A Uniform Resource Name (URN) Namespace for Federated Content
RFC 4198

[Ballot comment]
Spencer Dawkins said:

A quick look-through made me wonder about one detail - the whole point is to come up with a (domain name as of some date) convention to uniquely identify a content provider for all time - but the date only requires a four-digit year (CCYY), with month (MM) and day-of-month (DD) optional.

I'm sure this will work fine in practice, but my particular industry is off to the land of mergers and acquisitions again, so I would be more comfortable with requiring at least the month to be part of the date as well. The race condition is "name some resource on January 2, acquire some corporation on January 3, and start reusing domain name components for other purposes in October".

[Ballot comment]
Sepncer Dawkins said:

A quick look-through made me wonder about one detail - the whole point is to come up with a (domain name as of some date) convention to uniquely identify a content provider for all time - but the date only requires a four-digit year (CCYY), with month (MM) and day-of-month (DD) optional.

I'm sure this will work fine in practice, but my particular industry is off to the land of mergers and acquisitions again, so I would be more comfortable with requiring at least the month to be part of the date as well. The race condition is "name some resource on January 2, acquire some corporation on January 3, and start reusing domain name components for other purposes in October".

[Ballot comment]
If I read Bill's message correctly, the question is, should domain names embedded
within example URNs be required to use example.com, example.net, and friends?

I don't think so.  I think the reason we require it in the general run of documents is
to avoid folks reading the documents from coding the examples into the protocol
and to avoid references to domain names which may change owner, content, and
so on in ways which would surprise someone dereferencing the URIs.  Given the
delegation model of the URN, I don't think that's an issue here.

[Ballot discuss]
The MM ABNF allows months "01" through "09" or a month I've never heard of, "1012".  I suspect the second half of the alternation is meant to be ("1" ("0" / "1" / "2")) (the I-D is missing the slashes).

[Ballot discuss]
The MM ABNF allows months "01" through "09" or a month I've never heard of, "1012".  I suspect the second half of the alternation is meant to be ("1" %x30-32).

(my canonicalizer makes this error more obvoius, it turns the rule into:

MM = ( "0" %x31-39 ) / ( "1" "012" )

[Ballot discuss]
Minor ABNF error in section 3:

hex        = DIGIT | %x41-46 | %x61-66

should be:

hex        = DIGIT / %x41-46 / %x61-66

[Note]: 'RFC Editor note: Rules for Lexical Equivalence:       In addition to the rules defined in RFC 2141 [4], normalize the
      case of the ProviderId before comparison. Rules for Lexical Equivalence:       In addition to the rules defined in RFC 2141 [4], normalize the
      case of the ProviderId to lower case before comparison.' added by Ted Hardie

[Note]: 'RFC Editor note:

Rules for Lexical Equivalence:

      In addition to the rules defined in RFC 2141 [4], normalize the
      case of the ProviderId before comparison.

Rules for Lexical Equivalence:

      In addition to the rules defined in RFC 2141 [4], normalize the
      case of the ProviderId to lower case before comparison.' added by Ted Hardie