uvr — A fast R package manager, written in Rust

Performance

Install benchmarks

Parallel downloads, global package cache with CoW cloning, no R process for binary installs. macOS arm64, R 4.5, P3M binaries. Median of 5 runs, index caches warm. First-install comparison: uvr pre-resolves deps via uvr lock (separate step) and only uvr sync (install) is timed; install.packages/pak/renv resolve + install inline. For a like-for-like lockfile-restore comparison, see pak::lockfile_install() and renv::restore().

ggplot2

17 packages

uvr

0.21s

renv

0.54s

install.pkgs

4.4s

pak

5.6s

21x faster than install.packages() (first install)

tidyverse

99 packages

uvr

0.27s

renv

0.82s

pak

8.1s

install.pkgs

11.8s

44x faster than install.packages() (first install)

Features

What it does

Package resolution, lockfiles, R version management, and isolated project libraries.

⚡

Fast installs

Parallel downloads, native tar extraction, no R process for binary packages.

🔒

Reproducible

uvr.lock pins exact versions and checksums. Commit it, and everyone gets the same environment.

📦

Project-isolated

Each project gets its own library in .uvr/library/. No cross-project version conflicts.

🅐

R version management

Install and switch R versions with uvr r install and uvr r use. Pin per-project via .r-version.

🚀

CRAN + Bioconductor + GitHub

Install from any registry. Specify GitHub packages directly in your manifest.

⚙

CI-ready

--frozen mode ensures CI uses exactly the locked versions. No surprises in production.

🌐

Cross-platform

Works on macOS, Linux, and Windows. Same workflow everywhere your code runs.

💾

Single static binary

One ~8 MB binary with zero dependencies. Download it and you're done.

Quick Start

Up and running in seconds

Initialize a project

Creates a uvr.toml manifest with your R version and empty dependency list.

Add packages

Resolves dependencies, writes the lockfile, and installs everything in one step.

Sync the environment

Installs exactly what the lockfile says. Idempotent and fast on repeat runs.

Run your script

Executes R with the project library on the search path. No library() preamble needed.

            
            
            
          

            # Initialize a project
            $ uvr init analysis
            ✓ Created uvr.toml and .Rprofile
             
            # Add packages — resolve, lock, install in one step
            $ uvr add ggplot2 dplyr
            › Resolving dependencies
              + ggplot2 3.5.1
              + dplyr 1.1.4
              · 14 transitive deps resolved
            ✓ Locked 16 packages in 0.8s
            ✓ Installed 16 packages in 2.1s
             
            # Run a script in the isolated env — no library() needed
            $ uvr run analysis.R
          

Comparison

Feature comparison

How uvr compares to renv, pak, rv, rig, and pixi.

Feature	uvr	renv	pak	rv	rig	pixi
Declarative manifest	✓	✓†	✓†	✓	—	✓
Lockfile	✓	✓	✓	✓	—	✓
R version management	✓	—	—	—	✓	✓
Run scripts in isolated env	✓	—	—	—	—	✓
CRAN packages	✓	✓	✓	✓	—	✓*
Bioconductor packages	✓	✓	✓	✓	—	✓*
GitHub packages	✓	✓	✓	✓	—	—
System dep detection (Linux)	✓	—	✓	—	—	✓
CI mode (--frozen)	✓	✓	—	—	—	✓
No admin rights required	✓	✓	✓	✓	—	✓
Standalone CLI (no R required)	✓	—	—	✓	✓	✓
Windows support	✓	✓	✓	✓	✓	✓

* Via conda-forge channel, not native CRAN/Bioconductor integration. † Via DESCRIPTION-based workflow, not a dedicated manifest format.

Installation

Install

Single binary, no dependencies. Pick your method.

Quick install

macOS and Linux — downloads the latest release

$ curl -fsSL https://raw.githubusercontent.com/nbafrank/uvr/main/install.sh | sh

Arch Linux (AUR)

Install via your favorite AUR helper

$ yay -S uvr

From R

Install via the companion R package

> pak::pak("nbafrank/uvr-r")
> uvr::install_uvr()

From source

Build with Cargo (requires Rust toolchain)

$ cargo install --git https://github.com/nbafrank/uvr

R Integration

Use uvr from R

The companion R package lets you manage your project without leaving the R console. Initialize projects, add packages, sync environments, and manage R versions — all from familiar R syntax.

View uvr-r on GitHub →

            
            
            
          

            # Install the companion package
            > pak::pak("nbafrank/uvr-r")
            > library(uvr)
             
            # Manage your project from R
            > init()
            ✓ Created uvr.toml
            > add("ggplot2")
              + ggplot2 3.5.1 (+ 8 deps)
            ✓ Installed 9 packages in 1.4s
            > sync()
            ✓ Environment up to date
          

A fast R package manager,written in Rust.