Issues

• Native PartialMerge aggregate emits duplicate keys under memory pressure, so count(DISTINCT) over-counts

  area:aggregationbugpriority:criticalrequires-triage
  apache/datafusion-comet#6857 · Filter by author andygrove opened on Oct 10, 2026

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

  enhancementEPICperformancerequires-triage
  apache/datafusion-comet#6855 · Filter by author andygrove opened on Oct 10, 2026
  2

  

• Push Iceberg residual predicates on decimal columns into the native scan

  area:Icebergarea:scanenhancementperformancerequires-triage
  apache/datafusion-comet#6854 · Filter by author andygrove opened on Oct 10, 2026

• Follow-ups to native NULL short-circuiting of array and map functions

  area:expressionsenhancementrequires-triage
  apache/datafusion-comet#6850 · Filter by author andygrove opened on Oct 10, 2026

• Check the timezone label of every native expression that returns a timestamp

  area:expressionspriority:lowrequires-triagetest
  apache/datafusion-comet#6849 · Filter by author andygrove opened on Oct 10, 2026

• Nightly CI failed on 2026-10-10

  ci-nightly-failure
  apache/datafusion-comet#6848 · Filter by author github-actions opened on Oct 10, 2026

• Native date functions and date-to-string cast return NULL or fail for dates outside chrono's NaiveDate range

  requires-triage
  apache/datafusion-comet#6843 · Filter by author viirya opened on Oct 10, 2026
  1

• Native cast string to date returns NULL for years outside chrono's NaiveDate range

  requires-triage
  apache/datafusion-comet#6842 · Filter by author viirya opened on Oct 10, 2026

• cast(string as decimal) returns NULL for non-ASCII Unicode digits that Spark accepts

  requires-triage
  apache/datafusion-comet#6840 · Filter by author viirya opened on Oct 10, 2026

• Native string to float/double and decimal casts accept strings Spark rejects

  requires-triage
  apache/datafusion-comet#6839 · Filter by author viirya opened on Oct 10, 2026

• CometScanRule reads fs.comet.libhdfs.schemes from SQLConf, but native reads it from the Hadoop conf

  requires-triage
  apache/datafusion-comet#6837 · Filter by author dwsmith1983 opened on Oct 10, 2026
  1

  

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

  correctnessEPICperformancerequires-triage
  apache/datafusion-comet#6834 · Filter by author andygrove opened on Oct 10, 2026 · 1.2.0
  

• [followup] Broadcast build-side auto-bridge only accelerates the innermost join when a fact has multiple unsupported-format dimensions

  bugrequires-triage
  apache/datafusion-comet#6832 · Filter by author parthchandra opened on Oct 9, 2026
  

• Cast of a literal that fails under ANSI fails the query at planning time, even in a branch no row reaches

  requires-triage
  apache/datafusion-comet#6825 · Filter by author viirya opened on Oct 9, 2026

• IN evaluates later candidates over the whole batch, so an ANSI error can fire for a row that already matched

  bugrequires-triage
  apache/datafusion-comet#6823 · Filter by author viirya opened on Oct 9, 2026

• Native Iceberg scan returns wrong values for a column renamed after a VERSION AS OF snapshot

  area:Icebergarea:scanbugcorrectnessrequires-triage
  apache/datafusion-comet#6822 · Filter by author andygrove opened on Oct 9, 2026

• Native fanout Iceberg writes: sort the remaining rows by partition when the memory pool runs short, instead of writing small files

  area:Icebergarea:memoryarea:writerenhancementrequires-triage
  apache/datafusion-comet#6818 · Filter by author andygrove opened on Oct 9, 2026

• Native timezone conversions stop applying DST after 2099, while Spark keeps applying the zone's rules

  area:expressionsbugcorrectnesspriority:lowrequires-triage
  apache/datafusion-comet#6816 · Filter by author andygrove opened on Oct 9, 2026

• Non-AQE DPP subquery doesn't reuse the Comet broadcast when the dimension is a UNION

  area:joinsperformancerequires-triage
  apache/datafusion-comet#6815 · Filter by author andygrove opened on Oct 9, 2026

• Interrupt a native scalar Arrow UDF when its task is killed

  enhancementrequires-triage
  apache/datafusion-comet#6811 · Filter by author viirya opened on Oct 9, 2026
  1

  

• perf: without AQE, native operators read Comet shuffle through a JVM round trip per block, making small reduce tasks up to 4x slower than Spark

  area:ffiarea:shuffleenhancementperformancerequires-triage
  apache/datafusion-comet#6807 · Filter by author comphead opened on Oct 8, 2026
  

• perf: every Comet broadcast hash join probe task re-decodes the whole broadcast in the JVM

  area:joinsenhancementperformancerequires-triage
  apache/datafusion-comet#6806 · Filter by author comphead opened on Oct 8, 2026 · 1.2.0
  1

  

• ShuffleScan metrics are dropped under AQE: shuffle read and decode time missing on the direct-read path

  bugrequires-triage
  apache/datafusion-comet#6804 · Filter by author comphead opened on Oct 8, 2026
  

• CometIcebergWriteDetectionSuite assumes a split Iceberg write plan

  area:Icebergarea:writerrequires-triagespark sql tests
  apache/datafusion-comet#6803 · Filter by author andygrove opened on Oct 8, 2026

• Null-named attribute fails Arrow export with "field name cannot be null"

  bugrequires-triagespark sql tests
  apache/datafusion-comet#6802 · Filter by author andygrove opened on Oct 8, 2026