- From: <james@dydra.com>
- Date: Sun, 2 Apr 2017 18:19:37 +0000
- To: public-sparql-exists@w3.org
good evening;
> On 2017-04-02, at 20:01, james anderson <james@dydra.com> wrote:
>
> good evening;
> […]
>>
>> What is wrong with this analysis?
>
> your interpretation employs blank nodes in a manner which makes no sense.
it is evident, that something is not clear, despite that it should appear so obvious.
i propose a simple example.
take the query
select * where { ?x <http://example.org/p1> ?p1 . ?x <http://example.org/p2> ?p2 }
and apply it to a repository with blank node terms as subjects.
now consider how one would involve sip in the implementation.
in which domain must the interpreter perform the matches?
given that, in which domain must it perform the statement transformations prior to matching.
may one call those transformations “substitution”?
i suspect, if your execution model conflates blank nodes and nondistinguished variables in such a manner that they become indistinguishable, that will lead to problems beyond those claimed to exist with respect to exists.
best regards, from berlin,
>
> best regards, from berlin.
> ---
> james anderson | james@dydra.com | http://dydra.com
>
>
>
>
>
>
Received on Sunday, 2 April 2017 18:20:11 UTC