buraksenn opened a new pull request, #25057: URL: https://github.com/apache/datafusion/pull/25057
## Which issue does this PR close? - Closes #24620. ## Rationale for this change Parquet source and sink protobuf hooks accessed fields individually and silently dropped source metadata prefetch hints and sink sorting-column metadata. This changed performance characteristics and omitted metadata from files written after a plan round trip. ## What changes are included in this PR? - Exhaustively destructure `ParquetSource`, `ParquetSink`, and their protobuf messages. - Preserve `ParquetSource::metadata_size_hint` with a checked integer conversion. - Preserve `ParquetSink::sorting_columns`, including the distinction between `None` and an empty list. - Explicitly document source/sink fields that are reconstructed or runtime-only. #24930 already preserves `reverse_row_groups` and `sort_order_for_reorder`; this PR builds on that behavior rather than duplicating it. The protobuf changes are additive and backward compatible. ## What is the testing strategy for this PR? Extended physical-plan round-trip coverage for the metadata-size hint and Parquet sorting columns. Validated with: - `cargo test -p datafusion-proto --test proto_integration roundtrip_parquet_exec_with_pruning_predicate` - `cargo test -p datafusion-proto --test proto_integration roundtrip_parquet_sink` - `cargo clippy -p datafusion-datasource-parquet -p datafusion-proto --all-features --tests -- -D warnings` - `cargo fmt --all --check` ## Are there any user-facing changes? Parquet scan metadata prefetch hints and sink sorting-column metadata now survive protobuf plan round trips. The protobuf additions are backward compatible. -- 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]
