Re: Two thoughts on Turtle

>>> Would it be useful to allow bare dates/dateTimes?

Because this has been mentioned (elswhere, before) I wanted to get all 
the synatx possibilities on the list.

>>>
>>> :x dc:date 2010-06-08 .
>>
>> I would like to see that. It would complete the set of syntactic sugar for the commonly used XSD types, and I'd hope that it'd eventually also make its way into SPARQL implementations. Although there might be problems here; this is valid SPARQL:
>>
>>     SELECT (2011-06-09 as ?today) WHERE { }
>>
>> but it doesn't do what you think :-)

:-)

Well, it does for me, but maybe not what everyone might see:

It would be an xsd:date - greedy tokenizing means it is not a numeric 
expression:  2011 - 06 - 09

Seriously, some reduced, indicator syntax would be better:

date(2011-06-09)
dt(2011-06-09T00:00:00Z)

which is less nice but better than the clunky ^^xsd:dateTime.

But we're not starting from a  clean sheet so it probably fails the 
"sufficent value for change from submission" test.

> xsd:date is not an required SPARQL type, but xsd:datetime is, and 2011-06-09T00:00:00Z is not an integer :)
>
> That said, I'm not a parser expert, and I'm scared of things that seem like they ought to be fine, but cause crazy shift-reduce conflicts with sensible future syntax.
>
> SPARQL already caused itself a lot of pain by making WHERE optional, and not allowing "," between clauses e.g. in SELECT and ORDER BY, it makes the project expression syntax grotty, for one.
>
> - Steve

OK - It does not seem there is sufficient value and Turtle-SPARQL 
alignment is much more important.

	Andy

Received on Saturday, 11 June 2011 17:36:46 UTC