Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-6683

A parameter and a string literal with the same values are not recognized as equal by the optimizer

    XMLWordPrintable

Details

    • Bug
    • Status: Closed (View Workflow)
    • Major
    • Resolution: Fixed
    • 10.0.13
    • 10.0.14
    • None
    • None

    Description

      This bug is related to MDEV-6679, but for the cases when
      the column character set is different from the connection character set.

      This script:

      SET NAMES utf8, collation_connection=utf8_swedish_ci;
      CREATE TABLE t1 (a VARCHAR(10) CHARACTER SET latin1, b INT NOT NULL DEFAULT 0, key(a));
      INSERT INTO t1 VALUES ('a'),('b'),('c'),('d'),('¢');
      SET @arg='¢';
      PREPARE stmt FROM "EXPLAIN SELECT * FROM t1 WHERE a BETWEEN _utf8'¢' AND ?";
      EXECUTE stmt USING @arg;
      PREPARE stmt FROM "EXPLAIN SELECT * FROM t1 WHERE a between ? AND _utf8'¢'";
      EXECUTE stmt USING @arg;
      DEALLOCATE PREPARE stmt;

      returns

      +------+-------------+-------+-------+---------------+------+---------+------+------+--------------------------+
      | id   | select_type | table | type  | possible_keys | key  | key_len | ref  | rows | Extra                    |
      +------+-------------+-------+-------+---------------+------+---------+------+------+--------------------------+
      |    1 | SIMPLE      | t1    | range | a             | a    | 13      | NULL |    8 | Using where; Using index |
      +------+-------------+-------+-------+---------------+------+---------+------+------+--------------------------+

      for both EXPLAIN queries. Notice RANGE access type, which is wrong.

      If I change the queries to:

      EXPLAIN SELECT * FROM t1 WHERE a BETWEEN _utf8'¢' AND '¢';
      EXPLAIN SELECT * FROM t1 WHERE a between '¢' AND _utf8'¢';

      it returns:

      +------+-------------+-------+------+---------------+------+---------+-------+------+--------------------------+
      | id   | select_type | table | type | possible_keys | key  | key_len | ref   | rows | Extra                    |
      +------+-------------+-------+------+---------------+------+---------+-------+------+--------------------------+
      |    1 | SIMPLE      | t1    | ref  | a             | a    | 13      | const |    8 | Using where; Using index |
      +------+-------------+-------+------+---------------+------+---------+-------+------+--------------------------+

      Notice, REF access type, which means the two constants were
      recognized as the same and BETWEEN was changed to equality.

      The prepared statements should be fixed to use REF access,
      like the non-prepared do.

      Note, there is a subtle difference with MDEV-6679:

      • in MDEV-6679, the character set of the field is the same
        with the character set of the connection (utf8). The string literal
        with the introdcuder and the parameter values also have character
        set utf8, but they have different collations:
        utf8_general_ci vs utf8_swedish_ci respectively.
        So RANGE access is probably acceptable.
      • in this report, the field character set and the connection character set are different,
        so both literal with the introducer with the collation utf8_general_ci
        and the PS parameter with the collation utf8_swedish_ci are converted
        to latin1 with its default collation latin1_swedish_ci and thus loose the
        original collation difference and become equal. Thus REF access
        looks more suitable.

      Attachments

        Activity

          People

            bar Alexander Barkov
            bar Alexander Barkov
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Git Integration

                Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.