[
https://issues.apache.org/jira/browse/HIVE-6858?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13962753#comment-13962753
]
Jason Dere commented on HIVE-6858:
----------------------------------
Would you be able to fix groupby3_map_skew.q as well, which looks like it also
has a similar issue? For that one maybe you could replace:
SELECT dest1.* FROM dest1;
with:
SELECT c1, c2, c3, c4, c5, c6, c7, ROUND(c8, 5), ROUND(c9, 5) FROM dest1;
And hopefully the values generated do not show differences between the jdk6/7
formatting.
> Unit tests decimal_udf.q, vectorization_div0.q fail with jdk-7.
> ---------------------------------------------------------------
>
> Key: HIVE-6858
> URL: https://issues.apache.org/jira/browse/HIVE-6858
> Project: Hive
> Issue Type: Bug
> Reporter: Jitendra Nath Pandey
> Assignee: Jitendra Nath Pandey
> Attachments: HIVE-6858.1.patch
>
>
> Unit tests decimal_udf.q, vectorization_div0.q fail with jdk-7.
> {noformat}
> < -250.0 6583411.236 1.0 6583411.236 -0.004 -0.0048
> ---
> > -250.0 6583411.236 1.0 6583411.236 -0.0040 -0.0048
> {noformat}
> Following code reproduces this behavior when run in jdk-7 vs jdk-6. Jdk-7
> produces -0.004 while, jdk-6 produces -0.0040.
> {code}
> public class Main {
> public static void main(String[] a) throws Exception {
> double val = 0.004;
> System.out.println("Value = "+val);
> }
> }
> {code}
> This happens to be a bug in jdk6, that has been fixed in jdk7.
> http://bugs.java.com/bugdatabase/view_bug.do?bug_id=4511638
--
This message was sent by Atlassian JIRA
(v6.2#6252)
