Activity log for bug #901399

Date Who What changed Old value New value Message
2011-12-07 20:36:46 Elena Stepanova bug added bug
2011-12-07 20:37:29 Elena Stepanova maria: milestone 5.3
2011-12-07 20:37:38 Elena Stepanova maria: assignee Sergey Petrunia (sergefp)
2011-12-07 21:32:36 Elena Stepanova description The following query CREATE TABLE t1 ( c INT ) ENGINE=MyISAM; INSERT INTO t1 VALUES (0),(1),(2),(3),(4),(5), (6),(7),(8),(9),(10),(11),(12); CREATE TABLE t2 ( a INT, b INT, KEY(a)) ENGINE=MyISAM; INSERT INTO t2 VALUES (3,20),(2,21),(3,22); produces an extra row on test data if it's run with materialization=OFF and otherwise default optimizer_switch values, while it returns the correct result set with materialization=ON (or with semijoin=OFF, or with loosescan=OFF). EXPLAIN with materialization=OFF (wrong result): 1 PRIMARY alias2 ALL NULL NULL NULL NULL 13 100.00 1 PRIMARY alias3 ALL a NULL NULL NULL 3 66.67 Using where; LooseScan 1 PRIMARY alias4 ALL NULL NULL NULL NULL 3 100.00 Using where; FirstMatch(alias3) 1 PRIMARY alias1 ALL NULL NULL NULL NULL 13 100.00 Using where; Using join buffer (flat, BNL join) select `test`.`alias1`.`c` AS `c`,`test`.`alias2`.`c` AS `c` from `test`.`t1` `alias1` semi join (`test`.`t2` `alias3` join `test`.`t2` `alias4`) join `test`.`t1` `alias2` where ((`test`.`alias3`.`a` = `test`.`alias2`.`c`) and (`test`.`alias1`.`c` = `test`.`alias2`.`c`) and (`test`.`alias4`.`b` = `test`.`alias3`.`b`)) EXPLAIN with materialization=ON (correct result): 1 PRIMARY alias1 ALL NULL NULL NULL NULL 13 100.00 1 PRIMARY alias2 ALL NULL NULL NULL NULL 13 100.00 Using where; Using join buffer (flat, BNL join) 1 PRIMARY <subquery2> eq_ref distinct_key distinct_key 10 func,func 1 100.00 2 MATERIALIZED alias3 ALL a NULL NULL NULL 3 100.00 2 MATERIALIZED alias4 ALL NULL NULL NULL NULL 3 100.00 Using where select `test`.`alias1`.`c` AS `c`,`test`.`alias2`.`c` AS `c` from `test`.`t1` `alias1` semi join (`test`.`t2` `alias3` join `test`.`t2` `alias4`) join `test`.`t1` `alias2` where ((`test`.`alias2`.`c` = `test`.`alias1`.`c`) and (`test`.`alias4`.`b` = `test`.`alias3`.`b`)) Minimal optimizer_switch: materialization=off Full optimizer_switch: index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_merge_sort_intersection=off,index_condition_pushdown=on,derived_merge=on,derived_with_keys=on,firstmatch=on,loosescan=on,materialization=off,in_to_exists=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on,subquery_cache=on,mrr=off,mrr_cost_based=off,mrr_sort_keys=off,outer_join_with_cache=off,semijoin_with_cache=off,join_cache_incremental=on,join_cache_hashed=on,join_cache_bka=on,optimize_join_buffer_size=off,table_elimination=on Reproducible with MyISAM and InnoDB, but not Aria. Further reducing the test data eliminates the problem as the plan is not triggered anymore. Test case: SET optimizer_prune_level=0; SET optimizer_switch = 'materialization=off'; CREATE TABLE t1 ( c INT ) ENGINE=MyISAM; INSERT INTO t1 VALUES (0),(1),(2),(3),(4),(5), (6),(7),(8),(9),(10),(11),(12); CREATE TABLE t2 ( a INT, b INT, KEY(a)) ENGINE=MyISAM; INSERT INTO t2 VALUES (3,20),(2,21),(3,22); SELECT * FROM t1 AS alias1, t1 AS alias2 WHERE ( alias1.c, alias2.c ) IN ( SELECT alias3.a, alias3.a FROM t2 AS alias3, t2 alias4 WHERE alias3.b = alias4.b ); # End of test case # Notes on result: # The correct result is 2 rows # a a # 2 2 # 3 3 # The test produces 3 rows # a a # 2 2 # 3 3 # 3 3 bzr version-info revision-id: