Usingpipinside CondaPkg with Pixi backend (undocumented)

[mouad-tarif]

Hello everyone,

I want to share a method that seems undocumented but works perfectly.

The discovery

Inside CondaPkg.jl , when using the Pixi backend, you can directly call pip to install any PyPI package:

bash

/home/$USER/.julia/environments/v1.12/.CondaPkg/.pixi/envs/default/bin/python -m pip install pandas pyarrow

Why it matters

• Much faster than Conda backend
• Access to PyPI packages, not just Conda-forge
• Reproducible environment (via pixi.toml + pixi.lock )
• Isolated per Julia project

Minimal working example

1. Add CondaPkg to your project
2. Enable Pixi backend (automatic in latest versions)
3. Run the pip command above
4. Now PythonCall can import pandas !

Questions to the community

• Is this officially supported or a hack?
• Could this be documented properly?

I searched the docs and found no mention of this. Has anyone else tried this?

Looking forward to your insights.