[MDEV-13411] main.gis fails in buildbot on P8 with wrong result Created: 2017-07-30  Updated: 2022-10-08  Resolved: 2022-10-08

Status: Closed
Project: MariaDB Server
Component/s: GIS, Tests
Affects Version/s: 10.0
Fix Version/s: N/A

Type: Bug Priority: Major
Reporter: Elena Stepanova Assignee: Alexey Botchkov
Resolution: Cannot Reproduce Votes: 0
Labels: None


 Description   

http://buildbot.askmonty.org/buildbot/builders/p8-rhel6-bintar/builds/990/steps/test/logs/stdio

main.gis                                 w3 [ fail ]
        Test ended at 2017-07-25 08:28:36
 
CURRENT_TEST: main.gis
--- /home/buildbot/maria-slave/p8-rhel6-bintar/build/mysql-test/r/gis.result	2017-07-25 06:55:57.870020147 -0400
+++ /home/buildbot/maria-slave/p8-rhel6-bintar/build/mysql-test/r/gis.reject	2017-07-25 08:28:35.740020148 -0400
@@ -684,11 +684,11 @@
 select object_id, geometrytype(geo), ISSIMPLE(GEO), ASTEXT(centroid(geo)) from
 t1 where object_id=85998;
 object_id	geometrytype(geo)	ISSIMPLE(GEO)	ASTEXT(centroid(geo))
-85998	MULTIPOLYGON	1	POINT(115.2970604672862 -36.23335610879993)
+85998	MULTIPOLYGON	1	POINT(115.29706047613604 -36.23335611157958)
 select object_id, geometrytype(geo), ISSIMPLE(GEO), ASTEXT(centroid(geo)) from
 t1 where object_id=85984;
 object_id	geometrytype(geo)	ISSIMPLE(GEO)	ASTEXT(centroid(geo))
-85984	MULTIPOLYGON	1	POINT(-114.86854472054372 36.34725218253213)
+85984	MULTIPOLYGON	1	POINT(-114.86854470090232 36.34725217627852)
 drop table t1;
 create table t1 (fl geometry not null);
 insert into t1 values (1);
 
mysqltest: Result length mismatch



 Comments   
Comment by Daniel Black [ 2017-07-31 ]

Note based on discussion here https://github.com/mysql/mysql-server/pull/158, that MySQL with https://github.com/mysql/mysql-server/commit/0a417e84 (using CFLAGS ffp-contract=off instead of no-expensive-optimizations), the actual rounding on almost all tests can be eliminated (https://github.com/mysql/mysql-server/pull/162). Here x86_64 showed a different result from P8 and ARM64 (which where the same) in only 8 test cases.

With MariaDB having rdiff/combinations support, extending this to architectures could eliminate all differences (and the occasional result due to page size etc).

Comment by Elena Stepanova [ 2022-10-08 ]

No recorded occurrences since 2018.

Generated at Thu Feb 08 08:05:23 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.