1
0
mirror of https://github.com/MariaDB/server.git synced 2025-07-30 16:24:05 +03:00

BUG#15872: Don't run the range analyzer on "t1.keypart NOT IN (const1, ..., )", as that consumes

too much memory. Instead, either create the equvalent SEL_TREE manually, or create only two ranges that
strictly include the area to scan
(Note: just to re-iterate: increasing NOT_IN_IGNORE_THRESHOLD will make optimization run slower for big 
IN-lists, but the server will not run out of memory. O(N^2) memory use has been eliminated)


mysql-test/r/func_in.result:
  Testcase for BUG#15872
mysql-test/t/func_in.test:
  Testcase for BUG#15872
sql/item.cc:
  BUG#15872: Added Item_decimal::set_decimal_value()
sql/item.h:
  UG#15872: Added Item_decimal::set_decimal_value()
sql/item_cmpfunc.h:
  BUG#15872: Added in_vector::create_item(), in_vector::value_to_item() and their implementations in concrete
  classes.
sql/opt_range.cc:
  BUG#15872: Don't run the range analyzer on "t1.keypart NOT IN (const1, ..., )", as that 
  consumes too much memory. Instead, either 
   A) create the equivalent SEL_TREE manually, making use of the fact that item_not_in->array
      has an ordered IN-list, or
   B) create only two ranges: (-inf|NULL) < X < min_value_from_in_list,  max_value_from_in_list < X
  (Choose #B if the IN-list has > 10K elements)
This commit is contained in:
unknown
2006-04-25 23:33:31 +04:00
parent 2a138695bf
commit 15e00f1f3d
6 changed files with 364 additions and 14 deletions

View File

@ -1,4 +1,4 @@
drop table if exists t1;
drop table if exists t1, t2;
select 1 in (1,2,3);
1 in (1,2,3)
1
@ -225,3 +225,104 @@ a
46
DROP VIEW v1;
DROP TABLE t1;
create table t1 (a int);
insert into t1 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9);
create table t2 (a int, filler char(200), key(a));
insert into t2 select C.a*2, 'no' from t1 A, t1 B, t1 C;
insert into t2 select C.a*2+1, 'yes' from t1 C;
explain
select * from t2 where a NOT IN (0, 2,4,6,8,10,12,14,16,18);
id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t2 range a a 5 NULL 12 Using where
select * from t2 where a NOT IN (0, 2,4,6,8,10,12,14,16,18);
a filler
1 yes
3 yes
5 yes
7 yes
9 yes
11 yes
13 yes
15 yes
17 yes
19 yes
explain select * from t2 force index(a) where a NOT IN (2,2,2,2,2,2);
id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t2 range a a 5 NULL 912 Using where
explain select * from t2 force index(a) where a <> 2;
id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t2 range a a 5 NULL 912 Using where
drop table t2;
create table t2 (a datetime, filler char(200), key(a));
insert into t2 select '2006-04-25 10:00:00' + interval C.a minute,
'no' from t1 A, t1 B, t1 C where C.a % 2 = 0;
insert into t2 select '2006-04-25 10:00:00' + interval C.a*2+1 minute,
'yes' from t1 C;
explain
select * from t2 where a NOT IN (
'2006-04-25 10:00:00','2006-04-25 10:02:00','2006-04-25 10:04:00',
'2006-04-25 10:06:00', '2006-04-25 10:08:00');
id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t2 range a a 9 NULL 18 Using where
select * from t2 where a NOT IN (
'2006-04-25 10:00:00','2006-04-25 10:02:00','2006-04-25 10:04:00',
'2006-04-25 10:06:00', '2006-04-25 10:08:00');
a filler
2006-04-25 10:01:00 yes
2006-04-25 10:03:00 yes
2006-04-25 10:05:00 yes
2006-04-25 10:07:00 yes
2006-04-25 10:09:00 yes
2006-04-25 10:11:00 yes
2006-04-25 10:13:00 yes
2006-04-25 10:15:00 yes
2006-04-25 10:17:00 yes
2006-04-25 10:19:00 yes
drop table t2;
create table t2 (a varchar(10), filler char(200), key(a));
insert into t2 select 'foo', 'no' from t1 A, t1 B;
insert into t2 select 'barbar', 'no' from t1 A, t1 B;
insert into t2 select 'bazbazbaz', 'no' from t1 A, t1 B;
insert into t2 values ('fon', '1'), ('fop','1'), ('barbaq','1'),
('barbas','1'), ('bazbazbay', '1'),('zz','1');
explain select * from t2 where a not in('foo','barbar', 'bazbazbaz');
id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t2 range a a 13 NULL 7 Using where
drop table t2;
create table t2 (a decimal(10,5), filler char(200), key(a));
insert into t2 select 345.67890, 'no' from t1 A, t1 B;
insert into t2 select 43245.34, 'no' from t1 A, t1 B;
insert into t2 select 64224.56344, 'no' from t1 A, t1 B;
insert into t2 values (0, '1'), (22334.123,'1'), (33333,'1'),
(55555,'1'), (77777, '1');
explain
select * from t2 where a not in (345.67890, 43245.34, 64224.56344);
id select_type table type possible_keys key key_len ref rows Extra
1 SIMPLE t2 range a a 7 NULL 7 Using where
select * from t2 where a not in (345.67890, 43245.34, 64224.56344);
a filler
0.00000 1
22334.12300 1
33333.00000 1
55555.00000 1
77777.00000 1
drop table t2;
create table t2 (a int, key(a), b int);
insert into t2 values (1,1),(2,2);
set @cnt= 1;
set @str="update t2 set b=1 where a not in (";
select count(*) from (
select @str:=concat(@str, @cnt:=@cnt+1, ",")
from t1 A, t1 B, t1 C, t1 D) Z;
count(*)
10000
set @str:=concat(@str, "10000)");
select substr(@str, 1, 50);
substr(@str, 1, 50)
update t2 set b=1 where a not in (2,3,4,5,6,7,8,9,
prepare s from @str;
execute s;
deallocate prepare s;
set @str=NULL;
drop table t2;
drop table t1;

