Today — crates/oxidelake-storage/tests/pruning.rs:109,123 assert row_groups_pruned_statistics >= 90 and row_groups_pruned_bloom_filter >= 90. Page-index pruning is enabled (session.rs:241) but never asserted: grep -c page_index crates/oxidelake-storage/tests/pruning.rs → 0. The README says Parquet is pruned "by DataFusion's statistics, page index and Bloom filters … with proofs".
Fix — a test that writes a dataset with a sorted id column and a small data-page size, runs a narrow range predicate, and asserts page_index_rows_pruned > 0 (from the scan metrics, as the other two proofs do) and result equality with the pruning-disabled run.
Done when — the third proof exists beside the other two and the README sentence is true for all three.
Today —
crates/oxidelake-storage/tests/pruning.rs:109,123assertrow_groups_pruned_statistics >= 90androw_groups_pruned_bloom_filter >= 90. Page-index pruning is enabled (session.rs:241) but never asserted:grep -c page_index crates/oxidelake-storage/tests/pruning.rs→ 0. The README says Parquet is pruned "by DataFusion's statistics, page index and Bloom filters … with proofs".Fix — a test that writes a dataset with a sorted
idcolumn and a small data-page size, runs a narrow range predicate, and assertspage_index_rows_pruned > 0(from the scan metrics, as the other two proofs do) and result equality with the pruning-disabled run.Done when — the third proof exists beside the other two and the README sentence is true for all three.