Re: Updated definitions (constraint)

Dan Connolly wrote:
> On Jun 17, 2005, at 8:51 AM, Seaborne, Andy wrote:
> 
>>Dan Connolly wrote:
>>
>>>On Jun 16, 2005, at 7:44 AM, Dan Connolly wrote:
>>>
>>>>On Jun 16, 2005, at 7:15 AM, Seaborne, Andy wrote:
>>>>
>>>>
>>>>> In a group, all the group elements have to "pass" a solution for  
>>>>>the groiup to match a solution.
>>>>
>>>>Ah... that's what I was missing. Thanks.
>>>>
>>>>http://www.w3.org/2001/sw/DataAccess/rq23/#defn_GroupGraphPatter
>>>
>>>Hmm... what about this case?
>>>SELECT ?x WHERE { FILTER ( ?x < 10 ) }.
>>>That's got a bzillion solutions, right? Maybe we need a special case  
>>>for that?
>>>Or maybe it's already syntactically illegal or something?

If {} is a pattern with one solution of no variables, then

{ FILTER ( ?x < 10 ) } is the same as { {} FILTER ( ?x < 10 ) }

and has no solutions.  That would seem to be tieing FILTERS to graph patterns, 
rather than treating them as first class pattern.

So the answer to your question is partially dependent on the answer for the 
refined optionals.

	Andy

>>
>>I think it depends on the evaluation of filter.  As things stand, I  
>>read that as an attempt to do "?x < 10" with no bound x hence false.   
>>That right, Eric?
> 
> 
> Er... what's in Eric's head is not that relevant. Maybe Eric can find  
> part of the spec
> that explains, but...  My point is that the definition of constraint
> doesn't say anything about "bound x"; i.e. the definition does not say  
> "take the solutions
> you got from other parts of the pattern and check them" as I  
> proposed... and even
> if it did, the empty pattern has every solution, so it would still need  
> something else
> to say why this is false.
> 
> 
>>	Andy
>>
>>
>>See also:
>>http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2005Jun/ 
>>0034.html
> 
>

Received on Tuesday, 28 June 2005 17:13:44 UTC