Vivek1106-04 commented on PR #57504: URL: https://github.com/apache/spark/pull/57504#issuecomment-5306845289
Thanks for catching the version range. Agreed on dropping it. An upper bound (`>= "3.0.0" and < "3.0.5"`) would lock pandas-on-Spark to a pandas bug across a 4-patch window, and would also need a matching version guard in `test_frame_loc_setitem` so the test asserts the buggy no-op on 3.0.0-3.0.4 — both dead weight once those releases stop mattering. Since pandas fixed this in maintenance, matching fixed pandas seems like the right target. The tradeoff is the one you reproduced: on 3.0.0-3.0.4 pandas-on-Spark now applies the write while pandas still no-ops, so `test_frame_loc_setitem` fails there. CI is unaffected — `dev/requirements.txt` pins `pandas>=2.2.0,<3.0.0` and no workflow installs pandas 3. Updated the PR description to state the 3.0.0-3.0.4 vs 3.0.5 split, the reason for not adding an upper bound, and the divergence on those patch releases. -- 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]
