Details
-
Bug
-
Status: Open (View Workflow)
-
Minor
-
Resolution: Unresolved
-
11.8
-
None
-
None
Description
I saw result mismatches somewhat similar to MDEV-36890 on a Launchpad s390x build in https://launchpadlibrarian.net/835922834/buildlog_ubuntu-resolute-s390x.mariadb_1%3A11.8.5-2~bpo26.04.1~1765057349.29238fe8526+debian.latest_BUILDING.txt.gz
main.vector_innodb w4 [ fail ]
|
Test ended at 2025-12-06 22:49:02
|
|
|
CURRENT_TEST: main.vector_innodb
|
--- /<<PKGBUILDDIR>>/mysql-test/main/vector_innodb.result 2025-11-13 08:45:29.000000000 +0000
|
+++ /<<PKGBUILDDIR>>/mysql-test/main/vector_innodb.reject 2025-12-06 22:49:01.683738555 +0000
|
@@ -49,10 +49,10 @@
|
select id,vec_distance_euclidean(v, x'754b5f3ea2312b3fc169f43e4604883e1d20173e8dd7443f421b703fb11e0d3e') d from t1 order by d limit 5;
|
id d
|
2 0.94269
|
-33 0.94775
|
30 1.11140
|
10 1.11863
|
8 1.14057
|
+5 1.16305
|
create table t2 (id int auto_increment primary key, v vector(6) not null, vector index (v)) engine=innodb;
|
insert t2 (v) values
|
(x'45cf153f830a313f7a0a113fb1ff533f47a1533fcf9e6e3f'),
|
|
|
Result length mismatch
|
It was sporadic, as it passed in the same test run on retry:
main.vector_innodb w4 [ retry-pass ] 369
|
main.vector_innodb w4 [ retry-pass ] 362
|