Hello, Let say we create two tables:
create table a(num integer primary key);create unique index a_uq on a(num); create table b(num integer primary key);create unique index b_uq on b(num); Table a is populated with 1 million records and table b with 1000 records. We run analyze and look at sqlite_stat1 table: d|d_uq|1000 1 a|a_uq|1000000 1 Now I run two selects with different orders of joins: select * from a join b using(num); select * from b join a using(num); Corresponding timings are CPU Time: user 0.000000 sys 0.000000 CPU Time: user 0.343750 sys 0.000000 This shows that query analyzer is broken - it does not reorder simple inner join based on collected statistics. It worked OK in 3.6 and first release of 3.7 (although required _uq indexes in addition to primary key since no statistics on primaries were used). Please advice. Peter. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
