txwyy123 commented on issue #59787: URL: https://github.com/apache/spark/issues/59787#issuecomment-6405027200
Prototype branch with the initial fix for `SortMergeJoinExec` and `HashJoin` (inner + outer): https://github.com/txwyy123/spark/tree/wide-output-split-joins **What's done:** - `SortMergeJoinExec.doProduce`: wraps `consume()` for both `outputRow` and `guardOutputRow` in separate methods via `ctx.addNewFunction()`, following the existing `codegenFullOuter` / `consumeFullOuterJoinRow` pattern - `HashJoin.codegenInner` + `HashJoin.codegenOuter`: same wrapping for both unique-key and non-unique-key paths - Test in `WholeStageCodegenSuite`: wide 50+50 column inner join, outer join, and sort-merge join with correctness check **What's not yet done (follow-up):** - `HashAggregateExec` output (wide GROUP BY + many aggregates) - `ColumnarToRowExec` (wide columnar scan) - Semi/anti/existence joins (lower priority — output only streamed-side columns) - Benchmark numbers Complementary to #59678 (SPARK-60431, `ProjectExec` output splitting). -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
