- From: Steve Harris <steve.harris@garlik.com>
- Date: Mon, 8 Nov 2010 10:35:46 +0000
- To: Lee Feigenbaum <lee@thefigtrees.net>
- Cc: Axel Polleres <axel.polleres@deri.org>, SPARQL Working Group <public-rdf-dawg@w3.org>
On 2010-11-08, at 05:56, Lee Feigenbaum wrote:
> On 11/7/2010 12:55 PM, Axel Polleres wrote:
>> Just to state that (even if we consider RAND() useful, I am not 100% sure whether i like the example.
>>
>> SELECT ?s WHERE { ?s a<Foo> } ORDER BY RAND() LIMIT 10
>>
>> Strictly speaking this has precisely the same semantics as
>>
>> SELECT ?s WHERE { ?s a<Foo> } LIMIT 10
>>
>> i.e. the order is not prescribed if no ORDER BY is given.
>> I know that *for specific implementations* (probably most) the result of two consecutive calls of the former version may be different, whereas
>> two consecutive calls of the latter might not, but I don't think that we should - by such example - indicate that this is intended by the spec
>> and that this is a way which guarantees random results. Opinions?
>
> Yes, I share this thought. I have no a priori objection to RAND() but had the same reaction that this particular example can be pretty misleading as to what the meaning of the same query without ORDER BY RAND() is.
Agreed, I think. I wouldn't advocate using that as an example in the document, just that it's a common real world requirement, which we can't satisfy currently.
There are obvious issues in creating tests for RAND(), but we have an entire verb that's essentially untestable, so I don't think that should be a blocker.
- 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 Monday, 8 November 2010 10:38:09 UTC