Github user robbinspg commented on the issue:
https://github.com/apache/spark/pull/16841
@kevinyu98 Several of the new tests fail on Big Endian platforms. It
appears that rows are returned in a slightly different order but are still a
correct output from the query. For example in-joins query 4:
-- !query 4
SELECT Count(DISTINCT(t1a)),
t1b,
t3a,
t3b,
t3c
FROM t1 natural left JOIN t3
WHERE t1a IN
(
SELECT t2a
FROM t2
WHERE t1d = t2d)
AND t1b > t3b
GROUP BY t1a,
t1b,
t3a,
t3b,
t3c
ORDER BY t1a DESC
on Little Endian returns
1 10 val3b 8 NULL
1 10 val1b 8 16
1 10 val3a 6 12
1 8 val3a 6 12
1 8 val3a 6 12
wheras on big endian returns:
1 10 val3a 6 12
1 10 val3b 8 NULL
1 10 val1b 8 16
1 8 val3a 6 12
1 8 val3a 6 12
I believe GROUP BY does not define any ordering so both of these outputs
are valid for the query as the ORDER BY is only on t1a but obviously the big
endian output does not match your expected output so fails.
I'm trying to determine why the execution on big endian returns the rows in
a different order.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
