Re: Anyone in support of CONSTRUCT constraints?

On 3/25/2015 17:47, Jerven Tjalling Bolleman wrote:
> As a non WG member, I can only say I would very much miss the 
> CONSTRUCT constraints. Mostly because they allowed us to really write 
> nice error messages. But I would not mind seeing what happens in this 
> regard with just SELECT.

This would become

SELECT ?message ...
WHERE {...

so nothing will change really - we'll use variable names instead. The 
main things I see disappear would be
- multiple result values in the same sh:Error (e.g. multiple sh:value)
- how to create complex path expressions is unclear at this stage.

I am sure work-arounds to both cases exist, and I believe the overall 
trade-off is hopefully negligible compared to the improved simplicity 
and consistency.

Holger


>
> Regards,
> Jerven
>
> On 25/03/15 00:54, Holger Knublauch wrote:
>> In further attempts to simplify the spec, Dimitris had suggested that we
>> remove support for constraints backed by SPARQL CONSTRUCT. Since we have
>> already taken out ASK, this would only leave SELECT queries.
>>
>> I agree that we should assume simplicity for now, and we could bring
>> back CONSTRUCT at some later stage when we really have to.
>>
>> Does anyone have concerns about this process?
>>
>> Thanks,
>> Holger
>>
>>
>

Received on Wednesday, 25 March 2015 08:03:11 UTC