Commit d01ec5e7 authored by unknown's avatar unknown
Browse files

Bug #25551: inconsistent behaviour in grouping NULL, depending on index type 

 The optimizer takes away columns from GROUP BY/DISTINCT if they constitute
 all the parts of an unique index.
 However if some of the columns can contain NULLs this cannot be done 
(because an UNIQUE index can have multiple rows with NULL values).
 Fixed by not using UNIQUE indexes with nullable columns to remove
 grouping columns from GROUP BY/DISTINCT.


mysql-test/r/distinct.result:
  Bug #25551: inconsistent behaviour in grouping NULL, depending on index type
   - test case
mysql-test/t/distinct.test:
  Bug #25551: inconsistent behaviour in grouping NULL, depending on index type
   - test case
sql/sql_select.cc:
  Bug #25551: inconsistent behaviour in grouping NULL, depending on index type
   - UNIQUE NULL indices don't guarantee GROUP BY/DISTINCT.
parent c6d4b94d

mysql-test/r/distinct.result

+25 −1
Original line number Diff line number Diff line
@@ -530,7 +530,8 @@ id select_type table type possible_keys key key_len ref rows Extra 
EXPLAIN SELECT DISTINCT a,b FROM t1 GROUP BY a,b;

id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra

1	SIMPLE	t1	ALL	NULL	NULL	NULL	NULL	3	

CREATE TABLE t2(a INT, b INT, c INT, d INT, PRIMARY KEY (a,b));

CREATE TABLE t2(a INT, b INT NOT NULL, c INT NOT NULL, d INT, 

PRIMARY KEY (a,b));

INSERT INTO t2 VALUES (1,1,1,50), (1,2,3,40), (2,1,3,4);

EXPLAIN SELECT DISTINCT a FROM t2;

id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
@@ -644,3 +645,26 @@ SELECT COUNT(*) FROM 
COUNT(*)

2

DROP TABLE t1, t2;

CREATE TABLE t1 (a INT, UNIQUE (a));

INSERT INTO t1 VALUES (4),(null),(2),(1),(null),(3);

EXPLAIN SELECT DISTINCT a FROM t1;

id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra

1	SIMPLE	t1	index	NULL	a	5	NULL	6	Using index

SELECT DISTINCT a FROM t1;

a

NULL

1

2

3

4

EXPLAIN SELECT a FROM t1 GROUP BY a;

id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra

1	SIMPLE	t1	index	NULL	a	5	NULL	6	Using index

SELECT a FROM t1 GROUP BY a;

a

NULL

1

2

3

4

DROP TABLE t1;

mysql-test/t/distinct.test

+16 −1
Original line number Diff line number Diff line
@@ -364,7 +364,8 @@ EXPLAIN SELECT a FROM t1 GROUP BY a; 
EXPLAIN SELECT a,b FROM t1 GROUP BY a,b;

EXPLAIN SELECT DISTINCT a,b FROM t1 GROUP BY a,b;



CREATE TABLE t2(a INT, b INT, c INT, d INT, PRIMARY KEY (a,b));

CREATE TABLE t2(a INT, b INT NOT NULL, c INT NOT NULL, d INT, 

                PRIMARY KEY (a,b));

INSERT INTO t2 VALUES (1,1,1,50), (1,2,3,40), (2,1,3,4);

EXPLAIN SELECT DISTINCT a FROM t2;

EXPLAIN SELECT DISTINCT a,a FROM t2;
@@ -525,3 +526,17 @@ SELECT COUNT(*) FROM 
  (SELECT DISTINCT a FROM t2 WHERE a='oe' COLLATE latin1_german2_ci) dt;



DROP TABLE t1, t2;



#

# Bug #25551: inconsistent behaviour in grouping NULL, depending on index type

#

CREATE TABLE t1 (a INT, UNIQUE (a));

INSERT INTO t1 VALUES (4),(null),(2),(1),(null),(3);

EXPLAIN SELECT DISTINCT a FROM t1;

#result must have one row with NULL

SELECT DISTINCT a FROM t1;

EXPLAIN SELECT a FROM t1 GROUP BY a;

#result must have one row with NULL

SELECT a FROM t1 GROUP BY a;



DROP TABLE t1;

sql/sql_select.cc

+11 −8
Original line number Diff line number Diff line
@@ -848,10 +848,11 @@ JOIN::optimize() 
  }

  /*

     Check if we can optimize away GROUP BY/DISTINCT.

     We can do that if there are no aggregate functions and the

     We can do that if there are no aggregate functions, the

     fields in DISTINCT clause (if present) and/or columns in GROUP BY

     (if present) contain direct references to all key parts of

     an unique index (in whatever order).

     an unique index (in whatever order) and if the key parts of the

     unique index cannot contain NULLs.

     Note that the unique keys for DISTINCT and GROUP BY should not

     be the same (as long as they are unique).
@@ -11878,7 +11879,8 @@ test_if_subkey(ORDER *order, TABLE *table, uint ref, uint ref_key_parts, 
    of the table are referenced by a list : either the select list

    through find_field_in_item_list or GROUP BY list through

    find_field_in_order_list.

    If the above holds then we can safely remove the GROUP BY/DISTINCT,

    If the above holds and the key parts cannot contain NULLs then we 

    can safely remove the GROUP BY/DISTINCT,

    as no result set can be more distinct than an unique key.

 

  RETURN VALUE
@@ -11903,7 +11905,8 @@ list_contains_unique_index(TABLE *table, 
           key_part < key_part_end;

           key_part++)

      {

        if (!find_func(key_part->field, data))

        if (key_part->field->maybe_null() || 

            !find_func(key_part->field, data))

          break;

      }

      if (key_part == key_part_end)