JingsongLi commented on PR #340: URL: https://github.com/apache/paimon-rust/pull/340#issuecomment-4697198290
I found one issue that is easy to miss when this PR is reviewed independently from #374. In `crates/paimon/src/table/kv_file_reader.rs`, the merged batch is reordered with `RecordBatch::try_new(output_schema.clone(), columns)`. When the projection is empty, e.g. `COUNT(*)`, `columns` is empty and Arrow cannot infer the row count from the arrays, so the reordered batch can lose the row count even though the merge stream produced rows. This affects aggregation/PK KV reads because this PR routes aggregation tables through the KV merge path. Please either include the `RecordBatchOptions::new().with_row_count(Some(batch.num_rows()))` fix here, or make this PR explicitly depend on the PR that adds it, so `COUNT(*)` / empty projection reads do not regress when this lands. -- 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]
