1.3.0

Issues

• [EPIC] Faster Spark-to-Arrow conversion in CometSparkToColumnarExec

  area:scanenhancementEPICperformance
  apache/datafusion-comet#6565 · Filter by author andygrove opened on Oct 3, 2026 · 1.3.0
  

• [EPIC] Nested-type conversion performance: C2R, R2C and JVM shuffle slower than Spark

  area:scanarea:shuffleenhancementEPICperformance
  apache/datafusion-comet#6855 · Filter by author andygrove opened on Oct 10, 2026 · 1.3.0
  2

  

• [EPIC] Performance follow-ups from profiling the end-to-end Iceberg benchmark

  area:IcebergenhancementEPICperformance
  apache/datafusion-comet#6911 · Filter by author andygrove opened on Oct 12, 2026 · 1.3.0
  1

  

• Store offsets as lengths in shuffle blocks to shrink shuffle files

  area:shuffleenhancementperformance
  apache/datafusion-comet#6941 · Filter by author andygrove opened on Oct 13, 2026 · 1.3.0
  

• Do not accelerate a query stage that contains a columnar-to-row fallback (enable whole-stage revert by default)

  enhancementperformanceuser experience
  apache/datafusion-comet#5207 · Filter by author andygrove opened on Aug 1, 2026 · 1.3.0
  

• Nested field-id reads give one file column to both an id match and a name match; the root returns null

  area:scanbugpriority:low
  apache/datafusion-comet#6994 · Filter by author andygrove opened on Oct 14, 2026 · 1.3.0
  1

  

• Spark's task CPU time (executorCpuTime) leaves out native execution for plans with no JVM input

  enhancement
  apache/datafusion-comet#6935 · Filter by author andygrove opened on Oct 13, 2026 · 1.3.0

• Add a GroupsAccumulator for the native regr_* aggregates

  area:aggregationenhancementperformancepriority:medium
  apache/datafusion-comet#6886 · Filter by author andygrove opened on Oct 12, 2026 · 1.3.0
  1

  

• Closing the shuffle direct-read stream early fetches every remaining block

  area:shufflebugperformance
  apache/datafusion-comet#6865 · Filter by author andygrove opened on Oct 11, 2026 · 1.3.0

• [EPIC] In-memory cache audit: correctness and performance follow-ups for InMemoryTableScanExec support

  area:scanbugcorrectnessEPICperformancepriority:critical
  apache/datafusion-comet#6834 · Filter by author andygrove opened on Oct 10, 2026 · 1.3.0
  

• Native Parquet scan assigns row groups to splits differently from Spark

  area:scanenhancement
  apache/datafusion-comet#6512 · Filter by author andygrove opened on Oct 1, 2026 · 1.3.0
  1

  

• [EPIC] Match Spark's -0.0 and NaN semantics systematically instead of per expression

  area:expressionsbugcorrectnessEPICpriority:critical
  apache/datafusion-comet#6385 · Filter by author andygrove opened on Sep 29, 2026 · 1.3.0
  

• perf: broadcast hash join rebuilds the build-side hash table in every task

  enhancementperformance
  apache/datafusion-comet#6013 · Filter by author andygrove opened on Sep 18, 2026 · 1.3.0
  2

  

• Track nested-field statistics pruning and TIMESTAMP_MILLIS overflow parity in filtered scans

  area:scanbugpriority:critical
  apache/datafusion-comet#5739 · Filter by author peterxcli opened on Sep 6, 2026 · 1.3.0
  2

  

• [EPIC] DataFusion 55 upgrade follow-ups

  enhancementEPIC
  apache/datafusion-comet#5709 · Filter by author andygrove opened on Sep 5, 2026 · 1.3.0
  3

• Enable the split-operator plan and native Iceberg writes by default

  area:ciarea:Icebergarea:writerenhancement
  apache/datafusion-comet#5644 · Filter by author andygrove opened on Sep 2, 2026 · 1.3.0
  

• Checked TIMESTAMP_MILLIS overflow for nested fields and nested-predicate scans is blocked on DataFusion nested-field pruning

  area:scanbugcorrectnesspriority:critical
  apache/datafusion-comet#5553 · Filter by author peterxcli opened on Aug 29, 2026 · 1.3.0
  3

  

• Match Spark ObjectHashAggregate decimal AVG buffer semantics

  area:aggregationbugpriority:critical
  apache/datafusion-comet#5509 · Filter by author sunchao opened on Aug 27, 2026 · 1.3.0
  

• Date-to-timestamp casts can overflow or panic for wide dates

  area:expressionsbugcorrectnesspriority:critical
  apache/datafusion-comet#5456 · Filter by author sunchao opened on Aug 24, 2026 · 1.3.0
  

• Bound memory when a partition has many files

  area:scanenhancement
  apache/datafusion-comet#5343 · Filter by author parthchandra opened on Aug 12, 2026 · 1.3.0
  

• Broaden storage-partitioned-join tests, then enable reportPartitioning by default

  area:scanenhancement
  apache/datafusion-comet#5342 · Filter by author parthchandra opened on Aug 12, 2026 · 1.3.0
  

• Skip the merge wrapper for single-file partitions

  area:scanenhancement
  apache/datafusion-comet#5341 · Filter by author parthchandra opened on Aug 12, 2026 · 1.3.0
  

• Keep read concurrency in the sorted merge path

  area:scanenhancement
  apache/datafusion-comet#5340 · Filter by author parthchandra opened on Aug 12, 2026 · 1.3.0
  

• Support transformed sort orders (bucket / truncate / etc.)

  area:scanenhancement
  apache/datafusion-comet#5339 · Filter by author parthchandra opened on Aug 12, 2026 · 1.3.0
  

• Merge on sort keys even when they are not selected

  area:scanenhancement
  apache/datafusion-comet#5338 · Filter by author parthchandra opened on Aug 12, 2026 · 1.3.0