View File

@ -1,6 +1,6 @@
# Initialise
--disable_warnings
drop table if exists t1;
drop table if exists t1, t2;
--enable_warnings
#
# test of IN (NULL)
@ -128,3 +128,95 @@ SELECT * FROM v1;
DROP VIEW v1;
DROP TABLE t1;
# BUG#15872: Excessive memory consumption of range analysis of NOT IN
create table t1 (a int);
insert into t1 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9);
create table t2 (a int, filler char(200), key(a));
insert into t2 select C.a*2, 'no' from t1 A, t1 B, t1 C;
insert into t2 select C.a*2+1, 'yes' from t1 C;
explain
select * from t2 where a NOT IN (0, 2,4,6,8,10,12,14,16,18);
select * from t2 where a NOT IN (0, 2,4,6,8,10,12,14,16,18);
explain select * from t2 force index(a) where a NOT IN (2,2,2,2,2,2);
explain select * from t2 force index(a) where a <> 2;
drop table t2;
#
# Repeat the test for DATETIME
#
create table t2 (a datetime, filler char(200), key(a));
insert into t2 select '2006-04-25 10:00:00' + interval C.a minute,
'no' from t1 A, t1 B, t1 C where C.a % 2 = 0;
insert into t2 select '2006-04-25 10:00:00' + interval C.a*2+1 minute,
'yes' from t1 C;
explain
select * from t2 where a NOT IN (
'2006-04-25 10:00:00','2006-04-25 10:02:00','2006-04-25 10:04:00',
'2006-04-25 10:06:00', '2006-04-25 10:08:00');
select * from t2 where a NOT IN (
'2006-04-25 10:00:00','2006-04-25 10:02:00','2006-04-25 10:04:00',
'2006-04-25 10:06:00', '2006-04-25 10:08:00');
drop table t2;
#
# Repeat the test for CHAR(N)
#
create table t2 (a varchar(10), filler char(200), key(a));
insert into t2 select 'foo', 'no' from t1 A, t1 B;
insert into t2 select 'barbar', 'no' from t1 A, t1 B;
insert into t2 select 'bazbazbaz', 'no' from t1 A, t1 B;
insert into t2 values ('fon', '1'), ('fop','1'), ('barbaq','1'),
('barbas','1'), ('bazbazbay', '1'),('zz','1');
explain select * from t2 where a not in('foo','barbar', 'bazbazbaz');
drop table t2;
#
# Repeat for DECIMAL
#
create table t2 (a decimal(10,5), filler char(200), key(a));
insert into t2 select 345.67890, 'no' from t1 A, t1 B;
insert into t2 select 43245.34, 'no' from t1 A, t1 B;
insert into t2 select 64224.56344, 'no' from t1 A, t1 B;
insert into t2 values (0, '1'), (22334.123,'1'), (33333,'1'),
(55555,'1'), (77777, '1');
explain
select * from t2 where a not in (345.67890, 43245.34, 64224.56344);
select * from t2 where a not in (345.67890, 43245.34, 64224.56344);
drop table t2;
# Try a very big IN-list
create table t2 (a int, key(a), b int);
insert into t2 values (1,1),(2,2);
set @cnt= 1;
set @str="update t2 set b=1 where a not in (";
select count(*) from (
select @str:=concat(@str, @cnt:=@cnt+1, ",")
from t1 A, t1 B, t1 C, t1 D) Z;
set @str:=concat(@str, "10000)");
select substr(@str, 1, 50);
prepare s from @str;
execute s;
deallocate prepare s;
set @str=NULL;
drop table t2;
drop table t1;