[jira] [Comment Edited] (CALCITE-4925) AggregateReduceFunctionsRule should accept arbitrary predicates

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

    [ https://issues.apache.org/jira/browse/CALCITE-4925?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17455923#comment-17455923 ] 

Julian Hyde edited comment on CALCITE-4925 at 12/8/21, 6:44 PM:
----------------------------------------------------------------

This is a useful change, nicely implemented. Two minor changes, then I'll merge:
 * Add a {{<p>}} paragraph markers the javadoc. Javadoc is not smart enough to recognize a blank line as a paragraph break.
 * Could you make {{extraCondition}} mandatory? It could be the true predicate by default. Then you would not need {{{}actualExtraCondition{}}}.


was (Author: julianhyde):
This is a useful change, nicely implemented. Two minor changes:
 * Add a {{<p>}} paragraph markers the javadoc. Javadoc is not smart enough to recognize a blank line as a paragraph break.
 * Could  

> AggregateReduceFunctionsRule should accept arbitrary predicates
> ---------------------------------------------------------------
>
>                 Key: CALCITE-4925
>                 URL: https://issues.apache.org/jira/browse/CALCITE-4925
>             Project: Calcite
>          Issue Type: Improvement
>            Reporter: Will Noble
>            Assignee: Will Noble
>            Priority: Minor
>              Labels: pull-request-available
>             Fix For: 1.29.0
>
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> {{AggregateReduceFunctionsRule}} allows you to specify which functions to reduce, but only allows you to specify by function type. For instance, you can either reduce all {{SUM}} functions, or no {{SUM}} functions, but not a subset of {{SUM}} functions (say, only the {{SUM}} functions with distinct keys). Allowing the user to specify an arbitrary predicate would allow us to selectively apply the rule to functions without distinct keys, which matters in Looker due to the fact that distinct keys cause an aggregate to be impossible to roll up, and when that's the case, we'd like to avoid reducing the functions for the purpose of aggregate awareness.



--
This message was sent by Atlassian Jira
(v8.20.1#820001)