[ARROW-14214] [Python][CI] wheel-windows-cp36-amd64 nightly build failure

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 6.0.0
• Component/s: Continuous Integration, Packaging, Python
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/29797
• Assignee: Joris Van den Bossche
• Reporter: Antoine Pitrou
• Created: 04/Oct/21 09:53
• Updated: 11/Jan/23 08:38
• Resolved: 07/Oct/21 07:08

Description

See https://github.com/ursacomputing/crossbow/runs/3788550781?check_suite_focus=true#step:8:308

For some reason, Orc is disabled in the Python 3.6 Windows wheel build, but the test doesn't detect it:

__________________________ test_file_format_pickling __________________________

    def test_file_format_pickling():
        formats = [
            ds.IpcFileFormat(),
            ds.CsvFileFormat(),
            ds.CsvFileFormat(pa.csv.ParseOptions(delimiter='\t',
                                                 ignore_empty_lines=True)),
            ds.CsvFileFormat(read_options=pa.csv.ReadOptions(
                skip_rows=3, column_names=['foo'])),
            ds.CsvFileFormat(read_options=pa.csv.ReadOptions(
                skip_rows=3, block_size=2**20)),
            ds.ParquetFileFormat(),
            ds.ParquetFileFormat(dictionary_columns={'a'}),
            ds.ParquetFileFormat(use_buffered_stream=True),
            ds.ParquetFileFormat(
                use_buffered_stream=True,
                buffer_size=4096,
            )
        ]
        try:
>           formats.append(ds.OrcFileFormat())
E           AttributeError: module 'pyarrow.dataset' has no attribute 'OrcFileFormat'

Issue Links

• links to: GitHub Pull Request #11336