reticulate crashes R when importing pandas (numpy 2.2.6 + pandas 2.3.1)

[SpatLyu]

When importing pandas via reticulate, the entire R session crashes.
However, importing pandas directly in Python works without issue.

---

Environment:

• OS: Windows 11
• R version: 4.5.1
• reticulate version: 1.43.0
• Python version: 3.11.13 (conda-forge)
• numpy: 2.2.6 (conda-forge)
• pandas: 2.3.1 (conda-forge)

---

Reproducible example in R:

library(reticulate)

# point reticulate to conda env
use_python("D:/application/miniconda3/envs/geocompy/python.exe", required = TRUE)

# works:
py_run_string("import numpy; print(numpy.__version__)")

# crashes R session:
py_run_string("import pandas; print(pandas.__version__)")

Expected behavior:
pandas should import successfully (just like in Python).

Actual behavior:
R session immediately crashes without error message.

[t-kalinowski]

Does you see the same error outside of conda, in a venv?

[SpatLyu]

Does you see the same error outside of conda, in a venv?

I tested with a clean venv environment:

cd d:/
python -m venv r-venv
cd r-venv
Scripts\activate
python.exe -m pip install --upgrade pip
pip install numpy==2.2.6 pandas==2.3.1 joblib ruptures

Then test it in R

library(reticulate)
use_virtualenv("d:/r-venv", required = TRUE)
py_config()

> python:         D:/r-venv/Scripts/python.exe
> libpython:      C:/Users/31809/AppData/Local/R/cache/R/reticulate/uv/python/cpython-3.11.13-windows-x86_64-none/python311.dll
> pythonhome:     D:/r-venv
> version:        3.11.13 (main, Jul 11 2025, 22:36:59) [MSC v.1944 64 bit (AMD64)]
> Architecture:   64bit
> numpy:          D:/r-venv/Lib/site-packages/numpy
> numpy_version:  2.2.6

> NOTE: Python version was forced by use_python() function

py_run_string("import numpy; print(numpy.__version__)")
> 2.2.6
py_run_string("import pandas; print(pandas.__version__)")
> 2.3.1

This confirms that reticulate works correctly with venv.