github-actions[bot] commented on code in PR #65262:
URL: https://github.com/apache/doris/pull/65262#discussion_r3578688631
##########
fe/be-java-extensions/iceberg-metadata-scanner/src/main/java/org/apache/doris/iceberg/IcebergSysTableJniScanner.java:
##########
@@ -106,9 +103,10 @@ protected int getNext() throws IOException {
break;
}
StructLike row = reader.next();
- for (int i = 0; i < fields.size(); i++) {
- SelectedField field = fields.get(i);
- Object value = row.get(field.sourceIndex,
field.field.type().typeId().javaClass());
+ for (int i = 0; i < requiredFieldCount; i++) {
+ // FE keeps the fields requested by BE at the start of the
Iceberg projection.
+ // FileScanTask.schema() is not the row schema for every
DataTask implementation.
+ Object value = row.get(i, Object.class);
ColumnValue columnValue = new
IcebergSysTableColumnValue(value, timezone);
Review Comment:
This empty-projection path still loses the row count for `COUNT(*)`. With
`requiredFieldCount == 0`, the scanner consumes each `StructLike` and
increments the local `rows` counter, but it never appends anything to the
zero-column `VectorTable`, so `JniScanner.getNextBatchMeta()` either hits the
`numRows == vectorTable.getNumRows()` assertion or writes a metadata row count
of 0. A Java-only `appendVirtualData(rows)` fix is also not enough end-to-end:
the legacy scanner ultimately returns based on `Block::rows()`, which is 0 for
a block with no columns, and the v2 JNI path checks `output_block->rows() ==
original_rows` after filling zero columns. Please either keep a cheap required
output field for system-table `COUNT(*)` (avoiding `readable_metrics`) or add
explicit BE support for propagating virtual zero-column row counts, and make
these COUNT regressions assert the expected count.
--
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]
