The GitHub Actions job "Release Auditing" on texera.git/fix/7245-jsonl-offset has succeeded. Run started by GitHub user kz930 (triggered by kz930).
Head commit for run: 76712d3ee43a7ab775df1de6066894ec8cd85737 / kary zheng <[email protected]> fix(jsonl-scan): count the worker slice from the offset, not the file `open()` drops `offset` lines from the reader, so the iterator it keeps already begins at the requested row. The slice bounds were then computed as `offsetValue + ...` -- positions in the original file -- and applied to that already-offset iterator, skipping the offset a second time. When what is left is shorter than the offset, the worker emits nothing at all, with no error raised and nothing logged. Resolve the bounds against the iterator they are applied to. Over a five-line file, every combination of offset, limit and worker count was compared against `drop(offset).take(limit)` -- the reading ArrowSourceOpExec and this operator's own sourceSchema() already use. 36 of 105 were wrong before, none after. All 36 had an offset of 1 or more, which is why the existing spec, which covers the row limit but never the offset, stayed green. Closes #7245 Co-Authored-By: Claude Opus 5 (1M context) <[email protected]> Report URL: https://github.com/apache/texera/actions/runs/30726407953 With regards, GitHub Actions via GitBox
