Bug #109351 Value change caused by DISTINCT and date functions
Submitted: 13 Dec 2022 4:46 Modified: 13 Dec 2022 6:34
Reporter: ZongYin Hao Email Updates:
Status: Verified Impact on me:
None 
Category:MySQL Server: Optimizer Severity:S2 (Serious)
Version:8.0.28, 8.0.31 OS:Any
Assigned to: CPU Architecture:Any
Tags: date functions, distinct, regression

[13 Dec 2022 4:46] ZongYin Hao
Description:
In theory, the result of sql2(DISTINCT) ⊆ the result of sql1:

SELECT 1%`f1` FROM (SELECT DAYNAME('2010-07-20') AS `f1` FROM t WHERE DATE_ADD(TIMEDIFF('2000-08-04 02:09:39', '2002-08-11 22:00:35'), INTERVAL 1 YEAR) NOT IN (1)) AS `t2`; -- sql1
SELECT 1%`f1` FROM (SELECT DISTINCT DAYNAME('2010-07-20') AS `f1` FROM t WHERE DATE_ADD(TIMEDIFF('2000-08-04 02:09:39', '2002-08-11 22:00:35'), INTERVAL 1 YEAR) NOT IN (1)) AS `t2`; -- sql2

However, the value 0 changed to NULL after adding DISTINCT, seems like a logical bug:

mysql> select version();
+-----------+
| version() |
+-----------+
| 8.0.31    |
+-----------+
1 row in set (0.00 sec)

mysql> SELECT 1%`f1` FROM (SELECT DAYNAME('2010-07-20') AS `f1` FROM t WHERE DATE_ADD(TIMEDIFF('2000-08-04 02:09:39', '2002-08-11 22:00:35'), INTERVAL 1 YEAR) NOT IN (1)) AS `t2`;
+--------+
| 1%`f1` |
+--------+
|      0 |
+--------+
1 row in set, 1 warning (0.00 sec)

mysql> SELECT 1%`f1` FROM (SELECT DISTINCT DAYNAME('2010-07-20') AS `f1` FROM t WHERE DATE_ADD(TIMEDIFF('2000-08-04 02:09:39', '2002-08-11 22:00:35'), INTERVAL 1 YEAR) NOT IN (1)) AS `t2`;
+--------+
| 1%`f1` |
+--------+
|   NULL |
+--------+
1 row in set, 3 warnings (0.00 sec)

How to repeat:
drop table if exists t;
CREATE TABLE t (c1 int);
INSERT INTO t VALUES (1);

SELECT 1%`f1` FROM (SELECT DAYNAME('2010-07-20') AS `f1` FROM t WHERE DATE_ADD(TIMEDIFF('2000-08-04 02:09:39', '2002-08-11 22:00:35'), INTERVAL 1 YEAR) NOT IN (1)) AS `t2`;
SELECT 1%`f1` FROM (SELECT DISTINCT DAYNAME('2010-07-20') AS `f1` FROM t WHERE DATE_ADD(TIMEDIFF('2000-08-04 02:09:39', '2002-08-11 22:00:35'), INTERVAL 1 YEAR) NOT IN (1)) AS `t2`;

Suggested fix:
The following list may be helpful for your debugging:
1. The bug cannot be reproduced after deleting any date function;
2. We look for the first version of the bug in all docker images (with format x.x.x) of https://hub.docker.com/_/mysql/tags 
We found that the bug first occurred in mysql:8.0.28, it cannot be reproduced in mysql:8.0.27:

mysql> select version();
+-----------+
| version() |
+-----------+
| 8.0.27    |
+-----------+
1 row in set (0.00 sec)

mysql> SELECT 1%`f1` FROM (SELECT DAYNAME('2010-07-20') AS `f1` FROM t WHERE DATE_ADD(TIMEDIFF('2000-08-04 02:09:39', '2002-08-11 22:00:35'), INTERVAL 1 YEAR) NOT IN (1)) AS `t2`;
Empty set, 2 warnings (0.00 sec)

mysql> SELECT 1%`f1` FROM (SELECT DISTINCT DAYNAME('2010-07-20') AS `f1` FROM t WHERE DATE_ADD(TIMEDIFF('2000-08-04 02:09:39', '2002-08-11 22:00:35'), INTERVAL 1 YEAR) NOT IN (1)) AS `t2`;
Empty set, 2 warnings (0.00 sec)
[13 Dec 2022 6:34] MySQL Verification Team
Hello ZongYin Hao,

Thank you for the report and test case.

regards,
Umesh