ICYMI: RPweave — Unified R + Python + LaTeX Workflow (Powered by uv)

If you juggle R, Python, and LaTeX for research, you know the pain: fragmented scripts, mixed environments, manual copying, and fragile reproducibility.

I needed a setup/workflow that handles both languages, any LaTeX template, environment isolation, and a command-line–first workflow — so I assembled RPweave. Not a new idea, but a polished, modern take that just works.

RPweave (GitHub template here: https://github.com/Techtonique/RPweave) ties everything together using:

• knitr for R + Python chunks
• reticulate for seamless Python integration
• LaTeX for publication-ready typesetting
• uv for fast, reproducible Python environments
• Makefile automation for building and previewing
• A ready-to-clone Git template

Getting Started

The workflow starts by cloning the RPweave template repo, and listing the Python packages you need in requirements.txt. Then, set up the isolated environment and install R dependencies via make setup. Finally, write your .Rnw document mixing R and Python chunks, and build with make view.

git clone https://github.com/Techtonique/RPweave my-paper
cd my-paper
uv venv venv && source venv/bin/activate
make setup
make view

Why It Matters

RPweave lets you:

• Run R and Python in the same .Rnw document at the command line
• Easily share objects across languages
• Use any LaTeX template or academic style
• Keep everything reproducible with isolated environments
• Build your PDF with a single command (make view)

Minimal Example

The first chunk is mandatory.

<<setup>>=
library(knitr); library(reticulate)
use_python("venv/bin/python")
@

<<r-analysis>>=
ggplot(mtcars, aes(wt, mpg)) + geom_point()
@

<<python-analysis, engine='python'>>=
import pandas as pd
print(pd.DataFrame({'x': range(100)}).describe())
@

Ideal For

• Papers mixing R stats + Python ML
• Projects needing clean LaTeX output
• Reproducible workflows with pinned dependencies
• Researchers tired of context-switching between RStudio, Jupyter, and LaTeX editors

Pro Tips

• Use make view as your main loop — instant rebuild + preview
• Store long chunks in chunks/
• Keep generated files out of Git
• Pass data R → Python via py$object for smooth cross-language flows

Repo & Docs

• RPweave: https://github.com/Techtonique/RPweave
• uv: https://docs.astral.sh/uv/
• knitr: https://yihui.org/knitr/