Re: reference needed - versioned documents

Actually I was just trying to find an example of a URI for a document  
for which there was explicit *justification* in using the URI to name  
the document.  The question is not so much about any server "giving"  
any particular response, as about whether when the URI is used in an  
assertion, will anyone will understand what was meant by the URI.  
(Server behavior factors into any assessment of this, but is not  
definitive, as servers can behave inscrutibly, unpredictably, or  
incorrectly.) I think one would be hard pressed to find any  
intelligent, diligent person aware of the URI policy statement  
thinking that the dated URI refers to anything other than that  
particular document (version). The other URI, however, is open to  
many interpretations - maybe it names different things at different  
times (maybe you have to know when the use of the URI was written in  
order to be able to understand assertions using the URI), or maybe it  
names a Fieldingesque "resource" (a time-varying membership-of- 
representation-in-something function), or maybe it names some  
particular thing whose essential properties can be communicated, or  
maybe it names a network endpoint. In any of these interpretations  
there is little guidance in knowing *which* thing of whatever variety  
is being named. In other words, the URI is not of much use as a name  
at the present time.

I'm not arguing for or against any particular meaning for the undated  
URI - just observing that (as everyone in this list knows) there is  
no consensus. To me the question is not "what does the URI mean" but  
"how will the URI help us to communicate among ourselves and why".  
Our best hope for being able to use the undated URI as a name would  
be a credible statement from permitting, enabling, and  
encouraging us to use it in some particular way, as it has made for  
the dated URI.


On Mar 29, 2008, at 8:47 PM, Pat Hayes wrote:

> At 4:28 PM -0400 3/28/08, Jonathan Rees wrote:
>> Could someone please point me to the memo that explains the naming  
>> policy used by that prescribes that dereferencing
>> always gives the same thing but
>> might not?
> I believe the rationale is that both of them give the "same thing",  
> but different things. The dated URI always gives you the version of  
> the publication which was published on the date in question. The  
> undated URI always gives you the version of the publication which  
> is most current on the date you de-reference the URI. You might say  
> that the first gives you a dated draft, the second gives you the  
> actual publication. Actual publication are entities that are  
> themselves, of course, subject to change: but the change is not in  
> the reference mapping from the URI to the publication, but in the  
> publication itself. The very same publication might look different  
> on one date from how it looks on another, c.f. a blog or a newspaper.
> I am pretty sure that this is indeed explained in a W3C note  
> somewhere, but I can't get hold of it right at present.
> Pat
>> Thanks
>> Jonathan
> -- 
> ---------------------------------------------------------------------
> IHMC               (850)434 8903 or (650)494 3973   home
> 40 South Alcaniz St.       (850)202 4416   office
> Pensacola                 (850)202 4440   fax
> FL 32502                     (850)291 0667    cell

Received on Monday, 31 March 2008 12:41:25 UTC