This question is a follow up to Can CONSTRUCT return duplicate triples?, the accepted answer there states the following:
Does "When we communicate the results of a CONSTRUCT query" extend to using an API? The SPARQL working draft(s) probably have nothing to say about this, but what would you expect or demand from an API for executing sparql (construct) queries?
Specifically, would you accept an API construct, e.g. an iterator, that returns duplicate RDF triples or would you be horrified and dismayed?
Yes I would, ultimately behavior in this regard in going to be down to the API designer. Provided the API states what it's behavior is clearly I wouldn't mind either way.
Personally I would prefer an API that doesn't suppress duplicates because this likely makes the API more efficient (no need to maintain state information necessary to check for duplicates e.g. a Set of triples generated so far) and gives the consumer of the API the freedom to choose to suppress duplicates as and when they need to.
answered 06 Nov '12, 10:40
Rob Vesse ♦