Multiple addons giving numpy errors, Blender 4.4 Mac

Support Technical Support
2

this is the answer Grok gave, i think it should help:

The error you’re encountering when trying to import NumPy in Blender 4.4 on macOS (“Symbol not found: (_cblas_caxpy$NEWLAPACK)”) indicates a compatibility issue between the NumPy version installed and the macOS Accelerate framework, which provides BLAS and LAPACK functions. This is a common issue when NumPy’s C-extensions fail to link properly with macOS’s system libraries, often due to mismatched expectations about the BLAS/LAPACK implementation. Let’s break this down and solve it step by step.

Why This Error Happens

  • Blender’s Python Environment: Blender 4.4 uses Python 3.11, bundled within its application (at /Applications/Blender-4.4/Blender.app/Contents/Resources/4.4/python/bin/python3.11). When you install NumPy, it’s installed into this isolated Python environment.
  • NumPy and Accelerate Framework: NumPy uses BLAS and LAPACK for numerical computations, and on macOS, it typically links to Apple’s Accelerate framework (/System/Library/Frameworks/Accelerate.framework). The error Symbol not found: (_cblas_caxpy$NEWLAPACK) means that NumPy’s _multiarray_umath module is trying to call a function (_cblas_caxpy) that it expects to find in Accelerate, but the symbol has a suffix ($NEWLAPACK) that Accelerate doesn’t provide in your macOS version.
  • Root Cause: This issue often arises because:
    • The NumPy version (1.26.4) was built with a different BLAS/LAPACK implementation (e.g., OpenBLAS or a newer Accelerate version) than what your macOS provides.
    • macOS version differences: macOS Ventura (13) and later updated the Accelerate framework, but if you’re on an older macOS (e.g., Monterey 12), the symbols might not match.
    • NumPy 1.26.4 might have been compiled with assumptions about macOS 14 (Sonoma) or later, while your system might be running an older version.

Steps to Fix the Issue

1. Check Your macOS Version

  • The error mentions the Accelerate framework, which varies across macOS versions. Run the following in Terminal to check your macOS version:
    sw_vers
  • Example output:
    ProductName:	macOS
ProductVersion:	14.4
BuildVersion:	23E214
  • If you’re on macOS 12 (Monterey) or earlier, the Accelerate framework might not support the $NEWLAPACK symbols that NumPy 1.26.4 expects. If so, consider upgrading macOS to Ventura (13) or later, as NumPy 1.26.4 is likely built for newer macOS versions.

2. Downgrade NumPy to a Compatible Version

  • NumPy 1.26.4 might be incompatible with your macOS version or Blender’s Python setup. Downgrade to an earlier version of NumPy that is known to work with Blender 4.4 and your macOS version.
  • Steps:
    • Open Terminal and navigate to Blender’s Python bin directory:
      cd /Applications/Blender-4.4/Blender.app/Contents/Resources/4.4/python/bin
    • Uninstall the current NumPy version:
      ./python3.11 -m pip uninstall numpy
    • Install an older version of NumPy, such as 1.25.2, which has broader compatibility with macOS versions:
      ./python3.11 -m pip install numpy==1.25.2
    • Test in Blender by running a script in the Scripting workspace:
      import numpy
print(numpy.__version__)
  • If 1.25.2 works without errors, this is the simplest fix. If not, try even earlier versions like 1.24.4.

3. Force NumPy to Use OpenBLAS Instead of Accelerate

  • NumPy can be built to use OpenBLAS instead of Accelerate, which avoids the symbol mismatch. You’ll need to install OpenBLAS and rebuild NumPy.
  • Steps:
    • Install OpenBLAS via Homebrew:
      brew install openblas
    • Uninstall NumPy:
      cd /Applications/Blender-4.4/Blender.app/Contents/Resources/4.4/python/bin
./python3.11 -m pip uninstall numpy
    • Set environment variables to tell NumPy to use OpenBLAS:
      export BLAS=/opt/homebrew/opt/openblas/lib/libopenblas.dylib
export LAPACK=/opt/homebrew/opt/openblas/lib/libopenblas.dylib
    • Reinstall NumPy with OpenBLAS:
      ./python3.11 -m pip install numpy --no-binary numpy
    • This forces NumPy to build from source using OpenBLAS, avoiding the Accelerate framework entirely.

4. Ensure Proper Python Environment

  • Blender’s Python environment should be isolated. If you have a system-wide NumPy installation that conflicts, it might cause issues.
  • Steps:
    • Verify that Blender’s Python is using the correct site-packages:
      /Applications/Blender-4.4/Blender.app/Contents/Resources/4.4/python/bin/python3.11 -m site
      Ensure the path includes /Applications/Blender-4.4/Blender.app/Contents/Resources/4.4/python/lib/python3.11/site-packages.
    • If there’s a conflict with a system-wide NumPy, uninstall it:
      /usr/bin/python3 -m pip uninstall numpy
    • Reinstall NumPy only in Blender’s environment (as shown in step 2).

5. Upgrade macOS (If Possible)

  • If you’re on macOS 12 (Monterey) or earlier, the Accelerate framework might be too old for NumPy 1.26.4. Upgrading to macOS 13 (Ventura) or 14 (Sonoma) can resolve this, as newer Accelerate versions include the $NEWLAPACK symbols.
  • Steps:
    • Go to System Settings > Software Update and upgrade to the latest macOS.
    • After upgrading, reinstall NumPy in Blender’s Python environment:
      cd /Applications/Blender-4.4/Blender.app/Contents/Resources/4.4/python/bin
./python3.11 -m pip install numpy

6. Use a Virtual Environment (Advanced)

  • Create a virtual environment to isolate dependencies and avoid conflicts with Blender’s Python:
    • Create a virtual environment:
      /Applications/Blender-4.4/Blender.app/Contents/Resources/4.4/python/bin/python3.11 -m venv ~/blender_venv
    • Activate the virtual environment:
      source ~/blender_venv/bin/activate
    • Install NumPy in the virtual environment:
      pip install numpy==1.25.2
    • Modify your add-on to use the virtual environment by adjusting sys.path:
      import sys
sys.path.append("~/blender_venv/lib/python3.11/site-packages")
import numpy

7. Check for Conflicting Libraries

  • Other libraries in Blender’s Python environment might interfere with NumPy’s linking to Accelerate. For example, if another library (e.g., SciPy) uses a different BLAS implementation, it can cause conflicts.
  • Steps:
    • List all installed packages in Blender’s Python:
      /Applications/Blender-4.4/Blender.app/Contents/Resources/4.4/python/bin/python3.11 -m pip list
    • If you see other numerical libraries (e.g., SciPy, pandas), uninstall and reinstall them after fixing NumPy:
      ./python3.11 -m pip uninstall scipy pandas
./python3.11 -m pip install scipy pandas

Recommended Solution

The simplest and most reliable fix is to downgrade NumPy to 1.25.2 (step 2). This version is known to work with Blender 4.4 on macOS and avoids the $NEWLAPACK symbol issue. If you need the latest NumPy features, consider upgrading macOS (step 5) or using OpenBLAS (step 3).

Test After Fixing

After applying one of the fixes, test in Blender:

  • Open Blender, go to the Scripting workspace, and run:
    import numpy
print(numpy.__version__)
  • If there’s no error, the issue is resolved.

Let me know if you need further assistance!