[
https://issues.apache.org/jira/browse/HIVE-30131?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Denys Kuzmenko resolved HIVE-30131.
-----------------------------------
Fix Version/s: 4.3.0
Resolution: Fixed
> StatsOptimizer cannot answer MIN/MAX/COUNT(col) from statistics for DECIMAL
> columns
> -----------------------------------------------------------------------------------
>
> Key: HIVE-30131
> URL: https://issues.apache.org/jira/browse/HIVE-30131
> Project: Hive
> Issue Type: Bug
> Reporter: Denys Kuzmenko
> Assignee: Denys Kuzmenko
> Priority: Major
> Labels: pull-request-available
> Fix For: 4.3.0
>
>
> - select min(c) from t with c decimal(p,s) and accurate column statistics
> still plans a full scan, while count(*) folds.
>
> - StatsOptimizer.getType returns Unsupported for DECIMAL (only
> integral/double/float/binary/boolean/string/date are handled), so the MIN/MAX
> and COUNT(col) branches decline with "Unsupported
> type: decimal(p,s) encountered in metadata optimizer".
>
>
> - The metastore model already carries the bounds (DecimalColumnStatsData
> low/high/numNulls), and both HMS and the Iceberg handler (HIVE-29834) serve
> them — only the optimizer's type switch
> lacks the case.
>
>
>
> - Fix: add StatType.Decimal, getDecimalStats().getNumNulls(), and a MIN/MAX
> case converting thrift Decimal → HiveDecimal the way StatsUtils already does
> (:924-931). Tests: a decimal column in
> stats_part.q and iceberg_part_colstats.q, with a
> hive.compute.query.using.stats=false control and an all-NULL case.
>
> - Applies equally to native tables.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
