Re: URNs: syntax and registries

Keith Moore (moore@cs.utk.edu)
Tue, 28 Nov 1995 13:09:00 -0500


Message-Id: <199511281809.NAA15179@wilma.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: "William Y. Arms" <warms@CNRI.Reston.VA.US>
Cc: Jim Conklin <conklin@info.cren.net>, uri@bunyip.com, moore@cs.utk.edu
Subject: Re: URNs: syntax and registries 
In-Reply-To: Your message of "Tue, 28 Nov 1995 09:17:10 EST."
             <v02130500ace0c7df2760@[132.151.1.217]> 
Date: Tue, 28 Nov 1995 13:09:00 -0500

> We are in agreement here.  The naming-scheme has an obligation to ensure
> that names are unique.  In all the principal URN proposals, this is
> delegated to naming authorities and the format of <r-name> is of the form:
> 
>   <r-name> := <naming-authority><separator><locally-unique-string>
> 
> My reasons for not spelling this out in the posting were:
> 
> (a)  To allow for the inclusion of naming schems where there is only one
> naming authority (e.g., social security numbers), or where the naming
> authority is buried in the name (e.g., ISBNs).  Thus, the following might
> be a valid URN.
> 
>   URN:SSN:012-34-5678

If we do the registry right, we can still delegate portions of (say)
ISBN space to different servers, by allowing the registry to specify
which ranges of LUI space are sent where.

(Please don't propose SSNs as a type of URN.   They're abused far
too much already, and it's entirely inappropriate to use them as a 
general resource identifer.)
 
Keith