eldenmoon opened a new pull request, #67168:
URL: https://github.com/apache/doris/pull/67168

   ## What problem does this PR solve?
   
   When an old segment footer still contains a column UID that is absent from 
the current rowset schema, `Segment::new_column_iterator` finds the footer 
metadata and then fails schema lookup. The scan returns `NOT_FOUND` instead of 
using the existing default-value compatibility iterator.
   
   The failure is in the same partial-column-reader path touched by #53511. 
That PR changes reader ownership/caching, but does not cover a footer UID 
missing from the current rowset schema; the mismatch remains reproducible on 
current master.
   
   This change restores the default iterator fallback for that schema-evolution 
case and adds a segment regression test using a physical footer with UID 1 and 
a rowset schema containing only UID 0.
   
   ## Test
   
   - Current `apache/master` red: 
`SegmentIteratorExprZonemapTest.NewColumnIteratorUsesDefaultForFooterColumnMissingFromRowsetSchema`
 failed with `[NOT_FOUND] column not found in segment, col_uid=1`.
   - This PR green: the same test passed 1/1 in the physical environment.
   - Physical Cloud SQL red/green: with the same controlled 
historical-footer/rowset-schema mismatch and SQL, the parent returned 
`NOT_FOUND`; the fixed BE returned the expected rows.
   - `git diff --check` and clang-format passed.
   
   ## Release note
   
   Restore query compatibility for historical segments whose footer contains 
columns omitted from the rowset schema.
   
   ## Checklist
   
   - Test: BE unit test and physical end-to-end validation
   - Behavior changed: Yes, restores schema-evolution compatibility
   - Documentation: No


-- 
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