quox

With quox (pronouced like quarks), we envision a fundamentally new way to develop desktop applications for Windows, macOS, and Linux from the same code base.

What We Want

Electron was a major step forward towards cross-platform applications. However, it is successful despite a number of serious drawbacks, mainly regarding performance, developer experience, and security.

quox fixes all of them at once by rethinking the entire stack of technologies.

Instead of taking HTML, CSS, and JavaScript, and bundling both Node and Chromium with it into a gigantic executable, you can take a TSX file and run it directly. Windowing, layouting, composing, rendering, and input handling are all done via library calls.

Basically, with quox, a minimal example project has two lines of code and no build step.

// main.tsx
import { renderToWindow } from "https://quox.dev/mod.ts";

await renderToWindow(<h1>Hello, world!</h1>);

The renderToWindow function then takes your TSX and does everything a browser does in order to bring it to the screen. Just run

deno run --allow-ffi main.tsx

and you have sucessfully created your first cross-platform desktop application with quox.

As a result, quox works quite differently from Electron.

Electron	quox
Bundles Node	Runs inside Deno
Bundles Chromium	Calls into servo
Bundles HTML and CSS	Runs TSX natively
Has a complex build process	Has no build step
Runs JavaScript in several environments that differ slightly	Runs TypeScript in a single, clean environment
Spreads out an application across several processes, and connects them via IPC	Runs in a single process with wasm threads and web workers
Cannot fully benefit from Chromium's security	Fully benefits from Deno's security

Both quox and Electron have full system access, native graphics rendering performance, cross-platform support, and anything else you'd expect from a native application.

A direct consequence our approach is that quox can do things nobody else can do:

1. Run a full desktop app from nothing but a URL via deno run https://...
2. Deduplicate quox on disk (even across several indepdendent apps) if run from a URL
3. Secure your entire desktop application via a sandbox, with interactive permissions if desired
4. Have hot module replacement (HMR) out of the box during development

What We Have

quox is more than an idea. We have a concrete plan.

However, there is no demo yet. The above code snippet illustrates what we aim for, but it does not run yet.

We have implemented many individual parts of quox already, but there are still some core components that are missing.

Who We Are

We are a small team of volunteers with too little time who do open-source work. Progress is steady but slow.

This means that we cannot develop at a particularly high pace. Don't expect a release in the next few weeks.

It also means that we can give time to the underlying technologies to mature, and to impact their development in a direction that suits us. We are early, but not wrong.

Come join us!