vinishjail97 commented on issue #834: URL: https://github.com/apache/incubator-xtable/issues/834#issuecomment-5321158360
## Progress update Where the table version 9 work currently stands. ### Pull requests - XTable: #835 adds the `xtable.hudi.target.table_version` config, switches instant selection and ordering to completion time on version 9, sets the partition-stats config from apache/hudi#19111 so column stats stay on for partitioned tables, and parameterizes the column-stats tests over table versions 6 and 9. - Hudi: apache/hudi#19123 fixes reading a table partitioned on a nested column. It is still open and waiting on a committer review. This is the last upstream change this issue depends on. ### Upstream status - apache/hudi#19111 is merged. - apache/hudi#19126 is merged. - apache/hudi#19123 is open with green CI, rebased onto current master. Since this issue was filed, Hudi master also reorganized the `hudi-common` packages in apache/hudi#19193 and apache/hudi#19195. Two of those moves affect XTable, and #835 now follows them: - `org.apache.hudi.stats` moved to `org.apache.hudi.metadata.stats`, covering `ValueMetadata`, `ValueType` and `HoodieColumnRangeMetadata`. XTable's `XTableValueMetadata` sits inside that package to reach package-private members, so the file itself moves too. - `org.apache.hudi.avro.HoodieAvroUtils` moved to `org.apache.hudi.common.avro.HoodieAvroUtils`. The generated `org.apache.hudi.avro.model` classes did not move. ### Validation Hudi was built from `apache/hudi` master at `989badc395` plus apache/hudi#19123, for Spark 3.4 and Scala 2.12, and the XTable suite was run against it. Everything passes: - `ITConversionController` 59 of 59, including the nested partition column case and the un-partitioned Paimon to Hudi case, both of which #772 had disabled - `ITHudiConversionSource` 49 of 49 - `ITHudiConversionTarget` 18 of 18 - `ITConversionService` 8 of 8 - Hudi unit tests in `xtable-core` 97 of 97 Without apache/hudi#19123 the single remaining failure is the nested partition column case, which fails with `HoodieSchemaException: Illegal character in: nested_record.level` out of `HoodieFileGroupReaderBasedFileFormat.buildReaderWithPartitionValues`. A full green CI run covering every unit and integration test is here: https://github.com/vinishjail97/onetable/actions/runs/32073259817 All eleven IT classes ran in that build, not only the Hudi ones. It uses a temporary workflow step that builds the Hudi modules XTable needs from a pinned SHA. That step lives on a throwaway branch for validation only and is not part of #835. ### Remaining gate CI on #835 itself still cannot pass. The branch pins `hudi.version` to `1.3.0-SNAPSHOT`, and no repository publishes it: Maven Central serves no snapshots, and the Apache snapshot repository carries no Hudi 1.x snapshot at all. So the build fails at dependency resolution rather than at compilation. This unblocks when a Hudi release carries apache/hudi#19111, apache/hudi#19126, apache/hudi#19123 and the package reorganization above. The latest Hudi 1.x release is 1.2.0 from 2026-05-23, which predates all of them, and there is no `release-1.3.0` branch yet. -- 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]
