Details
-
Bug
-
Status: In Testing (View Workflow)
-
Critical
-
Resolution: Unresolved
-
N/A
-
None
Description
While it is probably a test-only problem, it must be fixed before the release because the test in this configuration runs and fails in the buildbot.
https://buildbot.mariadb.org/#/builders/639/builds/23223
|
bb-blob-monty 4d3e5d05080a4663ec33dea982ce4022c12a86ed |
$ perl ./mtr main.type_bit --mem --cursor-protocol
|
...
|
main.type_bit [ fail ]
|
Test ended at 2026-05-21 17:44:00
|
|
|
CURRENT_TEST: main.type_bit
|
--- /share8t/bld/testing/bb-blob-monty-debug/mysql-test/main/type_bit.result 2026-05-20 16:55:18.743542079 +0300
|
+++ /share8t/bld/testing/bb-blob-monty-debug/mysql-test/main/type_bit.reject 2026-05-21 17:44:00.844978771 +0300
|
@@ -648,13 +648,13 @@
|
INSERT INTO t1 (b) VALUES (1), (0);
|
SELECT DISTINCT b FROM t1;
|
Catalog Database Table Table_alias Column Column_alias Type Length Max length Is_null Flags Decimals Charsetnr
|
-def test t1 t1 b b 16 1 1 Y 36 0 63
|
+def test t1 t1 b b 16 1 1 Y 32 0 63
|
b
|
#
|
#
|
SELECT b FROM t1 GROUP BY b;
|
Catalog Database Table Table_alias Column Column_alias Type Length Max length Is_null Flags Decimals Charsetnr
|
-def test t1 t1 b b 16 1 1 Y 36 0 63
|
+def test t1 t1 b b 16 1 1 Y 32 0 63
|
b
|
#
|
#
|
|
|
Result content mismatch
|
Due to MDEV-38975 the query in the case below (which represents the fragment of the type_bit test) acquired the extra flag UNIQUE_KEY:
MariaDB [test]> CREATE TABLE t1 (b BIT); |
Query OK, 0 rows affected (0.031 sec) |
|
|
MariaDB [test]> INSERT INTO t1 (b) VALUES (1), (0); |
Query OK, 2 rows affected (0.013 sec) |
Records: 2 Duplicates: 0 Warnings: 0
|
|
|
MariaDB [test]> SELECT DISTINCT b FROM t1; |
Field 1: `b`
|
Org_field: `b`
|
Catalog: `def`
|
Database: `test` |
Table: `t1` |
Org_table: `t1`
|
Type: BIT |
Collation: binary (63) |
Length: 1
|
Max_length: 1
|
Decimals: 0
|
Flags: UNIQUE_KEY UNSIGNED
|
|
|
+------+ |
| b |
|
+------+ |
| |
|
| |
|
+------+ |
2 rows in set (0.001 sec) |
This change was reflected in main/type_bit.result.
However, apparently the flag doesn't appear when the same query is executed via cursor protocol, so the test case now fails with a result mismatch.
Probably the easiest way to avoid it is to disable cursor protocol for this part of the test, but first it should be confirmed that the behavior is expected.
Attachments
Issue Links
- is caused by
-
MDEV-38975 BLOBs in MEMORY (HEAP)
-
- In Testing
-