Re: BPR: Is redirection permitted?

On 10/15/12 10:15 AM, David Wood wrote:
> On Oct 15, 2012, at 5:20, Andy Seaborne <andy.seaborne@epimorphics.com> wrote:
>
>> On 11/10/12 18:02, Arnaud Le Hors wrote:
>>
>>> I actually agree with you on the fact that RDF and Linked Data are
>>> different - meaning there is a difference between the two, namely:
>>> URIs in Linked Data are dereferencable URLs
> Sorry to respond to the middle of a thread - I'm traveling.
>
> Arnaud's position seems odd to me.  Dereferencable URLs are always good IMO, but should Linked Data become somehow invalid if a URI doesn't dereference?  Of course not.  That suggests that dereferencable URLs are just a Good Thing, but not mandatory, just as they are in RDF.

Linked Data != RDF.

Linked Data mandates that -- via indirection -- a URI that denotes an 
Entity (a Thing) also resolves to content that actually describes the 
entity it denotes. Put differently, a URI resolves to content describing 
its Referent.

Net effect of my statement above is that when a URI denotes any Entity 
(modulo entities of type: Web Document) you end up with two (not one) 
routes to the same content since you have a generic URI that denotes an 
Entity plus another that denotes the Entity's description or descriptor 
document location.

The use of indirection to triangulate access to content (the RDF or EAV 
model based graph) is what Linked Data is all about. RDF isn't about 
data access triangulation via name based indirection it's all about 
representation which is why URIs don't need the behavioral specificity 
of Linked Data.

You end all of this confusion by loosely coupling RDF, EAV, and Linked 
Data.

>
> Sorry to be philosophic, but I think our shared mental model bears on Andy's questions.
>
>> Can a GET on a BPR return 303?
> I certainly hope so.

Ultimately, it's irrelevant since the focal point has to be the Linked 
Data URI that denotes an entity. Opting for implicit (via hash URIs) or 
explicit (via hashless URIs) indirection is a choice to be made by a 
Linked Data publisher.

Links:

1. http://lists.w3.org/Archives/Public/public-rdf-wg/2012Sep/0003.html 
-- what many of us assumed was closure re. this matter .


Kingsley
>
> Regards,
> Dave
>
>> Can BPR URIs have a fragment? (c.f. 4.1.2)
>>
>> The intro to section 4 says that BPRs come from linked data rules and the Linked Data page mentions fragment and 303.
>>
>> Or does the spec not care? (an example with a # would be good in that case)
>>
>>     Andy
>>
>> PS which triggers the thought (unrelated):
>>
>> Should a BPR respond 301 if not accessed by the canonical URL?
>>
>
>


-- 

Regards,

Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen