To:
wessorh@ar.com (Rick H Wesson)
Cc:
paf@cisco.com (Patrik Fältström), bmanning@isi.edu (Bill Manning), briansp@walid.com (Brian W. Spolarich), george@register.com (George Belotsky), shollenbeck@verisign.com (Hollenbeck Scott), ietf-provreg@cafax.se
From:
Bill Manning <bmanning@isi.edu>
Date:
Thu, 8 Mar 2001 16:07:03 -0800 (PST)
In-Reply-To:
<Pine.LNX.4.30.0103081537240.10073-100000@loki.ar.com> from "Rick H Wesson" at Mar 08, 2001 03:42:41 PM
Sender:
owner-ietf-provreg@cafax.se
Subject:
Re: Unique handle generation
% let me say one thing, the handle WM-9387-19790922:09:27:33 SUCKS! Tell me how you really feel... :) It was an example. % The uniqueness SHOULD NOT be obtained by appending a date/timestamp. Perhaps. Thats one way to increment. I think there should be two distinct salts for each seed. You have selected an abstract thing called Object Identifier e.g. CT... again YMMV. % Furthermore the handle MAY contain a Registry Identifer and the handle % SHOULD contain information that describes what kind of object it points % to. % % For instance WM9387-CT:RIPE % WM - bills initials % 9387 incrementor % CT differientiates this hadle as a contact object % RIPE The optional suffex (pick your seporator [%#@:]) of the Registry of % Record for the Object. % % -rick How are Registry Identifiers different from the handles we have been tossing about? I expect that handle concatination between objects may be useful to narrow scope in -limited- circumstances, the general rule should be to only expose the handle. None of <MY> handles is going to have, as part of <MY> handle, a conncationation that describes some association. Their handles might... :) -- --bill