R packages on Marburg OPSI PCs

This guide is intended for members of the University of Marburg who want to use a personal R package library on university-managed PCs.

Install R packages in your user account on University of Marburg OPSI PCs

OPSI is a system for managing computers and deploying software. This guide concerns R on the university’s OPSI-managed Windows PCs.

Requirements: a university Windows PC, a university account and access to your H: drive.

Goal: store R packages in your account’s H: drive so that the library is accessible on other university PCs where the same drive is available. Those PCs must use a compatible R version.

  1. Check the current package library paths.

     .libPaths() # The first entry is normally used for package installation.
    

    A path on C: is local to this computer. Check whether the first entry points to your H: drive; the default can vary between installations.

  2. Create your personal package folder on H:. You only need to create this folder once.

     mypath <- "H:/R/win-library"
     dir.create(mypath, recursive = TRUE, showWarnings = TRUE)
    
  3. Make the library path persistent in your user .Rprofile. First, ask R which home directory it uses and where the user profile should be placed:

     path.expand("~")
     file.path(path.expand("~"), ".Rprofile")
    

    If the home directory is H:/Documents, create or edit H:/Documents/.Rprofile. Otherwise, use the home directory reported by R. Add this line, preserving any existing content:

     .libPaths("H:/R/win-library")
    

    By default, R looks for .Rprofile in its startup working directory before looking in the user home directory. A project-specific .Rprofile therefore takes precedence over the home file; include the library setting there too if needed. R_PROFILE_USER can select a different profile, and startup options can disable user profiles. See the R startup documentation.

  4. Restart R and check the library paths again.

     .libPaths()
    

    H:/R/win-library should now be the first entry. Repeat this check on another university PC before relying on the shared setup. The H: drive mapping and R home directory must be suitable there too.

Help for package installation

The following commands open help pages; they do not install packages. Read the relevant help page for the package name, repository or version arguments needed for an installation. Help for devtools or remotes requires that package to be installed; these helper packages can be installed from CRAN with install.packages().

Packages from CRAN:

?install.packages

Development versions from GitHub, which may be unstable:

?devtools::install_github

Previous package versions:

?remotes::install_version

See also R’s documentation shortcuts.

Install Rtools

On Windows, Rtools is needed to build packages from source when they contain code that must be compiled, such as C, C++ or Fortran. Installing a Windows binary package normally does not require Rtools. Whether a package comes from CRAN or GitHub does not by itself determine whether compilation is required.

Choose the Rtools version that matches your R version from the CRAN Rtools page. The R version appears at the top of the console when you open RStudio. For example, R 4.4.x uses Rtools44, whose default installation folder is C:/rtools44.

Rtools installed on C: is available only on that computer. It must therefore be available separately on each university PC where you need to compile packages; storing your R package library on H: does not make Rtools available on other PCs.

Updated: