C Extension Switching¶

Overview¶

NumPy + SciPy Stress Test¶

This demo is the ultimate stress test for OmniPkg’s LibResolver. It forces the interpreter to load, unload, and reload incompatible C-extension binaries (shared objects/DLLs) within the same process.

Usage¶

Prerequisite: This demo requires Python 3.11 (due to specific binary wheels used in the test). Try it now: Cloud users can click the play button to run this demo instantly:

omnipkg demo 3

# Run Demo #3: The "Nuclear" Option
omnipkg demo 3

What You’ll Learn¶

1. C-Extension Hot-Swapping: How OmniPkg manages the memory lifecycle of shared libraries (.so / .pyd).
2. Binary Compatibility: How it ensures numpy==1.24.3 (linked to older BLAS) doesn’t crash when swapped for 1.26.4.
3. The “Nuclear” Swap: Watch it combine incompatible pairs (numpy==1.24.3 + scipy==1.12.0) vs (numpy==1.26.4 + scipy==1.16.1) in real-time.

The Code¶

This script performs a series of high-speed swaps between incompatible versions of NumPy and SciPy, verifying that the math operations (which rely on C code) return correct results every time.

from omnipkg.loader import omnipkgLoader
import numpy as np

# 1. Baseline: NumPy 1.26.4 (Main Env)
print(f"Baseline Version: {np.__version__}")
# Output: 1.26.4

# 2. The "Impossible" Swap: NumPy 1.24.3
# This involves unloading C-extensions and reloading older binaries
with omnipkgLoader("numpy==1.24.3"):
    import numpy as np

    # Verify we are running C-code from 1.24.3
    print(f"Swapped Version: {np.__version__}")
    # Output: 1.24.3

    # Verify math accuracy (ensure memory isn't corrupted)
    print(f"Array Sum: {np.array([1, 2, 3]).sum()}") 
    # Output: 6

# 3. Restore: Back to 1.26.4
# OmniPkg handles the cleanup and re-linking automatically
import numpy as np
print(f"Restored Version: {np.__version__}")
# Output: 1.26.4

Live Execution Log¶

Pay attention to the Activation Time. OmniPkg swaps these massive C-libraries in ~118ms.

(evocoder_env) minds3t@aiminingrig:~/omnipkg$ omnipkg demo 3

============================================================
  🚀 STEP 3: Executing the Nuclear Test
============================================================

💥 NUMPY VERSION JUGGLING:

⚡ Switching to numpy==1.24.3
   🧹 Purging 1 module(s) from memory...
   - 🔒 STRICT mode
   - 🔩 Activating binary path: .../numpy-1.24.3/bin
   ⚡ HEALED in 21,684.8 μs
   ✅ Version: 1.24.3
   🔢 Array sum: 6
   ⚡ Activation time: 118.52ms
   🎯 Version verification: PASSED

🌀 omnipkg loader: Deactivating numpy==1.24.3...
   ✅ Environment restored.
   ⏱️  Swap Time: 47,614.668 μs

⚡ Switching to numpy==1.26.4
   ✅ Version: 1.26.4
   🔢 Array sum: 6
   ⚡ Activation time: 119.48ms
   🎯 Version verification: PASSED

🤯 NUMPY + SCIPY VERSION MIXING:

🌀 COMBO: numpy==1.24.3 + scipy==1.12.0
   🧪 numpy: 1.24.3, scipy: 1.12.0
   🔗 Compatibility check: [1. 2. 3.]
   🎯 Version verification: BOTH PASSED!
   ⚡ Total combo time: 162.69ms

🌀 COMBO: numpy==1.26.4 + scipy==1.16.1
   🧪 numpy: 1.26.4, scipy: 1.16.1
   🔗 Compatibility check: [1. 2. 3.]
   🎯 Version verification: BOTH PASSED!
   ⚡ Total combo time: 130.39ms

🚨 OMNIPKG SURVIVED NUCLEAR TESTING! 🎇

How It Works¶

1. Memory Purge: Before swapping, OmniPkg forces a garbage collection and manually unregisters the C-extension modules from sys.modules.
2. Binary Path Injection: It injects the bin path of the target version into LD_LIBRARY_PATH (Linux) or PATH (Windows) to ensure the dynamic linker finds the correct .so / .dll files.
3. Strict Mode: For packages identified as “C-Heavy” (like NumPy), OmniPkg enables Strict Mode, which performs deeper verification to prevent segmentation faults.