[ARROW-4057] [Python] Revamp handling of file URIs in pyarrow.parquet

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: Python
• Labels: parquet
• External issue URL: https://github.com/apache/arrow/issues/15909
• Assignee: Unassigned
• Reporter: Wes McKinney
• Created: 18/Dec/18 02:59
• Updated: 11/Jan/23 07:31

Description

The way this is being handled currently is pretty brittle. If the HDFS cluster being used to run the unit tests does not support writes from $USER then the tests fail (e.g. the only permissioned user in the docker-compose cluster is "root", so the unit tests cannot be run)

I'm inserting various hacks to get the tests passing for now, but they are temporary. There is code relating to path and URI handling spread throughout the parquet module; it would be much better to centralize and clean this up