[FLINK-36035] Remove zipped dependency files from flink-python

• Type: Technical Debt
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: API / Python
• Labels: pull-request-available
• Assignee: Unassigned
• Reporter: Grace Grimwood
• Created: 12/Aug/24 06:33
• Updated: 12/Aug/24 06:53

Description

There are zipped versions of Py4J and Cloudpickle stored in version control for use with PyFlink (as part of the build and also to use with the shell). It looks like they were put there because of issues with them not being auto-added to the PYTHONPATH, but this issue can be fixed without needing zipped copies of the libraries to be present locally.

It would be ideal to avoid storing these zip files in version control because git cannot parse them (they're treated as binaries) and they can also pose a security risk. I believe this can be done without altering anything user-facing in flink-python.

Issue Links

• links to: GitHub Pull Request #25191