[
https://issues.apache.org/jira/browse/HIVE-26319?focusedWorklogId=780788&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-780788
]
ASF GitHub Bot logged work on HIVE-26319:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 13/Jun/22 11:46
Start Date: 13/Jun/22 11:46
Worklog Time Spent: 10m
Work Description: kasakrisz opened a new pull request, #3362:
URL: https://github.com/apache/hive/pull/3362
### What changes were proposed in this pull request?
Rewrite update statements of iceberg tables to multi insert statement
similarly in case of native acid tables.
When generating the rewritten statement:
* Get the virtual columns from the table's storage handler in case of non
native acid tables
* Include the old values to the select clause of the delete branch of the
multi insert statement.
When executing the multi insert:
* Two iceberg writers are used which produce a data delta file and a delete
delta file. The result of these writers should be merged into one
`FilesForCommit` if both writers are run in the same task.
* In case of more complex statements (ex. partitioned and/or bucketed) more
than one Tez task produces commit info so this patch enables storing all of
them.
* Every `FileSinkOperator` creates its own jobConf instance because the
iceberg write operation is stored in it and it is different in both instance.
### Why are the changes needed?
See #2855
+ Preparation for iceberg Merge implementation.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
```
mvn test -Dtest.output.overwrite -DskipSparkTests
-Dtest=TestIcebergLlapLocalCliDriver -Dqfile=update_iceberg_partitioned_orc2.q
-pl itests/qtest-iceberg -Piceberg -Pitests
```
Issue Time Tracking
-------------------
Worklog Id: (was: 780788)
Remaining Estimate: 0h
Time Spent: 10m
> Iceberg integration: Perform update split early
> -----------------------------------------------
>
> Key: HIVE-26319
> URL: https://issues.apache.org/jira/browse/HIVE-26319
> Project: Hive
> Issue Type: Improvement
> Reporter: Krisztian Kasa
> Assignee: Krisztian Kasa
> Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Extend update split early to iceberg tables like in HIVE-21160 for native
> acid tables
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
