1
0
mirror of https://github.com/MariaDB/server.git synced 2025-07-29 05:21:33 +03:00

Fix for BUG#8532.

The problem was in that the code that analyses the applicability of the
QUICK_GROUP_MIN_MAX access method for DISTINC queries assumed that there
are no duplicate column references in the DISTINCT clause, and it added
non-exiting key parts for the duplicate column references. 

The solution adds a test to check whether the select list already contained
a field with the same name. If such field was already present, then it was
already decided to use its key part for index access. In this such case we
must skip the duplicate field instead of counting it as a new field.
This commit is contained in:
timour@mysql.com
2005-02-28 16:20:02 +02:00
parent 21ca91aac6
commit 03e7620d3f
3 changed files with 53 additions and 2 deletions

View File

@ -6878,6 +6878,7 @@ get_best_group_min_max(PARAM *param, SEL_TREE *tree)
SEL_ARG *cur_index_tree= NULL;
ha_rows cur_quick_prefix_records= 0;
uint cur_param_idx;
key_map cur_used_key_parts;
for (uint cur_index= 0 ; cur_index_info != cur_index_info_end ;
cur_index_info++, cur_index++)
@ -6925,17 +6926,25 @@ get_best_group_min_max(PARAM *param, SEL_TREE *tree)
else if (join->select_distinct)
{
select_items_it.rewind();
cur_used_key_parts.clear_all();
while ((item= select_items_it++))
{
item_field= (Item_field*) item; /* (SA5) already checked above. */
/* Find the order of the key part in the index. */
key_part_nr= get_field_keypart(cur_index_info, item_field->field);
/*
Check if this attribute was already present in the select list.
If it was present, then its corresponding key part was alredy used.
*/
if (cur_used_key_parts.is_set(key_part_nr))
continue;
if (key_part_nr < 1 || key_part_nr > join->fields_list.elements)
goto next_index;
cur_part= cur_index_info->key_part + key_part_nr - 1;
cur_group_prefix_len+= cur_part->store_length;
cur_used_key_parts.set_bit(key_part_nr);
++cur_group_key_parts;
}
cur_group_key_parts= join->fields_list.elements;
}
else
DBUG_ASSERT(FALSE);