Slim Bouguerra created SAMZA-2554:
-------------------------------------
Summary: SQL Join Filter Handling is Broken when Table is remote
and the condition has more than one expression
Key: SAMZA-2554
URL: https://issues.apache.org/jira/browse/SAMZA-2554
Project: Samza
Issue Type: Bug
Components: sql
Reporter: Slim Bouguerra
The current implementation of remote table join can not handle more than one
condition.
Currently this will return wrong results silently by ignoring the filter
expression.
As a first step we need to fail such query to start with and then add the
support of 2 or more filter expression as part of the join condition.
Here is the test case
org.apache.samza.test.samzasql.TestSamzaSqlRemoteTable#testSourceEndToEndWithKeyWithNullForeignKeysRightOuterJoin
{code:java}
String sql =
"Insert into testavro.enrichedPageViewTopic "
+ "select pv.pageKey as __key__, pv.pageKey as pageKey,
coalesce(null, 'N/A') as companyName,"
+ " p.name as profileName, p.address as profileAddress "
+ " from "
+ " testavro.PAGEVIEW as pv left outer join
testRemoteStore.Profile.`$table` as p"
+ " on p.__key__ = pv.profileId and p.__key__ = pv.profileId + 1
and p.__key__ = pv.profileId - 1 ";
String sql =
"Insert into testavro.enrichedPageViewTopic "
+ "select pv.pageKey as __key__, pv.pageKey as pageKey,
coalesce(null, 'N/A') as companyName,"
+ " p.name as profileName, p.address as profileAddress "
+ " from "
+ " testavro.PAGEVIEW as pv left outer join
testRemoteStore.Profile.`$table` as p"
+ " on p.__key__ = pv.profileId ";
{code}
Both queries will return the same result and that is wrong
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
