[
https://issues.apache.org/jira/browse/CALCITE-1368?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15605088#comment-15605088
]
Atri Sharma commented on CALCITE-1368:
--------------------------------------
[~julianhyde] What is the expected behaviour for this please? Can you please
give more context here?
> Infer predicates for aggregate functions
> ----------------------------------------
>
> Key: CALCITE-1368
> URL: https://issues.apache.org/jira/browse/CALCITE-1368
> Project: Calcite
> Issue Type: Bug
> Reporter: Julian Hyde
> Assignee: Julian Hyde
>
> Currently when inferring predicates for Aggregate, Calcite only pulls up
> existing predicates, and it can only do that for columns that are part of the
> group key. This feature would infer predicates based on the behavior of
> aggregate functions and, sometimes, predicates of the arguments to those
> functions.
> Examples:
> * {{count\(*)}} and {{count\(x)}} are always > 0 (if groups are non-empty,
> and x is not null)
> * {{count\(*)}} and {{count\(x)}} are always >= 0
> * {{sum\(x)}} is >= 0 (if x is >= 0; it's OK if x is sometimes null)
> * {{min\(x)}} and {{max\(x)}} have the same upper and lower bounds as x
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
