Configuring RStudio Workbench#
Administrators can configure RStudio Workbench to automatically use RStudio Package Manager when users install packages.
The necessary configuration files are documented in the RStudio Workbench Administration guide, example configurations are provided here for common scenarios.
A Single Repository#
In the most common scenario, users will install all packages from a single RStudio Package Manager repository that may contain CRAN packages and internal packages. An admin can discourage users from changing this repository setting. In this scenario, configure RStudio Workbench:
The exact URL to use is available in the RStudio Package Manager Setup page for the repository.
(RStudio Workbench only) To disable the repository option menu and discourage users from changing the repository setting, also include in
Internal Packages and CRAN Packages#
Another common scenario is to set up two repositories in RStudio Package Manager, one to serve CRAN packages and one to serve internal packages. (This setup also applies if you're only using RStudio Package Manager for internal packages and want to allow users to access a public CRAN mirror).
/etc/rstudio/rsession.conf, remove any existing
- Create a file called
/etc/rstudio/repos.conf. The file should contain:
Internal can be replaced with whatever name makes sense for your organization, but the repo containing CRAN packages should be indexed with the keyword
Allow Users to Optionally Add Additional Repos#
In addition to the two scenarios described above, some organizations may have additional
repositories for certain groups of users, e.g. "bleeding edge" packages
available in a dedicated "dev" repository. Administrators can configure
Global Options menu to automatically offer RStudio Package Manager repositories as optional, secondary repos that a user can opt into using.
RStudio Package Manager supports serving precompiled binary packages for the CRAN source, which can significantly speed up CRAN package installations.
To enable binary packages for users, follow the Configuration Steps and the R Configuration Steps for Linux or Windows to set up your R environment. For Linux binaries, also configure RStudio Workbench to use the binary repository URLs.
RStudio Workbench can be configured to use a Bioconductor repository as the default Bioconductor mirror. Unlike CRAN repositories, the default Bioconductor mirror must be configured separately in an R startup file.
We recommend setting the default Bioconductor mirror in the site-wide R startup
Rprofile.site), which applies to all users for a single version of R.
- Find the location of the
R_HOME/etc/Rprofile.site. For a given version of R,
R_HOMEcan be found by running the
R RHOMEcommand at a terminal:
$ R RHOME << /opt/R/4.0.3/lib/R << # Rprofile.site is located at: << /opt/R/4.0.3/lib/R/etc/Rprofile.site
- Create the
R_HOME/etc/Rprofile.sitefile if it does not already exist.
R_HOME/etc/Rprofile.site, add the repository setup code from the Setup page of a Bioconductor repository:This may also include setup code to enable Bioconductor use in an offline environment.
# Set the default Bioconductor mirror options(BioC_mirror = "https://<package-manager-server-address>/<bioconductor-repo-name>")
For more information on configuring R startup files, see this article.
Precedence of Settings#
If you have a site-wide R startup file (
Rprofile.site) that modifies the
repository setting, that file will take precedence over RStudio Workbench's
Some Ubuntu and Debian R binaries (installed via
apt-get install) include
Rprofile.site file that modifies the repository option. Unless you
intended to have this file, we recommend removing
Users can still modify the repository setting for their project needs using a
user or project-level
.Rprofile file, which takes precedence over
Checking For Success#
Once edits have been made to the appropriate configuration files (see above), start a new RStudio Workbench session. Closing and reopening the browser tab or pressing the reload button on the browser does not start a new session.
At that point, running
options('repos') from the RStudio Workbench console
should show the URL of the RSPM instance. If you've configured the default
Bioconductor mirror, running
options('BioC_mirror') should also show the URL
of the RSPM instance.
For a complete test, you can also confirm that packages are installed from the RSPM instance by default. To test a CRAN or internal package repository, run the following commands to install a package in the repository:
# Install a package in the repository install.packages("test-package") # Check that the package's DESCRIPTION file includes "Repository: RSPM" packageDescription("test-package")
While you can install any package as a test, one that is small and has few
dependencies will be the quickest. The
A3 package on CRAN is the first
alphabetically that meets this criteria.
To test the Bioconductor mirror, you can download and install a Bioconductor package:
# Install the BiocManager package from CRAN install.packages("BiocManager") # Check that BiocManager uses RSPM for the Bioconductor repositories BiocManager::repositories() # Install a Bioconductor package BiocManager::install("BiocVersion", update = FALSE) # Check that the package's DESCRIPTION file includes "Repository: RSPM" packageDescription("BiocVersion")