Re: Clarification on rdf_reifies requirement for triple terms

Dear Dominik,

On 27/09/2024 14:47, Dominik George wrote:
> Hi RDF-star WG,
>
> exploring triple terms and reified triples more closely, I find that
> both [RDF12-TURTLE] and [RDF12-CONCEPTS] assume that triple terms
> are only ever used as the object in triples with the rdf:reifies
> predicate.
>
> Is this only a semantical requirement within the entailment regime
> assumed by these documents?
>
> Or is it a technical limitation? If so, how is it to be enforced?
>
> What, in the eyes of the WG, would happen if I used a triple term as the
> object of a triple with some arbitrary predicate?

At this stage, it depends who you ask in the working group! :-)

This was the topic of heated discussions, but my *personal* feeling (any 
official hat off) is that we are converging to:

* this use of triple terms will be allowed in the abstract syntax, but 
discourage by best practices
    (just like using rdf:first and rdf:rest to build ill-formed list is 
allowed, but generally frowned upon)

* some implementations may reject such triples, and still be considered 
compliant
   (just like some implementation today reject "foo"^^xsd:integer, 
despite it being a valid literal under the simple entailment)

pa

> Kind regards,
> Nik
>
> [RDF12-TURTLE]https://www.w3.org/TR/rdf12-turtle/

> [RDF12-CONCEPTS]https://www.w3.org/TR/rdf12-concepts/