Details
-
Bug
-
Status: Confirmed (View Workflow)
-
Major
-
Resolution: Unresolved
-
11.4.2, 10.11, 11.0(EOL), 11.2(EOL), 11.4, 11.5(EOL)
-
None
-
None
-
ubuntu 20.04
Description
In the following example, sql1 differs from sql2 in whether an index is created. This theoretically does not change the content of the result. There is a difference on the character '֣' between sql1 and sql3, which results in different results. I think there are some bugs in the indexing and handling of the '֣' character
--sql1
|
DROP TABLE IF EXISTS t0; |
CREATE TABLE t0(c0 TEXT); |
INSERT INTO t0(c0) VALUES(0); |
CREATE INDEX i0 ON t0(c0(1)); |
SELECT * FROM t0 WHERE t0.c0 NOT BETWEEN '֣k' AND NULL; |
--Empty set (0.00 sec)
|
|
|
--sql2
|
DROP TABLE IF EXISTS t0; |
CREATE TABLE t0(c0 TEXT); |
INSERT INTO t0(c0) VALUES(0); |
SELECT * FROM t0 WHERE t0.c0 NOT BETWEEN '֣k' AND NULL; |
+------+ |
| c0 |
|
+------+ |
| 0 |
|
+------+ |
1 row in set (0.00 sec) |
|
|
--sql3
|
DROP TABLE IF EXISTS t0; |
CREATE TABLE t0(c0 TEXT); |
INSERT INTO t0(c0) VALUES(0); |
CREATE INDEX i0 ON t0(c0(1)); |
SELECT * FROM t0 WHERE t0.c0 NOT BETWEEN 'k' AND NULL; |
+------+ |
| c0 |
|
+------+ |
| 0 |
|
+------+ |
1 row in set (0.00 sec) |
|