Skip to contents

Choosing webrarian

Source: vignettes/when-to-use.qmd

webrarian builds a whole R workspace as a website. Visitors get an editor, a console, files, plots and an environment, with your packages and files ready when the page opens, and they write and run their own code against what you prepared.

A browser window at my-analysis.netlify.app showing the workspace. Numbered callouts mark the editor holding analysis.R, the console after installing packages and running the script, the Files tab with the file tree under web_user, the Plots tab with a bar chart, the footer with Reset files, and the address bar
Figure 1: What a visitor gets, already set up. There is an editor (1), a console (2), the environment and files (3), plots and help (4), edits kept between visits (5) and an address of your choosing (6).

It suits four kinds of job.

  • course materials where students start with the data and packages loaded
  • a reproducible analysis readers can rerun and change
  • a workshop sandbox with nothing to install
  • an internal R console on a network without internet access

Four features set it apart.

  • You build the site from R (catalog(), acquire_package(), bind()), with no HTML or JavaScript.
  • Sites can be self-contained. With build.offline: true a site makes no request to another server, and collection_mirror() adds a package repository visitors can install from on closed networks.
  • It compiles your local or GitHub packages (with Docker), or uses an r-universe that has built them.
  • It sets up GitHub Pages and Netlify deployment, brands sites with brand.yml, and supports share links.

pyodidarian, the Python counterpart, is released separately.

Other tools

No tool does everything, and webrarian is the wrong one for a Shiny app. The table shows what each was built for.

webrarian shinylive quarto-live webR REPL JupyterLite
Built for a workspace site around your packages and files Shiny apps runnable cells in a Quarto document an R console to use now notebooks
How you make it R functions and _webrarian.yml shinylive::export() a Quarto format nothing to build a JupyterLite build
Languages R R and Python R and Python R Python, and R through xeus-r

Here is each in a sentence.

  • shinylive runs Shiny apps in the browser, which the webrarian workspace does not.
  • quarto-live makes code blocks in a Quarto document runnable, for tutorials where the prose leads.
  • The webR REPL is webR’s own console, ready with nothing to build. webrarian adds your packages and files to a site you host.
  • JupyterLite runs notebooks, and R runs there through xeus-r with packages from emscripten-forge rather than repo.r-wasm.org.

They combine well. A Quarto course can use quarto-live for lessons and link to a webrarian workspace for assignments.