Message-Id: <199410210225.WAA16589@cs.bu.edu>
Date: Thu, 20 Oct 1994 22:36:59 -0500
To: uri@bunyip.com
From: "David G. Durand" <dgd@cs.bu.edu> (David G. Durand)
Subject: Re: "why isn't IETF using FPIs?"
>Mitra,
>
>If there is a legacy issue, perhaps another approach would work better.
>
>First off, the Internet (IETF?) could obtain either or both of an object
>identifier (root) or a registered owner-name and assign object identifiers
>and/or public identifiers on the basis of that.
>
>Secondly, I don't know what the domain name syntax is, but if it is compatible
>>with the ISO 9070 object-name component syntax you could grandfather existing
>
>domain names into the 9070 public identifier space by establishing a
>registered
>Internet owner-name (say <Internet-name>) and mapping existing domain names to
>"<Internet-name>//domain-name" ("//" being the separateor between the
> owner-name and the object-name. The Internet, as owner of <Internet-name>,
>>could also append one or more owner-name components to <Internet-name) to
>create additional owner-names.
More in line w/ 9070, since each domain name is a registration authority
(a "name owner" in ISO-ese, would be to have an owner-name like this:
<Internet-name>::domain-name::object-ID-scheme
(the last component on the assumption that we may need several object
naming schemes for different purposes, LIFNs, etc. -- and that a site might
use names of more than one such form)
Object-names themselves would be assigned according either to some set
of official schemes, (like the MD5 proposals), or at the whim of the owner.
In the ISO model, it is the name-owner's responsibility to ensure that the
same object name is not re-used for a "different" object, and no guarantees
are made about the form of object-names (except uniqueness and syntax)
-- David