prewarm

Warm up Python environments by importing packages and triggering their one-time side effects — font caches, C extension loading, BLAS discovery, and more.

Usage

# Warm up via IPython (warms IPython's own init + packages)
prewarm matplotlib pandas numpy

# Or as a module
python -m prewarm matplotlib pandas numpy

# Skip IPython, just import directly
prewarm --no-ipython matplotlib pandas

API

from prewarm import warm

# Boot IPython, import modules, exit
warm(["matplotlib", "pandas"])

# Just import, no IPython
warm(["matplotlib", "pandas"], ipython=False)

Why?

First imports of heavy packages like matplotlib and pandas are slow — they build font caches, load C extensions, discover BLAS libraries, etc. Running prewarm in a fresh environment triggers all of these one-time costs up front so subsequent imports are fast.

When run with IPython (the default), it also warms IPython's startup path: traitlets configuration, magic commands, tab completion, and the display system. This is especially useful for Jupyter kernels where IPython boot time is part of the perceived startup latency.