Faster package installs on Linux with Package Manager beta release

R packages can take a long time to install, especially on Linux servers or Docker containers. To address this problem, the newest version of RStudio Package Manager includes access to pre-compiled packages for Linux that install significantly faster — imagine installing the tidyverse from scratch in under 2 minutes!

Get Started

If you are already using RStudio Package Manager, simply upgrade to the new version. We do not recommend using the new repository URLs in production during the beta period, but you can continue to use Package Manager in production while using the new repository URLs outside of production.

If you are new to RStudio Package Manager, you can download the product and immediately start a 45 day evaluation, which will include access to faster package installs!

Please leave your feedback or ask any questions below.

The Details

R packages come in different flavors. Often when you install a package, R must compile the package for your specific version of R and operating system. This step can take time. RStudio Package Manager 1.0.12 introduces beta support for the binary format of R packages. In this binary format, the repository’s packages are already compiled and can be “installed” much faster — “installing” the package amounts to just downloading the binary!

In a configured environment, users can access these packages using “install.packages” without any changes to their workflow. For example, if the user is using R 3.5 on Ubuntu 16 (Xenial):

> install.packages(“dplyr”, repos = “https://demo.rstudiopm.com/cran/__linux__/xenial/latest”)

Besides being lightning fast, all the user will notice is a change to the console output:

* installing *binary* package ‘dplyr’ …

To configure and prepare the environment, an administrator or R user will need to follow these steps:

  1. Ensure the version of R is compatible with the pre-compiled packages. We’ve confirmed the packages distributed by Package Manager are compatible with the default R installations from most Linux distributions. However, the best way to ensure consistency is to use the same R installation on your system that we used to compile the package. To make this easy, RStudio provides debian and rpm installers for R. Learn more here.

  2. Double check to ensure that the user agent reported by R in your environment correctly reports the version of R. Learn more here.

  3. Install any system requirements required by the packages you’ll want to use. This step isn’t just for pre-compiled packages, you’ll need to do this to use R packages on your system. Learn more here.

  4. Configure R to use binary packages by using the new repository URL provided by the Package Manager. Learn more here.

These steps ensure that if a user requests “ggplot2”, RStudio Package Manager can give them the binary that will work on their system. You wouldn’t want to give someone on Ubuntu the ggplot2 binary built on CentOS!

The binary format is currently available for R 3.4, 3.5, and 3.6 on CentOS/REHL 7, Ubuntu 16 Xenial, and Ubuntu 18 Bionic. More distributions, including SLES, are coming soon.

RStudio Package Manager does not yet support binary packages for Windows or Mac

The packages distributed in Package Manager are built on open source Docker images using an opinionated installation of R. You can learn more about this build process here.

8 Likes