Re: Graphs Design 6.2

On 25 Apr 2012, at 19:49, Sandro Hawke wrote:

> On Wed, 2012-04-25 at 19:36 +0100, Steve Harris wrote:
>> On 25 Apr 2012, at 14:28, Sandro Hawke wrote:
>> ...
>>>> Hm. I am not sure I understand this restriction. It forces the user to come up with a URI or a blank node for no good reason. Why is it a problem to say that if I say:
>>>> 
>>>> 
>>>> @union .
>>>> { a b c }
>>>> d { e f g }
>>>> h { i j k }
>>>> 
>>>> then the default graph is 
>>>> 
>>>> { a b c .
>>>> e f g .
>>>> i j k .
>>>> }
>>>> 
>>>> What is wrong with that? It is only a shorthand...
>>> 
>>> Yeah, I suppose it's probably okay.    I was thinking of union datasets
>>> as a somewhat different kind of dataset.   
>>> 
>>> If you loaded that into 4store, it would have to make up a graph name,
>>> but maybe that's fine.
>> 
>> It has to anyway, it's a quad store*.
> 
> Good point.
> 
>> But in any case, the idea of @union is antithetical to the way these systems are used, see my other mail on the subject.
> 
> Please note the meeting minutes or my summary -- as clarified during the
> meeting, @union is just syntactic sugar in TriG, shorthand for having to
> repeat all the triples in all the named graphs.
> 
> Does that change your mind about it, at all?

Not really.

The implication of that would be that if I see @union I have to disable an optimisation that's designed specifically to handle that (common) case.

I don't see how it makes ay sense to allow a data format to specify that behaviour, it should normally be up to the query author.

- Steve

-- 
Steve Harris, CTO
Garlik, a part of Experian 
1-3 Halford Road, Richmond, TW10 6AW, UK
+44 20 8439 8203  http://www.garlik.com/
Registered in England and Wales 653331 VAT # 887 1335 93
Registered office: Landmark House, Experian Way, NG2 Business Park, Nottingham, Nottinghamshire, England NG80 1ZZ

Received on Wednesday, 25 April 2012 18:55:12 UTC