This bug related to MDEV-27072 but happend because of a very different reason than MDEV-27072.

I create a table:

DROP TABLE IF EXISTS t1;

CREATE TABLE t1 (d TIME);

INSERT INTO t1 VALUES ('120:00:00'), ('20:00:00'), ('-120:00:00'), ('-220:00:00');

Now I run this query:

SELECT * FROM t1 WHERE d >= ALL (SELECT * FROM t1);

+-----------+

| d         |

+-----------+

| 120:00:00 |

| 20:00:00  |

+-----------+

The result is wrong. It should return only one row the maxumum value, which is '120:00:00'.

The same problem happens with this query:

SELECT * FROM t1 WHERE d <= ALL (SELECT * FROM t1);

+------------+

| d          |

+------------+

| -120:00:00 |

| -220:00:00 |

+------------+

It should return only one row with the minumum value, which is '-220:00:00'.

The problem happens because select_max_min_finder_subselect::cmp_str() is executed at some point during these queries. It compares values as strings, therefore:

• '120:00:00' is erroneously compared as less than '20:00:00'
• '-220:00:00' is erroneously compared as greater than '-120:00:00'

The TIME data type needs its own implementation, e.g. select_max_min_finder_subselect::cmp_time().