dwsmith1983 opened a new pull request, #53263:
URL: https://github.com/apache/spark/pull/53263
<!--
Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, please read our contributor guidelines:
https://spark.apache.org/contributing.html
2. Ensure you have added or run the appropriate tests for your PR:
https://spark.apache.org/developer-tools.html
3. If the PR is unfinished, add '[WIP]' in your PR title, e.g.,
'[WIP][SPARK-XXXX] Your PR title ...'.
4. Be sure to keep the PR description updated to reflect all changes.
5. Please write your PR title to summarize what this PR proposes.
6. If possible, provide a concise example to reproduce the issue for a
faster review.
7. If you want to add a new configuration, please read the guideline first
for naming configurations in
'core/src/main/scala/org/apache/spark/internal/config/ConfigEntry.scala'.
8. If you want to add or modify an error type or message, please read the
guideline first in
'common/utils/src/main/resources/error/README.md'.
-->
### What changes were proposed in this pull request?
This PR enables Dynamic Partition Pruning (DPP) optimization when joining
with CommandResult nodes (e.g., results from SHOW PARTITIONS).
Changes made to
sql/core/src/main/scala/org/apache/spark/sql/execution/dynamicpruning/PartitionPruning.scala:
1. Modified hasSelectivePredicate() to recognize CommandResult as selective
(line 212)
2. Modified calculatePlanOverhead() to return 0.0 overhead for CommandResult
since data is already materialized (lines 187, 199)
3. Modified hasPartitionPruningFilter() to exclude plans containing
CommandResult from being used as DPP filter sources (line 221)
Added test coverage in
sql/core/src/test/scala/org/apache/spark/sql/DynamicPartitionPruningSuite.scala
to verify DPP works correctly with CommandResult.
Built and tested against tag v4.0.1 locally to verify the results and Spark
plan as well.
https://issues.apache.org/jira/browse/SPARK-54554
### Why are the changes needed?
Previously, when using SHOW PARTITIONS results in a broadcast join, Spark
would perform full table scans instead of applying Dynamic Partition Pruning.
Example scenario where this matters:
val partitions = spark.sql("SHOW PARTITIONS fact_table")
.selectExpr("cast(split(partition, '=')[1] as int) as partition_id")
.agg(max("partition_id"))
spark.table("fact_table")
.join(partitions, col("partition_id") === col("max(partition_id)"))
Before this fix: Full table scan of all partitions
After this fix: DPP prunes to only the relevant partition(s)
### Does this PR introduce any user-facing change?
Yes. Queries that join partitioned tables with SHOW PARTITIONS results (or
other commands returning CommandResult) will now benefit from Dynamic Partition
Pruning, potentially improving performance by scanning fewer partitions.
The behavior change is transparent to users - existing queries will simply
run faster without any code changes required.
### How was this patch tested?
Added new test case "DPP with CommandResult from SHOW PARTITIONS in
broadcast join" in DynamicPartitionPruningSuite that verifies:
- DPP is applied when joining with CommandResult
- Correct query results are returned
- Plan contains DynamicPruningSubquery operator
Ran full DynamicPartitionPruning test suite (73 tests total) - all passed
Tested manually with local Spark build using various CommandResult scenarios
### Was this patch authored or co-authored using generative AI tooling?
No
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
