andygrove commented on PR #6992: URL: https://github.com/apache/datafusion-comet/pull/6992#issuecomment-6203937180
@dwsmith1983 @viirya, pushed ee66473fd. - `CometMapFromArrays` now mixes in `CodegenDispatchFallback`, so a `FLOAT` or `DOUBLE` key no longer drops the projection to Spark, and `LAST_WIN` is dispatched too. The `expressions.md` row now reads Hybrid. `map_from_arrays_dedup_policy.sql` disables the dispatcher so it still covers a real fallback, as `map_from_entries_dedup_policy.sql` does. - The dispatched `DUPLICATED_MAP_KEY` queries now use `expect_error_class`, including every query in the Spark 4.0 file. I also added `map_from_arrays` queries with `expect_dispatch` to the main and legacy fixtures. - The array-key zeros query now reads `id = 1`. The NaN-in-array-key case is its own query with its own comment. - The 35 `map` fixtures pass on Spark 3.5, 4.0 and 4.1, and `CometMapExpressionSuite` passes on 4.1. I did not re-run 3.4, 4.2 or `CometFloatSemanticsSuite` after this change. - Fixture names, the `Part of #6549` change and the doc trimming are answered in the inline threads. On the backport to `branch-1.1`: I'll do it after this merges, per the backporting guide. -- 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]
