[
https://issues.apache.org/jira/browse/FLINK-33950?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17801120#comment-17801120
]
Jacky Lau commented on FLINK-33950:
-----------------------------------
hi [~martijnvisser] [~twalthr] the reason of the ci of my pr why not passed
is when we migrate to new stack agg.
the max function accumulate type is RowData. and it will cause the batch plan
from hashAgg to SortAgg because the bellowing code in BatchPhysicalHashAggRule.
and it may cause a decrease in performance. So we need to reach a consensus on
whether changes are necessary， could i migrate to
new stach agg now.
{code:java}
// BatchPhysicalHashAggRule
// HashAgg cannot process aggregate whose agg buffer is not fix length
isAggBufferFixedLength(agg) && {code}
> Update max aggregate functions to new type system
> -------------------------------------------------
>
> Key: FLINK-33950
> URL: https://issues.apache.org/jira/browse/FLINK-33950
> Project: Flink
> Issue Type: Sub-task
> Components: Table SQL / Planner
> Affects Versions: 1.19.0
> Reporter: Jacky Lau
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.19.0
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