igor@askmonty.org-20111206214218-lev0vyidyt8h3431 date: 2011-12-06 13:42:18 -0800 build-date: 2011-12-08 00:34:55 +0300 revno: 3334 branch-nick: maria-5.3 Also reproducible on revno 3315 The following query SELECT * FROM t1 AS alias1, t1 AS alias2 WHERE ( alias1.c, alias2.c ) IN ( SELECT alias3.a, alias3.a FROM t2 AS alias3, t2 alias4 WHERE alias3.b = alias4.b ); produces an extra row on test data if it's run with materialization=OFF and otherwise default optimizer_switch values, while it returns the correct result set with materialization=ON (or with semijoin=OFF, or with loosescan=OFF). EXPLAIN with materialization=OFF (wrong result): 1 PRIMARY alias2 ALL NULL NULL NULL NULL 13 100.00 1 PRIMARY alias3 ALL a NULL NULL NULL 3 66.67 Using where; LooseScan 1 PRIMARY alias4 ALL NULL NULL NULL NULL 3 100.00 Using where; FirstMatch(alias3) 1 PRIMARY alias1 ALL NULL NULL NULL NULL 13 100.00 Using where; Using join buffer (flat, BNL join) select `test`.`alias1`.`c` AS `c`,`test`.`alias2`.`c` AS `c` from `test`.`t1` `alias1` semi join (`test`.`t2` `alias3` join `test`.`t2` `alias4`) join `test`.`t1` `alias2` where ((`test`.`alias3`.`a` = `test`.`alias2`.`c`) and (`test`.`alias1`.`c` = `test`.`alias2`.`c`) and (`test`.`alias4`.`b` = `test`.`alias3`.`b`)) EXPLAIN with materialization=ON (correct result): 1 PRIMARY alias1 ALL NULL NULL NULL NULL 13 100.00 1 PRIMARY alias2 ALL NULL NULL NULL NULL 13 100.00 Using where; Using join buffer (flat, BNL join) 1 PRIMARY <subquery2> eq_ref distinct_key distinct_key 10 func,func 1 100.00 2 MATERIALIZED alias3 ALL a NULL NULL NULL 3 100.00 2 MATERIALIZED alias4 ALL NULL NULL NULL NULL 3 100.00 Using where select `test`.`alias1`.`c` AS `c`,`test`.`alias2`.`c` AS `c` from `test`.`t1` `alias1` semi join (`test`.`t2` `alias3` join `test`.`t2` `alias4`) join `test`.`t1` `alias2` where ((`test`.`alias2`.`c` = `test`.`alias1`.`c`) and (`test`.`alias4`.`b` = `test`.`alias3`.`b`)) Minimal optimizer_switch: materialization=off Full optimizer_switch: index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_merge_sort_intersection=off,index_condition_pushdown=on,derived_merge=on,derived_with_keys=on,firstmatch=on,loosescan=on,materialization=off,in_to_exists=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on,subquery_cache=on,mrr=off,mrr_cost_based=off,mrr_sort_keys=off,outer_join_with_cache=off,semijoin_with_cache=off,join_cache_incremental=on,join_cache_hashed=on,join_cache_bka=on,optimize_join_buffer_size=off,table_elimination=on Reproducible with MyISAM and InnoDB, but not Aria. Further reducing the test data eliminates the problem as the plan is not triggered anymore. Test case: SET optimizer_prune_level=0; SET optimizer_switch = 'materialization=off'; CREATE TABLE t1 ( c INT ) ENGINE=MyISAM; INSERT INTO t1 VALUES   (0),(1),(2),(3),(4),(5),   (6),(7),(8),(9),(10),(11),(12); CREATE TABLE t2 ( a INT, b INT, KEY(a)) ENGINE=MyISAM; INSERT INTO t2 VALUES (3,20),(2,21),(3,22); SELECT * FROM t1 AS alias1, t1 AS alias2 WHERE ( alias1.c, alias2.c )   IN (       SELECT alias3.a, alias3.a         FROM t2 AS alias3, t2 alias4         WHERE alias3.b = alias4.b      ); # End of test case # Notes on result: # The correct result is 2 rows # a a # 2 2 # 3 3 # The test produces 3 rows # a a # 2 2 # 3 3 # 3 3 bzr version-info revision-id: igor@askmonty.org-20111206214218-lev0vyidyt8h3431 date: 2011-12-06 13:42:18 -0800 build-date: 2011-12-08 00:34:55 +0300 revno: 3334 branch-nick: maria-5.3 Also reproducible on revno 3315
