[ https://issues.apache.org/jira/browse/TRAFODION-2972?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16379089#comment-16379089 ]
David Wayne Birdsall commented on TRAFODION-2972: ------------------------------------------------- Thinking about the query a little more: I will guess that ANY and SOME comparisons are meant to be alternative ways of formulating EXISTS predicates. If this is true, then we should expect deterministic behavior from the query. > Follow up on supposed non-deterministic query in core/TEST002 > ------------------------------------------------------------- > > Key: TRAFODION-2972 > URL: https://issues.apache.org/jira/browse/TRAFODION-2972 > Project: Apache Trafodion > Issue Type: Bug > Components: sql-cmp > Affects Versions: 2.3 > Reporter: David Wayne Birdsall > Priority: Major > > Regression test core/TEST002 has a query that is commented out: > ``` > -- This test disabled since it is non -deterministic. > -- It will be enabled after further investigation. > -- ?section Genesis_10_000222_6892_p1 > -- SELECT 1 FROM T002T3 T1 > -- GROUP BY T1.A > -- HAVING T1.A >ANY > -- ( SELECT 2 FROM T002T1 T2 > -- WHERE T2.C >SOME > -- ( SELECT AVG (T1.A) FROM T002T1 T3 ) > -- ); > ``` > From git blame, it appears this has been commented out since before Trafodion > was moved to github, so we don't have any available pedigree or background > info on why this was commented out. > We should investigate why this is non-deterministic, and if it is not > inherently so, fix any bug that is making it non-deterministic. If it is > inherently non-deterministic, then the test should be modified to take that > into account. > Just looking at the query without any deep investigation, I will guess that > it is in fact inherently non-deterministic, as ANY and SOME make no > guarantees as to which row they will return; they just return some row. -- This message was sent by Atlassian JIRA (v7.6.3#76005)