"Simon Slavin" ... > > By the way, you might find that swapping the last two lines makes it > faster: > > >> select * from A as A1, A as A2 where A1.name=A2.name and >> A2.position between A1.position - 10 and A1.position + 10 and >> A1.position != A2.position ; > > But you might not: SQLite's optimizer may already have spotted it. > I had opted for the other because it had fewer VM instructions and I thought the != condition might be a faster process than the 'between' and eliminate some points from being processed by 'between'. Of course, the result is highly dependent on the correlation of data - if every point is unique then the != condition is a wasteful test.
Thanks for the feedback, Simon & Igor. Sorry, Peng, if it seems I highjacked the discussion. Hope it was helpful to you, too. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
