JingsongLi commented on code in PR #8068:
URL: https://github.com/apache/paimon/pull/8068#discussion_r3339236735
##########
paimon-spark/paimon-spark-common/src/main/scala/org/apache/paimon/spark/PaimonSparkTableBase.scala:
##########
@@ -118,6 +118,9 @@ abstract class PaimonSparkTableBase(val table: Table)
_metadataColumns.append(PaimonMetadataColumn.ROW_ID)
_metadataColumns.append(PaimonMetadataColumn.SEQUENCE_NUMBER)
}
+ if (!coreOptions.vectorField().isEmpty) {
+ _metadataColumns.append(PaimonMetadataColumn.VECTOR_SEARCH_SCORE)
Review Comment:
This advertises `__paimon_vector_search_score` on every table with a vector
field, but the read path can only populate it when a vector-search scan has
produced `ScoreRecordIterator`s. A normal scan can still resolve this metadata
column, for example:
```sql
SELECT gid, __paimon_vector_search_score
FROM my_db1.vector_test
WHERE date = '20260420'
```
That scan has no pushed vector search, so `PaimonRecordReaderIterator`
reaches the score branch and casts the normal file iterator to
`ScoreRecordIterator`, which should fail at runtime with `ClassCastException`.
Can we gate this metadata column to `VectorSearchTable` / active vector-search
scans, or reject it explicitly when no vector search is present?
--
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]
