RE: Comment on LDF spec

On 12 Jun 2015 at 10:37, Ruben Verborgh wrote:
>> Only results of CONSTRUCT
>>         (and thus not SELECT, DESCRIBE or ASK) SPARQL queries
>>         are considered Linked Data Fragments.

+1, appears to be the simplest "solution".


> While this is indeed the clarifying explanation,
> I would rather have it outside of the definition then.

What do you mean by that?


>> DESCRIBE is an informative part of the SPARQL 1.1 spec and is therefore
quite
> unpredictable.
> 
> Agree, but note that Linked Data documents are equally unpredictable.
> I.e., you don't get just <s> ?p ?o triples back; typically, also some ?o
?p <s> triples.
> So following that argument, we could also include DESCRIBE queries.
> 
> However, for the sake of clarity and simplicity (it's an example anyway),
> I would prefer the clarifying remark suggested above.
> Would that work?

Yes. We should also create a Wiki page (LDF FAQs?) capturing why this
decision has been made resp. what the arguments are.


--
Markus Lanthaler
@markuslanthaler

Received on Monday, 15 June 2015 20:21:11 UTC