Re: Aggregates question...

http://lists.w3.org/Archives/Public/public-rdf-dawg/2010OctDec/0002.html

has a proposed clarification for this.

 >> ListEval(undefined) = error.

[[
Definition: ListEval

ListEval(ExprList, μ) returns a list E, where Ei =
μ(ExprListi) _or error_.

_ListEval retains errors in evaluation of the list elements._

A variant, ListEvalE, is the same except that all elements of E which
are errors are removed.
]]

It's useful to be able to "group by" bound/error. It's very unhelpful to 
have a query abort just because of the data.  I don't know how that 
would work with the protocol.

 Andy

On 07/10/10 13:51, Axel Polleres wrote:
>
> On 7 Oct 2010, at 05:34, Steve Harris wrote:
>
>> On 2010-10-07, at 02:31, Axel Polleres wrote:
>>
>>> What happens when grouping for UNBOUND variables?
>>>
>>> E.g.
>>>
>>> Graph:
>>> :s1 :p :o1.
>>> :s1 :p :o2.
>>> :s1 :q :o3.
>>> :s2 :p :o1.
>>> :s2 :p :o2.
>>>
>>> Query:
>>>
>>>   SELECT ?Q (COUNT(?P) as ?CP)
>>>   WHERE { ?S :p ?P  OPTIONAL {?S :q ?Q } }
>>>   GROUP BY ?Q
>>>
>>> My current understanding is:
>>>
>>>   that this returns
>>>
>>>    ?Q | ?CP
>>>   =========
>>>   :o3 | 2
>>
>> That's correct. It could be made clearer in the text though.
>>
>>> rather than
>>>
>>>    ?Q | ?CP
>>>   =========
>>>   :o3 | 2
>>>       | 2
>>>
>>> yes?
>>>
>>>
>>> In general, what is unclear to me is what ListEval returns when for ExprList = ?var andμ(?var) undefined?
>>
>> ListEval(undefined) = error.
>
> As far as I read it, this is not entirely clear from the current definition of ListEval...
> ListEval takes an ExpressionList as input... ?Q is an expression list,
> if ?Q is not in the domain of&mu, it is not clear for me that
> μ(?Q) = error, it could likewise just beμ(?Q) = ?Q, or no?
> At least I don't think it's clear from the current definition.
>
> best,
> Axel
>
>
>>
>>> It seems, strictly following the definitionμ(?var) would then just amount to ?var , then again, I don't know what that
>>> should mean for the rest of the aggregation semantics?
>>
>> I'm not sure I understand.
>>
>> - Steve
>>
>> --
>> Steve Harris, CTO, Garlik Limited
>> 1-3 Halford Road, Richmond, TW10 6AW, UK
>> +44 20 8439 8203  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 Thursday, 7 October 2010 13:10:26 UTC