Re: select distinct and ordered from two graphs

Thanks Lee and Paul,
finally I think the unexpected results were (1) due to the shuffled  
labeValues and (2) due to my WHERE clause. The order is as expected in  
the query below.

SELECT DISTINCT ?labelValue FROM :graphA FROM :graphB
WHERE {    ?post :contains :givenTag .
  ?post :contains ?tag .
  ?tag :label ?labelValue .
  ?tag :updated ?date   // before: ?post :created ?date
  }
ORDER BY DESC (?date) ?labelValue
LIMIT 10

Not sure yet why this works and the former does not. Maybe the prob is  
in my data, I will have to re-check this.

Regards,
Mirko



Am 12.09.2009 um 05:27 schrieb Paul Gearon:

> I'm late on this, I know, but all the same...
>
> On Mon, Aug 31, 2009 at 6:06 AM, Lee Feigenbaum  
> <lee@thefigtrees.net> wrote:
>> SELECT DISTINCT should be unrelated to ORDER BY, which in turn  
>> should be
>> unrelated to the graphs that comprise the default graph in your RDF  
>> dataset.
>> If you are seeing a problem, perhaps you should contact the
>> maintainers/supporters of your specific SPARQL implementation.
>>
>> Lee
>
> SELECT DISTINCT will usually reorder data (depending on the
> implementation), but ORDER BY will always override this. In this case,
> it looks like the ordering on labelValue is being shuffled with SELECT
> DISTINCT, which is perfectly fine, since there is no ordering on it
> (only on dates).
>
> I would recommend updating the query to order by labelValue after
> date, thereby fixing the result:
>
> SELECT ?labelValue FROM :graphA FROM :graphB
> WHERE {    ?post :contains :givenTag .
>  ?post :contains ?tag .
>  ?tag :label ?labelValue .
>  ?post :created ?date
>  }
> ORDER BY DESC (?date) ?labelValue
> LIMIT 10
>
> Regards,
> Paul
>
>> Mirko wrote:
>>>
>>> Hi,
>>> I wonder how to solve this. I have the following query:
>>>
>>> SELECT ?labelValue FROM :graphA FROM :graphB
>>> WHERE {    ?post :contains :givenTag .
>>>  ?post :contains ?tag .
>>>  ?tag :label ?labelValue .
>>>  ?post :created ?date
>>>  }
>>> ORDER BY DESC (?date)
>>> LIMIT 10
>>>
>>> Now, I would like to have only distinct ?labelValues as results.
>>> Unfortunately, in my case SELECT DISTINCT messes up the order when  
>>> selecting
>>> from several graphs.
>>>
>>> How could I make the results (?labelValue) of the above query  
>>> distinct
>>> while keeping the order?
>>>
>>> Thanks,
>>> Mirko
>>>
>>>
>>
>>

Received on Wednesday, 16 September 2009 08:11:34 UTC