Re: CommentResponse:HK-3

On 03/06/11 11:02, Steve Harris wrote:
> On 2011-06-03, at 10:24, Andy Seaborne wrote:
>>
>> On 03/06/11 09:51, Steve Harris wrote:
>>> While the answer is correct, I don't think this fully answers
>>> his query. Seems like what he's asking for is for unknown
>>> datatypes to be treated as numeric types (unless there's more
>>> information).
>>
>> I know the second part does but the first part is about "more
>> openness and flexibility to SPARQL's handling of typed literals."
>>
>>> To be blunt, this sounds plain crazy to me, e.g.
>>> "23"^^xsd:hexBinary< "24"^^xsd:hexBinary and "23"^^xsd:hexBinary<
>>> "ff"^^xsd:hexBinary is a trivial example of where it will cause
>>> user confusion.
>>
>> I agree the proposal of auto-numeric is the wrong one.
>>
>> "2011-06+01:00"^^xsd:gYearMonth is "confusing" as a number.
>>
>> Suggested changes of wording?
>
> Maybe add something like:
>
> "The group feels that implicit casting of unknown datatypes to
> numerical types would be counter productive. The intended mechanism
> for datatype extension already exists in SPARQL 1.0..."
>
> There's also the issue that he's using the datatyping mechanism to
> fudge units, which as I understand it isn't really correct, but
> that's a separate issue.

I'm reading the second part of Holger's message as an idea-of-the-moment 
because he had/has a particular issue to deal with.

I'll made some changes including:
"""
The group feels that implicit casting of unknown datatypes to numerical 
types would be a less useful mechanism.
"""

Please (re)check,

	Andy

Received on Friday, 3 June 2011 10:16:58 UTC