Davis/ordering fields as precombine#570
Draft
Davis-Zhang-Onehouse wants to merge 26 commits into
Draft
Conversation
xushiyan
reviewed
Mar 31, 2026
| /// Initialize env_logger exactly once for the lifetime of the loaded shared library. | ||
| /// | ||
| /// Reads RUST_LOG from the environment at the time the first FileGroupReader is created. | ||
| /// Example: RUST_LOG=hudi_core=debug enables all hudi-core debug logs. |
Member
There was a problem hiding this comment.
can we split the changes into separate PR pls? the title suggests it's about support "ordering fields" as alias, so let's separate out logger changes?
Author
There was a problem hiding this comment.
sorry, all these PRs are still WIP, please ignore all PRs under my name.
xushiyan
reviewed
Mar 31, 2026
| let get_result = if matches!(self, Self::PrecombineField) { | ||
| get_result.or_else(|_| { | ||
| configs | ||
| .get("hoodie.table.ordering.fields") |
Member
There was a problem hiding this comment.
we need to add a new enum item called OrderingFields as a table config, and mark PrecombineField as deprecated. Then all applicable callers resolve ordering fields first then fall back to precombine
Root Cause The HoodieDeleteRecordList.avsc and HoodieDeleteRecord.avsc schemas in hudi-rs were outdated. Java's current Hudi codebase uses wrapper record types (BooleanWrapper, IntWrapper, LongWrapper, etc.) for the orderingVal union field, while hudi-rs still had raw Avro types (int, long, float, etc.). This caused a union index mismatch: - Java writes LongWrapper at union index 3 (null=0, BooleanWrapper=1, IntWrapper=2, LongWrapper=3, ...) - hudi-rs expected raw long at union index 2 (null=0, int=1, long=2, float=3, ...) When hudi-rs read index 3, it interpreted it as float (4 fixed bytes) instead of LongWrapper (a record with zig-zag encoded long). This consumed wrong number of bytes, shifting all subsequent reads off alignment, producing "Union index 24 out of bounds: 2" on the next record. Changes 1. schemas/HoodieDeleteRecordList.avsc — Updated orderingVal union from raw types to wrapper record types matching Java 2. schemas/HoodieDeleteRecord.avsc — Same update 3. src/schema/delete.rs — Updated unit tests to use wrapper record values 4. tests/file_group_reader_tests.rs — Added reproduction test + fixed empty partition path handling 5. crates/test/src/lib.rs — Added V9MorNonpart3Commits test table variant + id_name_price helper 6. crates/test/data/.../v9_mor_nonpart_3commits.zip — New test data from failing table
Add 4 new HoodieFileGroupReader e2e tests validating against gold data from Hudi Spark's TestMORFileSliceLayouts: - table_log_compaction: log-only with compacted log block - table_log_only: log-only with data + delete blocks - table_column_projection: base + log with full schema (23 columns) - table_all_data_types: base + 3 log files with all data types Fixes to support these tests: - Add Decimal128 support in Avro-to-Arrow array reader - Add Map type support (Avro Map -> Arrow Map, not Dictionary) - Fix Array/List element field naming to match Parquet convention - Add direct struct field extraction for nested types in Map/List values - Fix log-only merge schema resolution (use reader_schema when no base file) - Add schema reconciliation in records_to_batch for field name differences between Avro-derived (log) and Parquet-derived (base) schemas Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
How are the changes test-covered