kiancm opened a new pull request, #2629: URL: https://github.com/apache/iceberg-rust/pull/2629
## Summary - Add public setters `with_preload_column_index()` and `with_preload_offset_index()` on `ArrowReaderBuilder`, following the same pattern as the existing `with_range_coalesce_bytes`, `with_metadata_size_hint`, and `with_range_fetch_concurrency` setters. - Plumb these settings through `TableScanBuilder` and `TableScan` so that `to_arrow()` respects them, following the same pattern as `row_group_filtering_enabled` and `row_selection_enabled`. ## Motivation `ParquetReadOptions` already has `preload_column_index` (default `true`) and `preload_offset_index` (default `true`), but there was no way for callers to change these defaults. For **wide tables** (many columns), preloading column and offset indices for ALL columns adds significant I/O overhead when only a few columns are actually queried. In testing on a 99-column table, disabling these reduced per-query metadata reads from **54 MB to 16.5 MB**. Exposing these settings lets users opt out of preloading when it is not needed, which is the common case for selective column reads on wide tables. ## Changes - `ArrowReaderBuilder`: added `with_preload_column_index(bool)` and `with_preload_offset_index(bool)` methods - `TableScanBuilder`: added `preload_column_index: Option<bool>` and `preload_offset_index: Option<bool>` fields with corresponding builder methods - `TableScan`: added the same fields, threaded through from `TableScanBuilder::build()` - `TableScan::to_arrow()`: applies the settings to `ArrowReaderBuilder` when present 🤖 Generated with [Claude Code](https://claude.com/claude-code) -- 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]
