Today, I have put Apache Datafusion in the SQL Arena.
The exact implementation under test is a vanialla DataFusion operating directly on top of Parquet files. People interested in running queries on it can use the Docker container supplied in the DBProve repo.
At first glace, the DataFusion planner appears quite simplisitic. It scores low on nearly all compares with other query planners.
Note that because this DataFusion runs on top of pure Parquet files - it only has access to the statistics embedded in those files.