Re: [urn] URI Scheme with Complex Equality Rules

HI Ted,

I agree that a new URI registration may be most appropriate and if the IETF will accept a reference to a non-IETF specification instead of an RFC then that is viable option.

I will review the procedures and submit a request to uri-review@ietf.org<mailto:uri-review@ietf.org>

Thank you for the help,

Randy

From: Ted Hardie <ted.ietf@gmail.com>
Sent: Saturday, August 27, 2022 1:59 AM
To: Randy Armstrong (OPC) <randy.armstrong@opcfoundation.org>
Cc: urn@ietf.org
Subject: Re: [urn] URI Scheme with Complex Equality Rules

Hi Randy,

On Fri, Aug 26, 2022 at 3:18 PM Randy Armstrong (OPC) <randy.armstrong@opcfoundation.org<mailto:randy.armstrong@opcfoundation.org>> wrote:
Hi Ted,

Thanks for the reply.
I was thinking of a URI scheme instead of a URN NID, however,  it appears URI scheme requires that we create RFC which is a much more cumbersome process.

You don't need to produce an RFC; the registration procedures are laid out here:

https://www.rfc-editor.org/rfc/rfc7595.html#page-11

For a provisional, they are designed to be very simple (since the main goal is to avoid collision).  Even for a permanent registration, the additions do not include producing an RFC:

       1. Review the requirements in Section 3<https://www.rfc-editor.org/rfc/rfc7595.html#section-3>

       2.  Send a copy of the scheme registration request or a pointer

           to the document containing the request (with specific

           reference to the section that requests the scheme

           registration) to the mailing list uri-review@ietf.org<mailto:uri-review@ietf.org>,

           requesting review.  In addition, request review on other

           relevant mailing lists as appropriate.  For example, general

           discussion of URI syntactical issues can be discussed on

           uri@w3.org<mailto:uri@w3.org>; schemes for a network protocol can be discussed

           on a mailing list for that protocol.  Allow a reasonable time

           for discussion and comments.  Four weeks is reasonable for a

           'permanent' registration request.

      3.  Respond to review comments and make revisions to the proposed

           registration as needed to bring it into line with the

           guidelines given in this document.
If you would like to discuss the possibility, we can redirect the discussion to  uri-review@ietf.org<mailto:uri-review@ietf.org>

regards,

Ted Hardie

The OPC Foundation is author/maintainer of IEC 62541 and we are trying to fix the incorrect URN usage in the spec today.
The identifiers are used in the subjectAltName of X509 Certificates and uniquely identify the application that the Certificate is assigned to.

The uniqueness of our identifiers is within the scope of the system so an IP address can be system unique and static.
We specifically do not want to use ‘http’ because people tend to assume they are valid URLs with backing websites.

Regards,

Randy

From: Ted Hardie <ted.ietf@gmail.com<mailto:ted.ietf@gmail.com>>
Sent: Friday, August 26, 2022 10:08 PM
To: Randy Armstrong (OPC) <randy.armstrong@opcfoundation.org<mailto:randy.armstrong@opcfoundation.org>>
Cc: urn@ietf.org<mailto:urn@ietf.org>
Subject: Re: [urn] URI Scheme with Complex Equality Rules

Hi Randy,

Based on a skim of your use case, I can't yet see why minting a URI scheme would not work.

opcid://authority/parameters

The authority here would be the DNS name or IP address, and the definition of the scheme could establish that you wanted case insensitive matching for the parameters (where the context would be stored).  Is there something in IEC 62541 that would make this approach unviable? I admit that  I did not spend any francs to get it, so I don't have examples of what the context might look like, so if you have examples you can share, that would be useful.

Note that while URNs are guaranteed to be unique across the lifetime of an identifier, your scheme doesn't seem likely to want that, as it is relying on the DNS and IP address assignment, neither of which guarantees that property.

regards,

Ted

On Fri, Aug 26, 2022 at 1:47 PM Randy Armstrong (OPC) <randy.armstrong@opcfoundation.org<mailto:randy.armstrong@opcfoundation.org>> wrote:
I am sending this issue to this email as per the suggestion in RFC 8141.

The OPC Foundation (https://opcfoundation.org) has a need to uniquely identify OPC UA (IEC 62541) network resources but we do not want the complexity that comes with the various URN schemes.

Specifically we need:

  1.  Equality checks with case-insensitive string comparisons;
  2.  Human readable strings with uniqueness provided by a DNS name or IP address (always lower case).
  3.  A valid URI.
  4.  R/Q/F components not allowed.

We have been using urns of the form: urn:<dnsname>:<additional context>, however, this is not technically allowed by the URN RFC.
We do not want a syntax that needs to be parsed before any comparison can be done.

We could submit a request for our own nid but we don’t really have any syntax requirements other than the dns name and the case-insensitive string comparisons.

What is the best way forward?

Regards,

Randy Armstrong
OPC Foundation

_______________________________________________
urn mailing list
urn@ietf.org<mailto:urn@ietf.org>
https://www.ietf.org/mailman/listinfo/urn