- From: Seaborne, Andy <andy.seaborne@hp.com>
- Date: Tue, 12 Apr 2005 14:51:28 +0100
- To: Jeen Broekstra <jeen@aduna.biz>
- CC: Dan Connolly <connolly@w3.org>, RDF Data Access Working Group <public-rdf-dawg@w3.org>
Jeen Broekstra wrote:
> Dan Connolly wrote:
>
>
>>It's a bit odd that
>>
>> { ?R ?dummy [] }
>>
>>is allowed, and means the same thing, but { ?R [] [] } is
>>not allowed.
>
> >
>
>>>Turtle does not allow bNodes in the property slot. Comments?
>>
>>
>>I don't really have strong feelings either way. I'm happy for
>>the editors to flip a coin (or leave the coin flipped).
>
>
> Given that blank nodes fultill the role of (anonymous) variables, I
> would be in favor of adopting blank-nodes-in-property-slots for SPARQL.
>
> Another, somewhat related, beef I have with the current syntax is that
> it allows the use of blank node labels (_:a) as variables. This is not
> only redundant, but confusing as heck since undoubtedly people will
> misread this as being a blank node label occurring in the queried
> source (instead of a variable name). Is there a strong argument
> against simply removing this feature from SPARQL?
1/ Symmetry and consistency.
2/ Some graph shapes can't be written with [] alone (e.g. 2-connected blank node
objects).
3/ Blank nodes are written _:a in N-triples and can be written that way in
Turtle and N3.
Yes, it can lead to confusion. But it is much the same confusion that blank
node labels are not global in RDF.
Andy
>
> Jeen
Received on Tuesday, 12 April 2005 13:51:44 UTC