fix(mlflow): fix KeyError in error message for missing environment variables

[Copilot]

Description

Fixes a bug where the error message handler itself throws a KeyError when attempting to inform users about missing TMPDIR/SCRATCH environment variables.

What problem does this change solve?

Bugfix: The error handler accessed os.environ["SCRATCH"] and os.environ["TMPDIR"] directly when checking if they exist, causing a KeyError instead of showing the intended error message to the user.

What issue or task does this change relate to?

Closes #622

Additional notes

Changes:

Minimal fix to the error message formatting in export_log_output_file_path():

• Changed os.environ["TMPDIR"] → os.getenv("TMPDIR", "not set")
• Changed os.environ["SCRATCH"] → os.getenv("SCRATCH", "not set")

Before:

if not os.getenv("TMPDIR") and not os.getenv("SCRATCH"):
    error_msg = "Please set one of those variables TMPDIR:{} or SCRATCH:{} to proceed.".format(
        os.environ["SCRATCH"],  # KeyError if not set!
        os.environ["TMPDIR"],   # KeyError if not set!
    )
    raise ValueError(error_msg)

After:

if not os.getenv("TMPDIR") and not os.getenv("SCRATCH"):
    error_msg = "Please set one of those variables TMPDIR:{} or SCRATCH:{} to proceed.".format(
        os.getenv("TMPDIR", "not set"),  # Safe with default value
        os.getenv("SCRATCH", "not set"),  # Safe with default value
    )
    raise ValueError(error_msg)

This is a minimal 2-line change that fixes the issue without unnecessary refactoring.

As a contributor to the Anemoi framework, please ensure that your changes include unit tests, updates to any affected dependencies and documentation, and have been tested in a parallel setting (i.e., with multiple GPUs). As a reviewer, you are also responsible for verifying these aspects and requesting changes if they are not adequately addressed. For guidelines about those please refer to https://anemoi.readthedocs.io/en/latest/

By opening this pull request, I affirm that all authors agree to the Contributor License Agreement.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[HCookie]

So, copilot has done some strange things here....

[Copilot]

So, copilot has done some strange things here....

Agreed - I over-engineered the solution. Reverted to a minimal fix that just addresses the KeyError issue. Commit b30241a.

[Copilot]

Pull request overview

This PR fixes a bug in the error message handler for export_log_output_file_path() in mlflow_sync.py. When neither TMPDIR nor SCRATCH environment variables are set, the error handler was attempting to access these variables directly using os.environ[], which itself raised a KeyError instead of displaying the intended helpful error message to the user.

Changes:

• Fixed KeyError in error message formatting by using os.getenv() with default values instead of direct dictionary access
• Changed error message construction to safely handle missing environment variables

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.