nico at cryptonector.com
Thu May 22 15:29:04 PDT 2014
On Thu, May 22, 2014 at 5:16 PM, Dave Crocker <dhc at dcrocker.net> wrote:
> On 5/22/2014 3:09 PM, Nico Williams wrote:
>> Not really. I explicitly don't want URLs. I want something where
>> location is NOT part of the URI. That makes it a URN.
> Now that this thread has iterated on the theory which justifies the term
> URN, perhaps there can be attention paid to the underlying concern that
> has been expressed (at least twice) about having the string be... well,
> ummm, you know... useful.
> For example, what is the existing, automated mechanism that will make
> the URN resolve to the referenced document?
The mechanism in RFC2648 is basically algorithmic:
Process of identifier resolution:
A mirrored copy of the underlying documentation is required
to resolve these URNs. Resolution via HTTP is done by a set
of simple Perl cgi-bin scripts presented in Appendix A.
The I2L cgi-bin script resolves URNs to URLs that are... local to the
server running the I2L cgi-bin script.
An algorithmic mapping to URL local parts relative to the URL of any
bibxml service is all that is needed to find bibxml entries.
Ditto txt/HTML/PDF/... renderings of the documents.
Obviously some mirrors will be "blessed" in that they are run by the
IETF or the RFC-Editor. The RFC-Editor, in particular, must use such
a mirror when preparing a new RFC, for obvious reasons.
More information about the rfc-interest