[jira] [Commented] (CALCITE-1844) Statistic getRowCount() should return Long

["Julian Hyde (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CALCITE-1844?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16050751#comment-16050751 ] 

Julian Hyde commented on CALCITE-1844:
--------------------------------------

I made it a Double because it is an estimate (in statistics it is an expected value, i.e. an average). If you have two sources of estimates and one says 0 and the other says 1, then the best combined estimate is 0.5.

Also, using Double means we don't have to worry about overflowing 2^63. Estimates that large are not as uncommon as you think.

> Statistic getRowCount() should return Long 
> -------------------------------------------
>
>                 Key: CALCITE-1844
>                 URL: https://issues.apache.org/jira/browse/CALCITE-1844
>             Project: Calcite
>          Issue Type: Bug
>          Components: core
>            Reporter: darion yaphet
>            Assignee: Julian Hyde
>
> Currently Statistic.getRowCount() return a double type data as the approximate number of rows in the table . It seems should be a Long type . And I found the row count is  estimate value , use long type maybe better ?



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)