Hello Pierre, Yes, RDF_IRI and RDF_OBJ are not freed. If RDF_OBJ is free-text idneed then free-text data are freed only partially if CLEAR GRAPH is used and not freed at all if quads are deleted by other means.
We need leapfrog garbage collection for RDF_IRI and RDF_OBJ, that is not yet implemented. (Free-text garbage will be collected by trigger as soon as RDF_OBJ is collected.) Currently we simply pollute disk with all old scrap, that's not good. OTOH we've found that most of scrap is reused sooner or later, and slightly increased depth of the tree of index pages does not affect the speed too much. In addition, leapfrog is really efficient only when there's at least as much garbage as data in use. Leapfrog is not available before this summer, I guess, before we're focusing on RDF Views ATM. The reason is that the amount of relational data available via RDF Views is much greater than the amount of 'dynamic' quads; there are of course big data sets of Linking Open Data Project and the like but they're static so they do not require garbage collection. That's why you're probably the first user who noticed that the garbage is persistent. If needed, we can quickly write non-leapfrog garbage collection but it will be atomic operation or at least will block RDF data updates while running. If you need such then I'd like to know your database sizes (typical and the biggest you've used) and RAM available for Virtuoso (typical, the smallest and the one used for biggest database mentioned before). Best Regards, Ivan Mikhailov, OpenLink Software. On Sat, 2008-02-09 at 16:36 +0100, Pierre De Wilde wrote: > Hi, > > > When clearing a graph using SPARQL CLEAR GRAPH <iri>, it seems that > the table RDF_IRI is not cleared. > I understand that some IRIs may be used by other graphs but not > always. > > > How can I delete RDF_IRI records not used in any (G,S,P,O) columns of > the RDF_QUAD table? > Why not include RDF_IRI deletion as part of SPARQL CLEAR GRAPH > command? > > > Thanks beforehand for your help. > > > Pierre > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ Virtuoso-users mailing list > Virtuoso-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/virtuoso-users
