Manybubbles added a comment. I've removed this from blocking the BlazeGraph selection - we don't need to make this decision to confirm the selection, which we've already done.
In other news I've run a few tests comparing RDR to the more traditional RDF layout. The results were somewhat inconclusive. When the data is hot in the disk cache my tests were getting the same runtimes. It _felt_ like the RDR data could warm up slightly more quickly but that was probably just noise. I think this ultimately needs more investigation. When I was running my tests I realized a few useful things: 1. SPARQL's blank node syntax makes the value nodes model reasonably terse to query. In fact, so far as I can tell, RDR can get _less_ terse in some cases as well. This is a claim about syntactic sugar - not performance, in case that wasn't clear. 2. Getting the labels of things in a specific language or using a specific fallback chain is a bit clumsy. You have to fetch all these triple and filter them on language. Its much much slower than the actual query portion. TASK DETAIL https://phabricator.wikimedia.org/T90952 REPLY HANDLER ACTIONS Reply to comment or attach files, or !close, !claim, !unsubscribe or !assign <username>. EMAIL PREFERENCES https://phabricator.wikimedia.org/settings/panel/emailpreferences/ To: Manybubbles Cc: Haasepeter, Thompsonbry.systap, Beebs.systap, Manybubbles, Aklapper, Smalyshev, jkroll, Wikidata-bugs, Jdouglas, aude, GWicke, daniel, JanZerebecki _______________________________________________ Wikidata-bugs mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs
