ryan-williams commented on issue #39399: URL: https://github.com/apache/arrow/issues/39399#issuecomment-1873320952
Apologies, [runsascoded/parquet-diff-test](https://github.com/runsascoded/parquet-diff-test) is "public" now. [Here's](https://github.com/runsascoded/parquet-diff-test/actions/runs/7366550240) the GitHub Action that ran the test. [Here's](https://github.com/runsascoded/parquet-diff-test/blob/2f6767c161562baaa606de0a29f44287d8cbfff7/parquet_diff_test/cli.py#L55-L56) the code that generated the "empty" (0-row) Parquet files: ```python df = pd.DataFrame([{ 'a': 111 }]) empty_df = df.iloc[:0] ``` (the issue also occurs with non-empty Parquet files, this was just the smallest repro I found). Here are example Parquet files from each OS, using {`pyarrow`,`snappy`}: - [ubuntu](https://github.com/runsascoded/parquet-diff-test/blob/ubuntu/out/pyarrow/snappy/empty.parquet) - [windows](https://github.com/runsascoded/parquet-diff-test/blob/windows/out/pyarrow/snappy/empty.parquet) (matches ubuntu) - [macOS](https://github.com/runsascoded/parquet-diff-test/blob/macos/out/pyarrow/snappy/empty.parquet) ([diff](https://github.com/runsascoded/parquet-diff-test/compare/ubuntu..macos#diff-1aff51203a0bbf705859a61d542f15bfa553b121b30fea500f03024a8ae44258), via `xxd`) -- 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]
