Repository navigation

feat(scan): Support split and parallel scans - Part-2a: read Parquet files by task range - #1002

Merged

wgtmac merged 11 commits into

apache:mainfrom

kamcheungting-db:feat/file-scan-task-range-reading

Oct 21, 2026

Merged

feat(scan): Support split and parallel scans - Part-2a: read Parquet files by task range#1002
wgtmac merged 11 commits into
apache:mainfrom
kamcheungting-db:feat/file-scan-task-range-reading

Commits on Oct 21, 2026
- feat(scan): read Parquet files by task range
  
  kamcheungting-db
  authored and
  wgtmac
  committed
  4ca7e53
- docs(parquet): explain row-group split selection
  
  authored and
  wgtmac
  committed
  760fa08
- style(test): use ranges algorithms for split boundaries
  
  authored and
  wgtmac
  committed
  acf80e6
- fix(parquet): consume a row-group range only after its reader is installed
  
  authored and
  wgtmac
  committed
  1a765a8
- feat(parquet): assign row groups to splits by midpoint like Java Iceberg
  
  authored and
  wgtmac
  committed
  9aec9ff
- fix comment from part-1
  
  kamcheungting-db
  authored and
  wgtmac
  committed
  7c4939d
- fix(scan): simplify split readers and bound stale row estimates
  
  kamcheungting-db
  authored and
  wgtmac
  committed
  7d6323c
- fix(scan): preserve split estimates when file size changes
  
  kamcheungting-db
  authored and
  wgtmac
  committed
  34139a0
- fix(scan): cache split row estimates at task creation
  
  kamcheungting-db
  authored and
  wgtmac
  committed
  c7cfe94
- fix(scan): require stable metadata for row estimates
  
  kamcheungting-db
  authored and
  wgtmac
  committed
  be10adc
- test(parquet): generate split fixture at runtime
  wgtmac
  committed
  e6787c0