W3C home > Mailing lists > Public > uri@w3.org > February 2009

Re: URN duri and tdb spec updated

From: Ray Denenberg, Library of Congress <rden@loc.gov>
Date: Fri, 6 Feb 2009 17:00:46 -0500
Message-ID: <281001c988a6$5d947990$18af938c@lib.loc.gov>
To: <uri@w3.org>

Hi Larry.  I have two comments;  one brief, one a bit longer. Posting to uri 
list only, as requested.


1. Why is the date not based on ISO 8601? Why invent a new date format?


2. I'm having trouble understanding how one determines what a tbd 
identifies.  Take the case of the ietf identifier, used as an example in the 
document.  I suppose it is easy to surmise what

                          urn:tdb:2001:http://www.ietf.org

identifies.  If you go to   www.ietf.org, you get a web page that clearly 
indicates that the IETF is "described" and so I presume we are to conclude 
that IETF is what the URI identifies.

(And nevermind the date, 2001 in this case. I'm leaving the presence of the 
date within the tbd out of this discussion for now because I haven't yet 
decided how I feel about it. It does add a dimension of complexity, though.)

But the underlying assumption seems to be that the URI within the tdb urn is 
going to resolve to a resource that tells you right up front what it 
describes. If that's an assumption, it isn't clear from the document. And if 
it is, then most of the URIs in the world are not candidates for constuction 
of a tbd.

But more to the point,  what then does

                       urn:tdb:2008:http://en.wikipedia.org/wiki/IETF

identify?  Also the IETF?

Then is one of the following statements true:
(1) The two URIs identify the same thing, and can thus be used 
interchangeably.  (Again, forget about the date for purposes of this 
discussion.)
(2) They identify different things. The first identifies the IETF as 
described by the IETF. The second identifies the IETF as described by 
wikipedia.
(3) Only one of the two is an identifier for the IETF (presumably the first, 
because it is somehow authoritative).

I assume (3) is not a possibility (it wasn't a serious suggestion).   I 
suspect the answer is not (2) because I can't see how this is useful. So I 
conclude the answer is (1).   I think this means that there can potentially 
be an infinite number of URIs (not even taking into account the date 
dimension) coined to identify the IETF. Is this a good thing?

Anyway, I'm simpy suggesting that this may warrant clarification in the next 
draft.

--Ray


----- Original Message ----- 
From: "Larry Masinter" <LMM@acm.org>
To: <danbri@danbri.org>; <www-tag@w3.org>; <uri@w3.org>; "'Harry Halpin'" 
<hhalpin@ibiblio.org>
Cc: <urn-nid@ietf.org>
Sent: Tuesday, February 03, 2009 11:06 PM
Subject: URN duri and tdb spec updated


>
> ((Please follow up only on uri@w3.org (DO NOT REPLY ALL)))
>
> By recent popular demand, I updated (slightly)
>
>         http://larry.masinter.net/duri.html (and .txt and .xml)
> and submitted it to the internet-drafts repository as
>
>           draft-masinter-dated-uri-05.txt.
>
> The only substantial change I made since the 2004 draft was to change the
> interpretation of the date from "first instant" to "last instant", based 
> on
> a comment by Al Gilman in 2004.
>
> Replies to recent comments:
> ======
> Ray Denenberg asked:
>> http://info-uri.info/   Have you considered 'info:' for duri and tbd?
> Yes, I considered 'info:/' instead of URN namespaces. I don't think it 
> fits
> into 'info' because 'duri' and 'tdb' aren't really naming authorities in 
> the
> same way. This is more like urn:uuid:...
>
> I thought about switching to plain URI schemes (without "urn:", e.g.,
> uri:2008:whatever instead of urn:duri:2008:whatever), but it would be a
> larger change to the draft, and require more explanation.
> ========
> Pat Hayes wrote:
>> ... the date should be optional.
>
> There's no point in 'duri' without a date. I think even 'tdb' needs a date
> of interpretation, because even when resources are unchanging, the date of
> interpretation matters.
> ========
> Stuart Williams wrote:
>>  ..  something of a year 10K (or maybe 100K) problem
>
> The reference to RFC 2550 hints at how to solve that problem.
> =========
> Harry Halpin and Tim Kindberg wrote reviews or comments in 2004
> which I responded to in email, but I didn't update the document
> based on their comments and my replies.
>
> Larry
> -- 
> http://larry.masinter.net
>
>
>
>
>
> 
Received on Friday, 6 February 2009 22:01:31 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 13 January 2011 12:15:42 GMT