Victoria Markman created DRILL-3801:
---------------------------------------
Summary: Extend support for scalar subqueries
Key: DRILL-3801
URL: https://issues.apache.org/jira/browse/DRILL-3801
Project: Apache Drill
Issue Type: New Feature
Reporter: Victoria Markman
In current drill implementation we only consider subquery scalar when scalar
aggregate is present and as I discovered recently, we also consider subquery to
return scalar result if you have 1 column, and 1 row (limit 1)
However, if subquery does return one row during runtime, we currently don't
detect that and we throw an error early during planning time:
{code}
0: jdbc:drill:schema=dfs> select * from t2 where a2 = 0;
+-----+------+-------------+
| a2 | b2 | c2 |
+-----+------+-------------+
| 0 | zzz | 2014-12-31 |
+-----+------+-------------+
1 row selected (0.283 seconds)
0: jdbc:drill:schema=dfs> select * from t1 where a1 >= (select a2 from t2 where
a2 = 0);
Error: UNSUPPORTED_OPERATION ERROR: Non-scalar sub-query used in an expression
See Apache Drill JIRA: DRILL-1937
[Error Id: 9e26af25-b0c4-42f9-a839-b76f759c94ab on atsqa4-133.qa.lab:31010]
(state=,code=0)
{code}
Because of this limitation we can't run TPCDS query 44 (inner subquery has
just one group):
{code}
...
...
SELECT ss_item_sk item_sk,
Avg(ss_net_profit) rank_col
FROM store_sales ss1
WHERE ss_store_sk = 4
GROUP BY ss_item_sk
HAVING Avg(ss_net_profit) > 0.9 *
(SELECT Avg(ss_net_profit)
rank_col
FROM store_sales
WHERE ss_store_sk = 4
AND ss_cdemo_sk IS
NULL
GROUP BY ss_store_sk))V2)
{code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
