Yes, it is TDB2 and a lot of order triples. I did not try this combination, it works. The documentation really needs an update for this... Thank you very much.
On Sat, Dec 18, 2021 at 8:37 PM Andy Seaborne <a...@apache.org> wrote: > > > On 18/12/2021 16:58, Dragan Lesic wrote: > > Hello, > > I'm trying to update (delete+insert) some data which has also reification > > in it (order of cast members for a movie for example). > > When doing delete all triples for a subject that order data seems not to > be > > deleted. > > For example i need to change the order. > > After inserting new data, when doing queries I get the new and old data > > back. > > Executing a delete for RDF-star data is extremely slow. > > > > > > Queries: > > Query data (with order): > > > > PREFIX sub: <https://example.org/movie/> > > PREFIX shema: <https://schema.org/> > > SELECT DISTINCT ?order ?name > > WHERE { > > sub:SOMEID shema:cast ?o {|shema:order ?order|} . > > ?o shema:name ?name . > > } > > ORDER BY ?order > > > > Returns: > > order > > name > > > > "0"^^xsd:integer > > "Sigourney Weaver"@en > > > > "1"^^xsd:integer > > "Annie Potts"@en > > > > "1"^^xsd:integer > > "Annie Potts"@en > > ... > > This is wrong, I have the order 1 twice because of an earlier insert. I > > want to delete it and insert correctly. > > > > > > Then delete everything: > > PREFIX sub: <https://example.org/movie/> > > DELETE > > WHERE { sub:SOMEID ?p ?o . } > > > > Then when I insert all data again, with the correct data and no > duplicates > > (I've checked) and i get the duplicate reference again. > > > > When trying to delete all references first it works but is extremely > slow, > > takes minutes to complete: > > > > PREFIX sub: <https://example.org/movie/> > > PREFIX shema: <https://schema.org/> > > delete > > where { << sub:SOMEID shema:cast ?o >> shema:order ?order } > > > > Any ideas what I am doing wrong, or how to make it faster? > > Thanks in advance. > > > > Hi, > > What's the storage? TDB2? > How much data? > > I guess there are a lot of "shema:order" triples. > > <<>> isn't indexed (there have been no changes in database layout for > RDF-star) so it needs to do: > > ?X shema:order ?order > then check ?X with << sub:SOMEID shema:cast ?o >> > > You could give it a starting point: > > > DELETE > { << sub:SOMEID shema:cast ?o >> shema:order ?order } > WHERE { > sub:SOMEID shema:cast ?o {|shema:order ?order|} ; > } > > or the use case in one request: > > ---- > DELETE WHERE { > sub:SOMEID shema:cast ?o {|shema:order ?order|} . > } > ; > DELETE WHERE { sub:SOMEID ?p ?o . } > ---- > > Andy >
