[teiid-issues] [JBoss JIRA] (TEIID-3573) Infinispan-dsl-cache translator: Operator <> incorrectly handles NULL values

[ https://issues.jboss.org/browse/TEIID-3573?page=com.atlassian.jira.plugin... ] Van Halbert edited comment on TEIID-3573 at 7/28/15 9:13 AM: ------------------------------------------------------------- My thinking was wrong on this, the result is correct, no rows should have been returned. Because, intNum <> 23 removed the rows where intNum = <null>, therefore adding "and intNum is Null" had no intersecting rows to choose from. Based on these tests and to get the right results, both SupportNot and SupportNull need to be set to false. As for performing a full scan for a negated predicate, it appears JDG is not doing a full scan on an indexed attribute. Based on the timing using squirrel, full scan: Total: 3.722, SQL query: 3.719, scan on index: Total: 0.133, SQL query: 0.129 Test query: select intKey,intNum, stringNum, floatnum from smallA where intNum <> 23 And this is only with 49 rows being returned. was (Author: van.halbert): My thinking was wrong on this, the result is correct, no rows should have been returned. Because, intNum <> 23 removed the rows where intNum = <null>, therefore adding "and intNum is Null" had no intersecting rows to choose from. Based on these tests and to get the right results, both SupportNot and SupportNull need to be set to false. As for performing a full scan for a negated predicate, it appears JDG is not doing a full scan on an indexed attribute. Based on the timing using squirrel, full scan: Total: 3.722, SQL query: 3.719, scan on index: Total: 0.133, SQL query: 0.129 And this is only with 49 rows being returned. -- This message was sent by Atlassian JIRA (v6.3.15#6346)