09Catho commented on issue #55486: URL: https://github.com/apache/arrow/issues/55486#issuecomment-6786422553
Removing the marks alone isn't enough, because the body is stale and fails once it runs. I copied the test body out without the marks and ran it against the released wheel (pyarrow 26.0.1, numpy 2.4.6, pandas 3.1.1), not a build of main. The first block fails: ``` E assert [<pyarrow.Int...t64Scalar: 3>] == [0, 1, 2, 3] E At index 0 diff: <pyarrow.Int64Scalar: 0> != 0 ``` Iterating the array yields `Int64Scalar` objects, so `np.asarray([_ for _ in arr]).tolist()` returns scalars, not ints. With that assertion changed to `[x.as_py() for x in np_arr] == [0, 1, 2, 3]` and the isinstance check changed to `pa.Int64Scalar`, the test passes. The rest of the body passes unchanged. #55264 edits the same lines (`Int64Value` to `Int64Scalar`) but keeps the `tolist()` assertion, so that line would still fail once the marks are fixed. I haven't opened a separate PR because it would overlap with #55264. If you'd rather review it on its own, I can send one with the marker fix too (drop `pandas`/`nopandas`, add `numpy`). -- 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]
