dajiaohuang opened a new pull request, #18659: URL: https://github.com/apache/iceberg/pull/18659
## Summary - Read `_row_id` and `_last_updated_sequence_number` from stored ORC columns before applying file-level constants. - Fall back to the first row ID plus the batch offset and selected row position when `_row_id` is not stored. - Add direct vectorized-reader coverage and a Spark table-level ORC vectorized-read test for row lineage. Closes #18543. Related prior attempt: #17635 was closed automatically for inactivity; this PR adds the requested Spark table-level coverage. ## Test plan - Direct ORC reader: verify explicit and inherited row IDs and sequence numbers, including selected-row filtering. - Spark table read: write an ORC-backed v3+ table and verify vectorized metadata-column results against the committed file's first row ID and data sequence number. ## Verification - `:iceberg-spark:iceberg-spark-4.1_2.13:test --tests org.apache.iceberg.spark.data.TestVectorizedOrcDataReader` — passed (4 tests). - `:iceberg-spark:iceberg-spark-4.1_2.13:test --tests org.apache.iceberg.spark.source.TestSparkMetadataColumns.rowLineageValuesWithVectorizedOrcRead` — passed (v3 and v4 ORC/vectorized cases; other parameter combinations skipped by test assumptions). - Spark 4.1 `spotlessApply` completed; `git diff --check` passed. AI disclosure: AI assistance was used in drafting the implementation and test scaffolding. The behavior was reviewed against the row-based ORC reader and validated with the targeted tests above. -- 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]
