manuzhang opened a new pull request, #17263:
URL: https://github.com/apache/iceberg/pull/17263

   ## Summary
   
   Backports #17039 to `1.10.x`.
   
   - Fixes `_last_updated_sequence_number` inheritance to use a data file's 
`data_sequence_number` in scan constants.
   - Adds regression coverage for a v2 rewrite/compaction followed by a v3 
upgrade and row ID assignment.
   - Applies the Spark regression to the versions supported on this branch: 
3.4, 3.5, and 4.0.
   
   ## Problem
   
   A v2 rewrite can add a replacement file with a newer file sequence number 
while preserving the original data sequence number. If the table is later 
upgraded to v3, existing files do not contain materialized row-lineage columns, 
so readers inherit `_last_updated_sequence_number` from scan constants.
   
   Using `file_sequence_number` reports the rewrite snapshot instead of the 
sequence associated with the row content. The row-lineage specification 
requires inheritance from the data sequence number.
   
   ## Impact
   
   This preserves accurate row update sequence information for tables compacted 
before upgrading from v2 to v3, improving row-lineage auditability.
   
   ## Backport notes
   
   The production and core test changes are logically equivalent to #17039. 
Spark coverage is adapted from the versions on `main` to the Spark 3.4, 3.5, 
and 4.0 modules present on `1.10.x`.
   
   ## Testing
   
   - Local tests were not run, per repository instructions.
   - An independent code review found no issues and verified equivalence with 
the upstream fix.
   - CI validation is pending.
   
   ---
   
   **AI Disclosure**
   
   OpenAI Codex assisted with preparing and reviewing this backport 
publication. The commit includes the required co-author trailer.
   


-- 
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]

Reply via email to