Uploaded image for project: 'Spark'
  1. Spark
  2. SPARK-45926

The InMemoryV2FilterBatchScan and InMemoryBatchScan are not implementing equals and hashCode correctly

    XMLWordPrintableJSON

Details

    • Bug
    • Status: Open
    • Major
    • Resolution: Unresolved
    • 4.0.0, 3.5.1
    • None
    • SQL

    Description

      These InMemoryV2FilterBatchScan and InMemoryBatchScan test classes are not implementing hashCode and equals correctly as they are not taking into account the pushed runtime filters. As a result they are unable to expose the TPCDS test issues which can show whether the reuse of exchange is happening correctly or not.
      If these classes implement equals and hashCode taking into account the pushed runtime filters, we would see that TPCDS Q14b which should ideally be reusing the exchange containing Union , is not happening due to multiple bugs which surface in AQE.

      Actual V2 DataSources like iceberg correctly implement equals and hashCode taking into account pushed runtime filters , which also expose the same issue of reuse of exchnage not happening

      Attachments

        Issue Links

          Activity

            People

              Unassigned Unassigned
              ashahid7 Asif
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

                Created:
                Updated: