Re: HTTP Update and HTTP verbs

On 12 Oct 2009, at 16:46, Chimezie Ogbuji wrote:
>
>> Section 5.1 says that a POST is equivalent to:
>> INSERT [ INTO <uri_of_knowledge> ] { .. RDF payload .. }
>> This implies that a POST to a graph that has not already been PUT to
>> will fail, which seems like it would be both surprising and
>> inconvenient. As a user I would expect something more like:
>>
>> CREATE SILENT GRAPH <uri>
>> INSERT [ INTO <uri> ] { .. RDF payload .. }
>>
>> Nothing that I can see in RFC 2616 implies that it should be an error
>> to POST to a currently non-existent resource.
>
> Understood.  I was just held up by the redundancy between the use of  
> POST on
> a non-existent Request-URI and the use of PUT in the same scenario  
> (they
> amount to the same operation).  However, if HTTP is ambiguous about  
> this,
> then the redundancy is the same as that of any other application that
> supported both these verbs.  I have updated the language, example  
> SPARQL
> update language, and removed the note.

It's only redundant if you know that the graph is empty, you can't  
always be sure of that.

>> Also, I find the term "URI of knowledge" a bit opaque. It's not a
>> phrase that appears often enough on Google for e.g. to be in common
>> usage, so I'd have though something like "graph URI" would be more
>> easily understood? If that's what it means.
>
> I was trying to be consistent in discussing the 'information resources
> identified by URIs used in the HTTP operations that are represented  
> by RDF
> graphs in a dataset.'  This is a mouthful and the term 'Networked RDF
> knowledge' was meant to be a shorthand for this.  In addition the  
> subtle
> (but confusing) fact that URIs of graphs in a dataset don't identify  
> the
> graphs themselves but what the graphs represents further confuses this
> notion.
>
> I don't claim it is the best phrase to use (and in fact I would  
> appreciate
> *any* suggestions), but I think distinguishing all information  
> resources
> from those that are represented by RDF graphs is important to tie  
> down with
> specific lexicon.

I see. That seems like quite a difficult to express concept, I don't  
have a suggestion for a better phrase I'm afraid. Kjetil's terminology  
seemed to be getting there, but I didn't quite follow the implications  
of it.

- Steve

-- 
Steve Harris
Garlik Limited, 2 Sheen Road, Richmond, TW9 1AE, UK
+44(0)20 8973 2465  http://www.garlik.com/
Registered in England and Wales 535 7233 VAT # 849 0517 11
Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10  
9AD

Received on Monday, 12 October 2009 16:17:12 UTC