Installation guide uses Python 3.8, which is incompatible with current OpenML requirements

Start with the installation guide and pyproject.toml, comparing the documented Python version and editable test-install command with the declared requirements and test extra. Reproduce the reported failures, including missing openml_sklearn and minio, and inspect the affected tests. Done means supported shell-compatible setup works, dependencies resolve, and optional imports are skipped appropriately.

Description

The current installation guide and dependency configuration make it difficult to set up the development environment and run tests successfully. There are multiple inconsistencies that can lead to installation failures or missing dependencies.

Issues Identified

1. Outdated Python Version in Installation Guide

The installation instructions recommend using Python 3.8:

uv venv --seed --python 3.8 ~/.venvs/openml-python

However, the project requires:

Python >= 3.10,<3.15

This results in dependency resolution errors when installing:

uv pip install -e ".[test]"

---

2. Shell Compatibility Issue with Installation Command

The documentation suggests:

uv pip install -e .[test]

However, in shells like zsh, this results in:

zsh: no matches found: .[test]

because [] are treated as glob patterns.

This can be fixed by quoting or escaping:

uv pip install -e ".[test]"

---

3. Incomplete or Incorrect Test Dependencies

Installing test dependencies via:

uv pip install -e ".[test]"

does not always provide a fully working test environment.

For example, running tests can fail with:

ModuleNotFoundError: No module named 'openml_sklearn'

The test extra includes openml-sklearn, but this package does not appear to be available on PyPI, leading to unresolved dependencies.

---

4. Missing Core Dependency at Runtime (minio)

Although minio is listed as a core dependency in pyproject.toml, running the tests can result in:

ModuleNotFoundError: No module named 'minio'

This suggests that dependencies may not always be installed correctly via the current installation instructions, or that additional guidance is needed to ensure all runtime dependencies are properly installed.

---

5. Missing Dependency Handling in Tests

Some tests import optional dependencies (e.g., openml_sklearn) without guarding against missing packages, causing the entire test suite to fail.

It would be better to handle such imports using:

pytest.importorskip("openml_sklearn")

so that tests are skipped when optional dependencies are not available.

---

Expected Behavior

• Installation guide should use a supported Python version (>=3.10)
• Installation commands should be shell-compatible
• Installing .[test] should provide all necessary dependencies to run tests
• All core dependencies (e.g., minio) should be reliably installed
• Optional dependencies should be handled gracefully in tests

---

Suggested Fixes

1. Update installation guide to use Python >= 3.10 (e.g., 3.11)

2. Update installation command to:

   uv pip install -e ".[test]"
   

3. Review and correct test dependencies:

   • Verify or remove openml-sklearn

4. Ensure all required dependencies (including minio) are installed correctly

5. Use pytest.importorskip for optional dependencies in tests

---

Impact

These issues can prevent new contributors from successfully setting up the project and running tests, which may discourage contributions.

---

Additional Context

I would be happy to help update the documentation and fix the test setup if needed.