iliasabk commented on PR #51394: URL: https://github.com/apache/arrow/pull/51394#issuecomment-5818170811
Update on the earlier test failures: they were a bug in the new test, not in the reader changes. `ColumnBuilder` emits one chunk per processed block, including blocks that contain no complete row. In the first case, the first block (`"zzz…\r` ending inside an unfinished quoted field) yields an empty chunk, so `column(0)->chunk(0)` was empty and `GetString` read past the offsets buffer — hence the `""` result and the UBSan crash in the ASAN runs. Same shape in the second case (`chunk(0)` holds only the first row). The new commit calls `CombineChunks()` before asserting on `chunk(0)`; the reader changes themselves are unchanged. Could a maintainer approve the workflow runs when convenient? Thanks. -- 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]
