[
https://issues.apache.org/jira/browse/HIVE-29637?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18086395#comment-18086395
]
Denys Kuzmenko edited comment on HIVE-29637 at 6/5/26 2:57 PM:
---------------------------------------------------------------
WHERE ds = `HIVE_DEFAULT_PARTITION` is a Hive-specific quirk and not something
I'd rely on semantically. __HIVE_DEFAULT_PARTITION is essentially Hive's
internal representation for partition values that are NULL.
Iceberg is an interoperable table format that can be accessed by multiple
engines, and other engines are not expected to expose or depend on Hive's
internal partition naming conventions. So I'd consider ds IS NULL the portable
and semantically correct way to express the predicate.
was (Author: dkuzmenko):
WHERE ds = `HIVE_DEFAULT_PARTITION` is a Hive-specific quirk and not something
I'd rely on semantically. __HIVE_DEFAULT_PARTITION is essentially Hive's
internal representation for partition values that are NULL.
>From a data model perspective, the partition value is still NULL.
For Iceberg tables in particular, I'd strongly prefer thinking in terms of NULL
values rather than `HIVE_DEFAULT_PARTITION`. Iceberg is an interoperable table
format that can be accessed by multiple engines, and other engines are not
expected to expose or depend on Hive's internal partition naming conventions.
So I'd consider ds IS NULL the portable and semantically correct way to express
the predicate.
> Iceberg: Incorrect Results for NULL Predicate on Partition Column
> -----------------------------------------------------------------
>
> Key: HIVE-29637
> URL: https://issues.apache.org/jira/browse/HIVE-29637
> Project: Hive
> Issue Type: Bug
> Reporter: Denys Kuzmenko
> Priority: Major
> Labels: pull-request-available
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
