Setup webR with bun

The idea is to show how to use webR with a more complex example (more then the hello world with a build in HTTP server) using R dependencies containing Fortran code (so the worst case possible :D). The endpoint minpack is based on Nonlinear Least-Squares Data Fitting - Example D.2 .

Prerequisites

You will need to either install bun or use a docker image mounting the src folder into the image. I will use docker since it is easier for me to work with.

Docker setup

Get the image that you like the most. Example

docker pull oven/bun:1.3.10-debian

Run the image mounting the src folder into the container

docker container run --rm -it -p 3000:3000 -v ./src:/home/bun/app oven/bun:1.3.10-debian bash

VS Code setup

Install the dev containers extension in vs code. Once you open this project vs code will prompt you to open the project in the bun:1.3.10-debian container.

Recreate setup from scratch

Initialize bun project

bun init 

leading to the setup

package name (app): webrbun
entry point (index.ts): 

Done! A package.json file was saved in the current directory.
 + index.ts
 + .gitignore
 + tsconfig.json (for editor auto-complete)
 + README.md

To get started, run:
  bun run index.ts

Add the dependencies

bun add elysia @elysiajs/swagger webr

add dev dependencies for linting and formatting

bun add --dev --exact @biomejs/biome
bunx biome init

Check everything is working

bun run ./index.ts

output

Hello via Bun!

Add script(s) to the package.json file for easier calling of repeated steps

"scripts": {
  "runREST": "bun run ./index.ts"
}

REST API description

I will build a REST api using

• ElysiaJS - backend/routing etc.
• @elysiajs/swagger - plugin to auto-generate openapi page
• @elysiajs/static - plugin to serve static content
• WebR - model fitting and random number generation
• biomejs - lint and format
• vanjs - minimal reactive frontend library

Running

bun runREST

will make the REST API and WEB UI available under localhost:3001.

You can test the endpoints with the default values using the openapi endpoint localhost:3001/swagger.

Acknowledgments

Microsoft. (2025). Copilot (GPT-4) [Large Language Model]. Retrieved from https://copilot.microsoft.com