mirror of
https://github.com/GreptimeTeam/greptimedb.git
synced 2026-01-03 11:52:54 +00:00
f46cdbd66b
* fix/fast-path-for-single-region-bulk-insert: ### Commit Summary - **Refactor `try_decode` Method**: Updated the `try_decode` method in `FlightDecoder` to accept a reference to `FlightData` instead of consuming it. This change affects multiple files including `database.rs`, `region.rs`, `flight.rs`, `bulk_insert.rs`, `stream.rs`, and `region_request.rs`. - **Optimize Bulk Insert Handling**: Added a fast path for handling bulk inserts when only one region is involved in `bulk_insert.rs`. * fix/fast-path-for-single-region-bulk-insert: Improve `FlightDecoder` usage in tests - Updated `try_decode` method calls in `flight.rs` to remove unnecessary references for `d1`, `d2`, and `d3`. - Ensured consistency in handling `FlightMessage` variants within test cases. * fix/fast-path-for-single-region-bulk-insert: **Enhancement: Skip Empty Regions in Bulk Insert** - Updated `bulk_insert.rs` to improve efficiency by skipping regions without data during the bulk insert process. This change ensures that regions with a `true_count` of zero are not processed, optimizing resource usage and performance. * fix/fast-path-for-single-region-bulk-insert: ### Commit Summary - **Refactor `RegionMask` Handling**: - Introduced `RegionMask` struct to encapsulate boolean array and selected rows count. - Updated methods to use `RegionMask` instead of `BooleanArray` for region selection. - Affected files: `bulk_insert.rs`, `multi_dim.rs`, `partition.rs`, `splitter.rs`. - **Optimize Region Selection**: - Removed unnecessary checks for empty regions in `bulk_insert.rs`. - Improved logic for handling default regions in `multi_dim.rs`. - **Update Tests**: - Modified test cases to accommodate `RegionMask` changes. - Affected files: `multi_dim.rs`, `splitter.rs`. * fix/fast-path-for-single-region-bulk-insert: **Enhancements to MultiDimPartitionRule Logic and Tests** - **`multi_dim.rs`**: Improved the logic for selecting rows in `MultiDimPartitionRule` by optimizing the selection process when only one region is present. - **Tests**: Added new test cases to verify the behavior of default regions with unselected rows, existing default regions, and scenarios where all rows are selected. These tests ensure robust handling of partition rules and validate the correct assignment of rows to regions.