2011-12-07 22:09:53 Elena Stepanova description The following query SELECT * FROM t1 AS alias1, t1 AS alias2 WHERE ( alias1.c, alias2.c ) IN ( SELECT alias3.a, alias3.a FROM t2 AS alias3, t2 alias4 WHERE alias3.b = alias4.b ); produces an extra row on test data if it's run with materialization=OFF and otherwise default optimizer_switch values, while it returns the correct result set with materialization=ON (or with semijoin=OFF, or with loosescan=OFF). EXPLAIN with materialization=OFF (wrong result): 1 PRIMARY alias2 ALL NULL NULL NULL NULL 13 100.00 1 PRIMARY alias3 ALL a NULL NULL NULL 3 66.67 Using where; LooseScan 1 PRIMARY alias4 ALL NULL NULL NULL NULL 3 100.00 Using where; FirstMatch(alias3) 1 PRIMARY alias1 ALL NULL NULL NULL NULL 13 100.00 Using where; Using join buffer (flat, BNL join) select `test`.`alias1`.`c` AS `c`,`test`.`alias2`.`c` AS `c` from `test`.`t1` `alias1` semi join (`test`.`t2` `alias3` join `test`.`t2` `alias4`) join `test`.`t1` `alias2` where ((`test`.`alias3`.`a` = `test`.`alias2`.`c`) and (`test`.`alias1`.`c` = `test`.`alias2`.`c`) and (`test`.`alias4`.`b` = `test`.`alias3`.`b`)) EXPLAIN with materialization=ON (correct result): 1 PRIMARY alias1 ALL NULL NULL NULL NULL 13 100.00 1 PRIMARY alias2 ALL NULL NULL NULL NULL 13 100.00 Using where; Using join buffer (flat, BNL join) 1 PRIMARY <subquery2> eq_ref distinct_key distinct_key 10 func,func 1 100.00 2 MATERIALIZED alias3 ALL a NULL NULL NULL 3 100.00 2 MATERIALIZED alias4 ALL NULL NULL NULL NULL 3 100.00 Using where select `test`.`alias1`.`c` AS `c`,`test`.`alias2`.`c` AS `c` from `test`.`t1` `alias1` semi join (`test`.`t2` `alias3` join `test`.`t2` `alias4`) join `test`.`t1` `alias2` where ((`test`.`alias2`.`c` = `test`.`alias1`.`c`) and (`test`.`alias4`.`b` = `test`.`alias3`.`b`)) Minimal optimizer_switch: materialization=off Full optimizer_switch: index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_merge_sort_intersection=off,index_condition_pushdown=on,derived_merge=on,derived_with_keys=on,firstmatch=on,loosescan=on,materialization=off,in_to_exists=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on,subquery_cache=on,mrr=off,mrr_cost_based=off,mrr_sort_keys=off,outer_join_with_cache=off,semijoin_with_cache=off,join_cache_incremental=on,join_cache_hashed=on,join_cache_bka=on,optimize_join_buffer_size=off,table_elimination=on Reproducible with MyISAM and InnoDB, but not Aria. Further reducing the test data eliminates the problem as the plan is not triggered anymore. Test case: SET optimizer_prune_level=0; SET optimizer_switch = 'materialization=off'; CREATE TABLE t1 ( c INT ) ENGINE=MyISAM; INSERT INTO t1 VALUES   (0),(1),(2),(3),(4),(5),   (6),(7),(8),(9),(10),(11),(12); CREATE TABLE t2 ( a INT, b INT, KEY(a)) ENGINE=MyISAM; INSERT INTO t2 VALUES (3,20),(2,21),(3,22); SELECT * FROM t1 AS alias1, t1 AS alias2 WHERE ( alias1.c, alias2.c )   IN (       SELECT alias3.a, alias3.a         FROM t2 AS alias3, t2 alias4         WHERE alias3.b = alias4.b      ); # End of test case # Notes on result: # The correct result is 2 rows # a a # 2 2 # 3 3 # The test produces 3 rows # a a # 2 2 # 3 3 # 3 3 bzr version-info revision-id: igor@askmonty.org-20111206214218-lev0vyidyt8h3431 date: 2011-12-06 13:42:18 -0800 build-date: 2011-12-08 00:34:55 +0300 revno: 3334 branch-nick: maria-5.3 Also reproducible on revno 3315 The following query SELECT * FROM t1 AS alias1, t1 AS alias2 WHERE ( alias1.c, alias2.c )   IN (       SELECT alias3.a, alias3.a         