comphead commented on PR #6725: URL: https://github.com/apache/datafusion-comet/pull/6725#issuecomment-6068866520
Rebased on main and pushed 4125e4291a. It addresses the review comments (replies inline) and has two more changes: - Reviewing it again, I found a regression in this PR. A partition source or equality key the query does not project is appended to the pruned schema, and `schemaWithRequiredFields` resolved it from `table.schemas()`, which lists the oldest schema first. A renamed column came back under its old name, which can clash with a column that took that name over. It now resolves from the current schema first, the same order as `FieldLookup` in Iceberg's Spark 4.1 `BaseReader`. Before this PR, the base schema was the full table schema except for `VERSION AS OF` reads of a dropped column, so live columns rarely went through that lookup. - The result checks moved to a SQL fixture, `sql-tests/iceberg/nested_schema_pruning.sql`, which also covers the rename case. The Scala suite keeps what a fixture cannot express: bytes scanned, equality deletes, `VERSION AS OF`, the schema pool, and files without field ids. -- 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]
