voonhous commented on PR #19181: URL: https://github.com/apache/hudi/pull/19181#issuecomment-4912902316
The above bug highlighted is per-basefile, not per query, IIUC, it only triggers if: 1. A single .lance base file holds >512 rows, and 2. the read projects the BLOB column, and 3. one readAll FFI stream drains past row 512 (crossing Lance's internal BLOB page boundary). So, total rows ingested is irrelevant. What matters is rows-per-base-file. I think the demos spread >500 rows across multiple base files. No single file's readAll ever crossed the page boundary, which is why it never triggered the bug above. -- 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]
