it MUST NOT start with the
>         string "xn--" or any other string consisting of two letters
>         followed by two hyphens; such strings are reserved for potential
>         representation of DNS A-labels and similar strings in the future
>         [RFC5890].
> 
> (Note that "li" is the country code for Liechtenstein.)
> 
> I'm going to try again with some of our contacts at Microsoft.

After much digging, I might have found a repository where I could file 
an issue with the LinkedIn team, so I've done that:

https://github.com/linkedin/rest.li/issues/976

Next, Amazon!

Peter