Python 3.6 Compatibility: dataclasses dependency in filesystem module

[jeremymanning]

Problem

The clustrix filesystem module uses Python 3.7+ dataclasses but some clusters (e.g., ndoli.dartmouth.edu SSH) run Python 3.6.8, causing:

No module named 'dataclasses'

Evidence

Failing Test Scripts:

• scripts/test_ssh_packaging_with_1password.py (filesystem/complex tests)
• scripts/test_slurm_packaging_jobs.py (filesystem/complex tests when paramiko issue is resolved)

Error Location:

• clustrix/filesystem.py imports dataclasses for FileInfo, DiskUsage classes
• Remote execution fails when extracting packages with filesystem utilities

Clusters Affected:

• SSH: ndoli.dartmouth.edu (Python 3.6.8)
• Potentially other older cluster environments

Solutions

Option 1: Compatibility Layer (Recommended)

try:
    from dataclasses import dataclass
except ImportError:
    # Python 3.6 compatibility
    def dataclass(cls):
        return cls

Option 2: Remove dataclasses dependency

• Replace dataclass objects with regular classes
• Maintain same API but use manual __init__ methods

Option 3: Version Detection

• Detect Python version in packaging
• Include dataclasses backport when needed

Acceptance Criteria

• Filesystem module works on Python 3.6.8+
• scripts/test_ssh_packaging_with_1password.py filesystem test passes
• No regression in Python 3.7+ environments
• All dataclass functionality preserved (FileInfo, DiskUsage)

Priority

HIGH - This blocks complete validation of the packaging system filesystem operations.

Related Issues

• Issue Core Architecture: Function Serialization and Dependency Management for Remote Execution #64: Function serialization architecture (core system working)
• Issue [VALIDATION] External API and Service Integration Testing Tracker #63: External API validation (completed)

[jeremymanning]

We don't actually need support for Python 3.6 or 3.7 (it's fine to support only 3.8+). But if we encounter a system with an unsupported Python version, we should run the necessary functions in a virtual environment with a supported version of python that we set up locally.