Description

lilpetamoix

opened on May 17, 2025

Issue body actions

Hi, I've imported the following libraries in my pyscript.

import pymc as pm
import arviz as az

When running the script, I get the following warning message:
"WARNING (pytensor.configdefaults): g++ not available, if using conda: conda install gxx
WARNING (pytensor.configdefaults): g++ not detected! PyTensor will be unable to compile C-implementations and will default to Python. Performance may be severely degraded. To remove this warning, set PyTensor flags cxx to an empty string."

So I installed Anaconda which should have those components that pytensor is looking for.

However, the problem is I'm having difficulty in figuring out how to "point" WinPython to the Anaconda directory so that pytensor can utilize g++ properly.

Thanks in advance.

Activity

1. stonebig commented on May 17, 2025

   

   stonebig

   on May 17, 2025

   Contributor

   More actions

   I'm not sure you can succeed quickly on this.
   Did you try on pytensor forum ?
   Or better nowodays, make an AI try to solve that question.
   If I had to do it for my local use on an isolated planet with no AI, I would look into adding g++ external tool to a VisualStudio https://stackoverflow.com/questions/59923813/how-to-use-g-compiler-in-visual-studio

   ... a gordian node-like method would be to use Anaconda: can't you ?
2. lilpetamoix commented on May 17, 2025

   

   lilpetamoix

   on May 17, 2025

   Author

   More actions

   so I asked ChatGPT the following:
   "I initially used WinPython to run pyscripts. I switched to PyCharm because I was able to run the pyscripts in the Anaconda environment. Is there a way to run pyscripts in WinPython but somehow link it to Anaconda environment?"

   Response:
   ✅ Two Working Solutions
   🔹 Option 1: Use WinPython IDE (Spyder or WinPython Control Panel) with Conda's Python
   Find your Anaconda env's Python path
   For pymc-env, it's usually:

   makefile
   Copy
   Edit
   C:\Users<you>\anaconda3\envs\pymc-env\python.exe
   Configure WinPython to use it
   If you're using WinPython Control Panel:

   Open the WinPython Control Panel

   Click “Register/Unregister” → point it to the conda env’s Python.exe

   Alternatively, open WinPython Command Prompt, then launch:

   bash
   Copy
   Edit
   your_script.py

   🔹 Option 2: Add Your Anaconda Env to WinPython Jupyter Kernel (if using notebooks)
   If you want to use Jupyter from WinPython, but execute in your conda env:

   bash
   Copy
   Edit
   conda activate pymc-env
   python -m ipykernel install --user --name pymc-env --display-name "Python (pymc-env)"
   Then open Jupyter in WinPython, and select the "Python (pymc-env)" kernel.

   I tried Option 1 by opening WPPM and tried to target the pymc-env in Anaconda but that failed, The Register functionality, correct me if I'm wrong, just registers the WinPython to the Windows environment but not necessarily set to Anaconda's environment.
3. stonebig commented on May 17, 2025

   

   stonebig

   on May 17, 2025

   Contributor

   More actions

   ok, LLM are still not always perfect, or too perfect for us.

   Option 1 doesn't make sense to me
   Option 2 is unlikely Windows territory: it starts per "bash", as Anaconda is using a sort of compatibility layer to Windows

   What I don't understand is the constraint that force you to try that exotic inter-breeding of Python stacks.
   With a clearer picture of the problematic, maybe a clearer path may appear.

   You need a master degree in IT, or a paid OpenAI, imho, to explore that world

   Most users/electrons/watts are stuck on linux/anaconda/nvidia for these things, it's a heavy leafting.
4. stonebig

   added

   for experimentators

   on May 25, 2025

Issue actions