Re: "do nothing" proposal for plain literal/xsd:string

On 03/05/11 14:13, Steve Harris wrote:
> On 2011-05-03, at 13:50, Lee Feigenbaum wrote:
>> On 5/2/2011 3:25 PM, Andy Seaborne wrote:
>>> We need to decide what to do about simple literals / xsd:strings
>>> thread:
Nothing is a possibility but it needs to be an active decision, not a
>>> silent default. We then need to feed that back to RDF-WG.
>> I'd propose we do nothing.
>> We already have that datatype("foo") is xsd:string.
> CONCAT() could maybe do with changing, to reduce implementor
> confusion. The only example that stands out is:
> concat("foo"^^xsd:string, "bar"^^xsd:string) ->
> "foobar"^^xsd:string
> "If all input literals are typed literals of type xsd:string, then
> the returned literal is also of type xsd:string"
> We could just remove that phrase, and change the example to "foobar",
> or we could just leave it as it is.

Leave - if it really does get all xsd:strings, not converted to simple 
literals we might as well give back an xsd:string.

>> The other (more important) thing that comes up is BGP matching. BGP
>> matching is defined via "subgraph", which does not seem to
>> reference any formal definition from RDF concepts (please correct
>> me if I'm wrong).
>> ...Actually, in section 2 this is informatively (why?) defined as
>> being based on "RDF graph equivalen[ce]".
>> It's my hope that when/if the RDF WG goes forward with this change,
>> the changes made to RDF will be such that :s :p "foo" and :s :p
>> "foo"^^xsd:string are "RDF graph equivalent", and therefore SPARQL
>> will simply "inherit" that change.
>> Is this hope unreasonable?
> No.

Less sure:
.... Recommend that systems silently convert xs:string data to plain 
so I think SPARQL Update is affected and potentially whether a query 
engine ought to convert syntax.

The latter (query) is less important and we can leave to "common sense". 
  It's SPARQL Update that I think is less clear.

I also think the consequences of the decision in RDF-WG have not been 
discussed enough and so the decision is not immediately easy; there may 
be "problems arising" e.g. tools that expect xsd:string because they 
wrote xsd:strings.


Received on Tuesday, 3 May 2011 14:07:36 UTC