FROM t2 AS alias3, t2 alias4         WHERE alias3.b = alias4.b      ); produces an extra row on test data if it's run with materialization=OFF and otherwise default optimizer_switch values, while it returns the correct result set with materialization=ON (or with semijoin=OFF, or with loosescan=OFF). EXPLAIN with materialization=OFF (wrong result): 1 PRIMARY alias2 ALL NULL NULL NULL NULL 13 100.00 1 PRIMARY alias3 ALL a NULL NULL NULL 3 66.67 Using where; LooseScan 1 PRIMARY alias4 ALL NULL NULL NULL NULL 3 100.00 Using where; FirstMatch(alias3) 1 PRIMARY alias1 ALL NULL NULL NULL NULL 13 100.00 Using where; Using join buffer (flat, BNL join) select `test`.`alias1`.`c` AS `c`,`test`.`alias2`.`c` AS `c` from `test`.`t1` `alias1` semi join (`test`.`t2` `alias3` join `test`.`t2` `alias4`) join `test`.`t1` `alias2` where ((`test`.`alias3`.`a` = `test`.`alias2`.`c`) and (`test`.`alias1`.`c` = `test`.`alias2`.`c`) and (`test`.`alias4`.`b` = `test`.`alias3`.`b`)) EXPLAIN with materialization=ON (correct result): 1 PRIMARY alias1 ALL NULL NULL NULL NULL 13 100.00 1 PRIMARY alias2 ALL NULL NULL NULL NULL 13 100.00 Using where; Using join buffer (flat, BNL join) 1 PRIMARY <subquery2> eq_ref distinct_key distinct_key 10 func,func 1 100.00 2 MATERIALIZED alias3 ALL a NULL NULL NULL 3 100.00 2 MATERIALIZED alias4 ALL NULL NULL NULL NULL 3 100.00 Using where select `test`.`alias1`.`c` AS `c`,`test`.`alias2`.`c` AS `c` from `test`.`t1` `alias1` semi join (`test`.`t2` `alias3` join `test`.`t2` `alias4`) join `test`.`t1` `alias2` where ((`test`.`alias2`.`c` = `test`.`alias1`.`c`) and (`test`.`alias4`.`b` = `test`.`alias3`.`b`)) Minimal optimizer_switch: materialization=off Full optimizer_switch: index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_merge_sort_intersection=off,index_condition_pushdown=on,derived_merge=on,derived_with_keys=on,firstmatch=on,loosescan=on,materialization=off,in_to_exists=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on,subquery_cache=on,mrr=off,mrr_cost_based=off,mrr_sort_keys=off,outer_join_with_cache=off,semijoin_with_cache=off,join_cache_incremental=on,join_cache_hashed=on,join_cache_bka=on,optimize_join_buffer_size=off,table_elimination=on Reproducible with MyISAM and InnoDB, but not Aria. Further reducing the test data eliminates the problem as the plan is not triggered anymore. Test case: SET optimizer_prune_level=0; SET optimizer_switch = 'materialization=off'; CREATE TABLE t1 ( c INT ) ENGINE=MyISAM; INSERT INTO t1 VALUES   (0),(1),(2),(3),(4),(5),   (6),(7),(8),(9),(10),(11),(12); CREATE TABLE t2 ( a INT, b INT, KEY(a)) ENGINE=MyISAM; INSERT INTO t2 VALUES (3,20),(2,21),(3,22); SELECT * FROM t1 AS alias1, t1 AS alias2 WHERE ( alias1.c, alias2.c )   IN (       SELECT alias3.a, alias3.a         FROM t2 AS alias3, t2 alias4         WHERE alias3.b = alias4.b      ); # End of test case # Notes on result: # The correct result is 2 rows # c c # 2 2 # 3 3 # The test produces 3 rows # c c # 2 2 # 3 3 # 3 3 bzr version-info revision-id: igor@askmonty.org-20111206214218-lev0vyidyt8h3431 date: 2011-12-06 13:42:18 -0800 build-date: 2011-12-08 00:34:55 +0300 revno: 3334 branch-nick: maria-5.3 Also reproducible on revno 3315
2011-12-07 22:51:14 Sergey Petrunia maria: status New Confirmed
2011-12-07 22:51:17 Sergey Petrunia maria: importance Undecided High
2011-12-15 23:46:41 Sergey Petrunia maria: status Confirmed Fix Committed
2012-03-18 00:00:03 Elena Stepanova maria: status Fix Committed Fix Released