[jira] [Created] (KYLIN-1689) bug when a column being dimension as well as in a sum metric

["hongbin ma (JIRA)" <ji...@apache.org>]

hongbin ma created KYLIN-1689:
---------------------------------

             Summary: bug when a column being dimension as well as in a sum metric
                 Key: KYLIN-1689
                 URL: https://issues.apache.org/jira/browse/KYLIN-1689
             Project: Kylin
          Issue Type: Bug
            Reporter: hongbin ma
            Assignee: hongbin ma


Hi all,
 
I recently built a cube named c1, use 2 columns as dimensions ,”rule_name”,” PARTNER_GAIN_PAY_PT_DOC_CNT”, also use ” sum(PARTNER_GAIN_PAY_PT_DOC_CNT)” as measure. C1 was built successfully.
 
So, I made a query sql to test, that is “select rule_name,PARTNER_GAIN_PAY_PT_DOC_CNT ,count(*),sum(PARTNER_GAIN_PAY_PT_DOC_CNT) from CUB_PARTNER_GAIN_PAY_PT_PRE0_AT0_S where rule_name='1号店3C产品' group by rule_name,PARTNER_GAIN_PAY_PT_DOC_CNT;”, but the result is not probably exactly.
RULE_NAME  PARTNER...  EXPR$2 EXPR$3
1号店3C产品    1860     30    1860
1号店3C产品    700      2     700
1号店3C产品    7410     38    7410
1号店3C产品    2940     60    2940
 
In my opinion,”count(*)” means the amount of records with the same rule_name and PARTNER_GAIN_PAY_PT_DOC_CNT, so I think sum(PARTNER…) equals
Count(*) * PARTNER_GAIN_PAY_PT_DOC_CNT, but the truth is not , I wonder if there is something wrong with my understanding?
 
       Insight snapshot as below:



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)