github-actions[bot] commented on code in PR #70416:
URL: https://github.com/apache/doris/pull/70416#discussion_r4545753152
##########
be/src/storage/segment/variant/sparse_column_merge_iterator.cpp:
##########
@@ -157,6 +161,12 @@ void
SparseColumnMergeIterator::_merge_to(MutableColumnPtr& dst) {
dst_sparse_column_values,
src_path, row);
}
+ // The interleaving above assumes the source sparse cells and the
materialized subcolumns
+ // are byte-ordered. Assert the merged row keeps that invariant for
downstream readers;
+ // debug-only, so release compaction pays nothing.
+ DCHECK(is_variant_sparse_row_sorted(dst_sparse_column_paths,
+
dst_sparse_column_offsets[ssize_t(row) - 1],
Review Comment:
[P2] Limit the destination assertion to the row just appended
`row` restarts at zero for each source batch, but `_merge_to` appends to
`dst`. `SegmentIterator::_read_columns_by_index` can call this iterator
repeatedly on the same destination for 256-row chunks of discontinuous row IDs,
for example after delete-bitmap filtering. On the second call, `row == 0` makes
this start offset `dst_sparse_column_offsets[-1] == 0`, so valid rows
containing `z` followed by a new row containing `a` are checked as one sequence
and abort a debug/ASAN BE. Capture the destination key count before appending
each row (or use its actual last offset) and check only that row's range; add
coverage that appends a second batch to a nonempty destination.
--
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]
