Re: Annotation syntax [was: SPARQL* test suite]

On 30/08/2020 19:08, Andy Seaborne wrote:
>
>
> On 29/08/2020 20:54, Gregg Kellogg wrote:
>> On Aug 29, 2020, at 12:35 PM, Andy Seaborne <andy@apache.org> wrote:
>>>>>> On 28 Aug 2020, at 03:06, Holger Knublauch 
>>>>>> <holger@topquadrant.com> wrote:
>>>>>
>>> ...
>>>
>>>>> I, for one, keep being unimpressed by the Turtle syntax, which 
>>>>> ends up becoming quite redundant because depending on the 
>>>>> interpretation a triple needs to be stated twice (once for the 
>>>>> actual assertion and once for the reifications). We are for now 
>>>>> using a syntax such as
>>>>>
>>>>> ex:subject ex:predicate ex:object [[
>>>>>      ex:created "2010-10-10"^^xsd:date ;
>>>>> ]]
>>>
>>> Yes, there is room for annotation syntax. Stardog has the annotation 
>>> after the predicate, before the object (guess: to avoid a problem 
>>> with TriG). What happens about object lists?
I wouldn't worry too much about edge cases as long as the primary use 
case (annotation triples about normal triples) is covered. Putting these 
blocks after the triple makes it pretty to print for multiple rows of 
annotations.
>>>
>>> ]] can occur in Turtle data.
>>
>> [[ can occur in Notation-3, which will eventually want compatibility 
>> with RDF*, too.
>>
>>> Given that {} is used for grouping in TriG and SPARQL:
>>>
>>> {%...%}
>>> {|...|}
>>> {!...!}
>>> {{...}}
>>
>> Same for {{ in N3., but I do like the other suggestions.
>
> Thanks for the reminder.
>
> Once upon a time, {$...$} was a proposed N3 syntax for sets as literals.

Any above syntax would work fine as long as any of them gets de-facto 
standard support. For now we have used [[ ... ]] in our tools yet this 
is easy to change, and if this would make it into a future Jena version 
then we can migrate to that implementation instead of our home-baked hack.

Does anyone here see problems with adding such a TTL extension?

Thanks,
Holger

Received on Tuesday, 1 September 2020 23:29:04 UTC