All pull requests

Pull requests

• fix(file): weigh cached footers by the memory they hold

  #11120·gitRasheed opened Nov 2, 2026Contributor·

  ·

• fix(spark): reject a file column read as a type it cannot be read as

  #11119·gitRasheed opened Nov 2, 2026Contributor·

  ·

• perf(spark): split large files into row ranges

  #11118·gitRasheed opened Nov 2, 2026Contributor·

  ·

• perf(spark): read Utf8View values in place instead of copying each one

  #11117·gitRasheed opened Nov 2, 2026Contributor·

  ·

• fix(layout): reject list layout offsets that cannot match the row count

  #11116·gitRasheed opened Nov 2, 2026Contributor·

  ·

• Decline DataFusion projections the Vortex scan does not evaluate

  #11113·mhk197 opened Nov 2, 2026Contributor·

  ·

• Build Arrow dictionaries for primitives without arrow_cast

  changelog/performance
  #11111·mhk197 opened Nov 2, 2026Contributor·

  ·
  15

• Faster Spark reads and smaller JNI and stats fixes

  changelog/performance
  #11109·robert3005 opened Nov 2, 2026Contributor·

  ·
  1

• Grouped SumV2 kernel for run-end list elements

  changelog/performance
  #11107·mhk197 opened Nov 2, 2026Contributor·4/4·

  ·

• Grouped SumV2 kernel for sparse list elements

  changelog/performance
  #11106·mhk197 opened Nov 2, 2026Contributor·3/4·

  ·

• list_sum benchmarks for bool, run-end, chunked and sparse elements

  changelog/chore
  #11105·mhk197 opened Nov 2, 2026Contributor·1/4·

  ·

• Keep Arrow dictionaries on write as one dictionary per column

  changelog/performance
  #11103·mhk197 opened Nov 2, 2026Contributor·

  ·
  15

• Reduce monomorphization in sequence and run-end take

  changelog/chore
  #11102·robert3005 opened Nov 2, 2026Contributor·

  ·
  6

• Join chunked dictionaries on Arrow export without rehashing

  changelog/performance
  #11101·mhk197 opened Nov 2, 2026Contributor·

  ·
  17

• Share dictionary values between V2 filters and projections

  #11098·joseph-isaacs opened Nov 2, 2026Contributor·

  ·
  19

• Push every Vortex-capable operation into the Spark file format

  changelog/feature
  #11096·robert3005 opened Nov 2, 2026Contributor·

  ·

• Register parent-reduce rules with the session kernels

  changelog/break
  #11092·robert3005 opened Nov 2, 2026Contributor·4/4·

  ·

• Deprecate the session-less optimize and keep constructors lazy

  changelog/deprecation
  #11091·robert3005 opened Nov 2, 2026Contributor·3/4·

  ·

• Optimize arrays bottom up and as execution reaches them

  changelog/performance
  #11090·robert3005 opened Nov 2, 2026Contributor·2/4·

  ·
  1

• Group filter conjuncts by shared fields, and decode a shared column once per unit

  #11087·joseph-isaacs opened Nov 2, 2026Contributor·

  ·
  8

• fix(layout): rebase sliced list chunks before writing the list layout

  #11082·gitRasheed opened Nov 2, 2026Contributor·

  ·
  8

• Push DataFusion dynamic filters into V2 scans

  changelog/performance
  #11073·joseph-isaacs opened Oct 31, 2026Contributor·

  ·
  9

• fix(duckdb): handle table filters in aggregate pushdown

  changelog/fixext/duckdb
  #11071·jackylee-ch opened Oct 31, 2026Contributor·

  ·
  2

  

• fix(jni): retain Java Arrow buffers independently

  changelog/fix
  #11069·ChristopherWirt opened Oct 30, 2026Contributor·

  ·
  8

• ci: run GPU jobs only after formatting and lints pass

  #11066·joseph-isaacs opened Oct 30, 2026Contributor·

  ·