"Max Vlasov" <max.vla...@gmail.com> wrote in message news:7cb963ca1001120149u550da7fr56cfc0ed261a9...@mail.gmail.com... > ... Are you sure you have > identical tests for your comparison? I mean these are the same data sets > with the same scheme?
Absolutely, errr, to the best of my knowledge. Same database, same query, but different SQLite managers having different versions of SQLite. But the fact that the SQLite manager is different is immaterial - the developer of one of them was the one to identify that it is the SQLite version that is the controlling variable: 3.6.17 and somewhat earlier - better; 3.6.18 and later - worse. The volatility of SQLite performance over these version changes and with the use of ANALYZE is very disquieting for me. The database structure is that of a popular genealogy software. Having run ANALYZE, now not only has my fastest query been degraded by 70:1, so has the speed of the 'owner' software in generating its reports. Attention needs to be paid to what changed between 3.6.17 and 3.6.18 that has adversely affected performance and why is it that ANALYZE is deleterious. Is it just this database structure that has been affected? Tom _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users