Re: DB-5 response

In Jena, the union graph is read-only.  There have been different user 
expectations (even different expectations from the same user!).

e.g. deleting a union-graph triple deleted all copies,
vs deleted one copy (AKA reference counting); some expectation that the 
delete applied each whole instantiated template to each graph, not 
triple-by-triple.  All can be reasonable use cases.

The current outcome has been a read-only union graph.

	Andy


On 20/12/11 09:40, Steve Harris wrote:
> Looks fine to me, with one minor note (probably not worth adding to
> the reply, but it struct me when reading it):
>
> “While queries operate on a dataset that is defined as a merge of
> multiple graphs, any updates must necessarily modify a single graph
> at a time. So it is not possible to state that updates operate on RDF
> Datasets.
>
> While a single INSERT or DELETE template may refer to multiple
> graphs, the triples being specified are always for individual graphs.
> So to remove the same triples from graphs<foo>  and<bar>  there is no
> way to do it with a single pattern in a template, but rather both
> graphs must be mentioned explicitly with that template. ie.:
>
> DELETE { GRAPH<foo>  { ... } GRAPH<bar>  { ... }} …”
>
> There is some possible overlap with "default graph" behaviour
> relating to this comment.
>
> In 4store (and 5store I believe) operations on the "default graph"
> will affect all graphs, as the default graph is the RDF Union of the
> named graphs, so:
>
> DELETE { ?s a<Class> } WHERE { ?s a<Class> }
>
> will affect all named graphs including a triple matching the
> pattern.
>
> I don't know if this is compatible with what's written in the
> document, but it appears to be what users expect.
>
> If other systems have this behaviour then it might be worth
> mentioning, but probably not.
>
> - Steve
>
> On 19 Dec 2011, at 22:56, Paul Gearon wrote:
>
>> Hi,
>>
>> In response to DB-5:
>> http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2011Jul/0017
>>
>>
>>
I've created:
>> http://www.w3.org/2009/sparql/wiki/CommentResponse:DB-5
>>
>> In point 8 there were some comments on the lack of definition of
>> how to define the URI of a graph store. I seem to recall that when
>> this was mentioned a few weeks ago that it had been intentionally
>> specified at its current level. However, since the comments refer
>> to the service description and protocol documents, I'd appreciate
>> any input from Greg or Chimezie.
>>
>> Also, David suggested that the text in section 3.1.3 may benefit
>> from having a decision tree to describe the currently operating
>> dataset (depending on the use of WITH, USING, and GRAPH). I have
>> not done this, but would appreciate any feedback as to whether this
>> ought to be included.
>>
>> The changes mentioned in the response are all in the Overview.xml
>> in CVS.
>>
>> Regards, Paul Gearon
>>
>

Received on Tuesday, 20 December 2011 11:41:29 UTC