afs
commented
3 years ago sameTerm
17.4.1.7 RDFterm-equal defers to RDF Concepts (section 6) for term-equality.
Presumably the definition for RDF-star term equality is
sameTerm(<<S1 P1 O1>>, <<S2 P2 O2>>) == sameTerm(S1, S2) && sameTerm(P1, P2) && sameTerm(O1, O2)
This goes in section 2 of the RDF-star report.
value-equality
17.3 Operator Mapping "A = B" is defaults to "sameTerm(A,B)" as it does for URIs.
The question becomes whether to add a dispatch for "=" that is value-sensitive on literals in a embedded triple.
True or false: << :s :p 123 >> = <<:s :p 123.0>>.
Because embedded triples are triples, and two triples compare by term-equality in a "set of triples" i.e. a graph, we could leave this as term-equality.
Is there a use case for value-equality? If so, does it merit defining "=" to be value-equality of component terms given that graphs are set-term-equality.
Note: DATATYPE(<<:s :p :o>>) is error, just like DATATYPE(<uri>).
ORDER BY
defines an ordering between term types:
(Lowest) no value assigned to the variable or expression in this solution. Blank nodes IRIs RDF literals
Adding to that list should be done.
Comparision of embedded triple terms
What is
<< :s :p 123 >> < <<:s :p 456>> (other than hard to write)
or
?t1 < ?t2 where ?t1 is << :s :p 123 >> and ?t2 is << :s :p 456 >> ?
This pairs with value-equality.
Proposal
- Add embedded-triple term equality to section 2, defined as recursively on the three components.
- Define "=" as
sameTerm(that is, leave as-is). - Add RDF-star terms to the partial order list of
ORDER BY.
JervenBolleman
As pointed out by @lisp and @jeenbroekstra on the mailling list, we need to define how RDF-star triples are compared with each other and with other terms
ORDER BY(Section 15.1 of the SPARQL 1.1 spec)=operator (Section 17.4.1.7)sameTermfunction (Section 17.4.1.8)