W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > January to March 2005

Re: ACTION: review "11 Testing Values"

From: Seaborne, Andy <andy.seaborne@hp.com>
Date: Tue, 15 Feb 2005 16:55:18 +0000
Message-ID: <42122976.2090800@hp.com>
To: jos.deroo@agfa.com
CC: public-rdf-dawg@w3.org

jos.deroo@agfa.com wrote:
> | ACTION: JosD to review <http://www.w3.org/2001/sw/DataAccess/rq23/#tests>
> Just one observation at this moment (I really have to study XPath
> more in depth before I can make a reasonable review..) is that I
> have implementation experience that following triples are the case
> "10.1"^^xsd:decimal math:greaterThan "8.4"^^xsd:decimal.
> "P1Y2M"^^xsd:duration math:notGreaterThan "P14MT10H"^^xsd:duration.
> "01:41:00+01:00"^^xsd:time math:lessThan "05:41:00Z"^^xsd:time.
> "1956-01-10"^^xsd:date math:notLessThan "1956-01-10"^^xsd:date.

Allowing for the literals as subject, a small problem if the query itself contains:

   ("10.1"^^xsd:decimal math:greaterThan ?x)

then the use of predicates for tests is perfectly legal in SPARQL - it matches 
patterns to an RDF graph and is that graph happens to contain

"10.1"^^xsd:decimal math:greaterThan "8.4"^^xsd:decimal

then a macth will happen.  It need no support from the spec - it just happens.

In principle, it is permissible for an implementation to compile all constraints 
to triple patterns.  Implementation choice.  Thought of this way, the syntax is 
merely to allow humans to write complex expressions in a natural way.

Note that an implmentation does not have to do the converse - turn triple 
patterns into constraints.  Triple patterns would be matched against the graph 
as normal and if the graph does not contain them, no matching occurs.

It was implementation choice to compile to triple patterns.

[I don't see what happens about the value testing extensibility but its 
extensibility and hence not required of an implementation.  The rules engine is 
may need to understand the type promotion rules.]


> and that one can query such math: and string: and so on [1]
> triples like any other triples..
Received on Tuesday, 15 February 2005 16:57:59 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:00:33 UTC