Recent upgrade to the pyarrow/arrow-cpp packages lead to a lot of breakage.

[xhochy]

We have several users reporting problems in the Apache Arrow project that are related to the arrow-cpp and pyarrow packages after they have been updated to 3.0.0 in Anaconda. These are issues that are not occuring in the conda-forge versions of them.

I'm using this issue to redirect user reports and to accumulate links to reported issues.

[xhochy]

HDFS doesn't seem to be supported anymore: https://issues.apache.org/jira/browse/ARROW-11555

[xhochy]

Snappy (the most important/default Parquet compression codec) isn't built-in anymore: https://stackoverflow.com/questions/66017811/python-error-using-pyarrow-arrownotimplementederror-support-for-codec-snappy

[xhochy]

The pyarrow.dataset modules wasn't built: https://stackoverflow.com/questions/66026964/pyarrow-3-0-0-and-1-0-0-reading-a-parquet-file-fails-when-passing-the-cont

[xhochy]

Some users see pyarrow as installed from pypi_0 even though they pulled it from defaults, e.g. here https://stackoverflow.com/questions/66047129/pyarrow-library-is-not-installed-but-i-am-able-to-import-it-downloading-data-f