The state of building user interfaces in Rust

How about some screenshots?

Its very difficult to compare X to Y anywhere on this site. Its just an aggregator, not really an exemplary resource.

egui is the clear winner for making desktop applications. I've built a complex application recently (think of it like an AI powered image editor, doing plenty of editor logic and communicating with several python backends for the AI part) and it's been smooth sailing. It would be nice to have a family of components that look native on every platform but nowadays the desktop experience is anyway wildly inconsistent (and web-centric).

Using qt bindings is a good option too, but depending on non rust code means you are more likely to catch some weird crash. My experience with Qt in Rust is years old, so I can't comment on stability.

For frontend development, leptos is really nice and it feels familiar coming from react - but the whole chain is too heavy, your target directory quickly balloons to GBs and that's unacceptable, especially if you have several frontend projects.

I vibe coded a proof of concept leptos (including islands) with a minimal runtime and no dependencies and the size was much more contained. There is margin for improvements but today I would stick with solid.js for frontend development.

The other big hurdle for Rust on the web is the need to compile to wasm. That means that any Rust application will be heavier than a similar app in another JS framework. If we could target js instead of wasm, maybe we could have apps with small bundles.

The downside is that the dependency stack you need to do gui programming with rust is massive and the compile times are brutal. You can’t beat the application performance, though. It’s crazy how nice it feels compared to bloated electron apps.

The question of “are we gui yet” is definitely yes, at least on the desktop. The problem is that developers are too lazy to build apps with anything other than web frameworks.

Also shoutout to ratatui because even though it's technically a TUI not GUI, it's superb.

Strengths:

  - Message passing model with separate Model and View paths
  - Async / Sync landscape (sync on the GUI thread, tasks / subscriptions for async stuff with messages returned to the GUI thread)
  - Writing custom Widgets is quite easy! In 0.14 stateful widgets got a revamp and they are quite nice
  - Performance is great
  - Despite being very capable, the framework is not that large. Learning it is not a daunting task
  - Documentation might be sparse, but due to how it's written, I was able to just read the code and understand how it works without issue

  - Lots of changes between releases (still pre 1.0)
  - Theming. Despite being (somewhat) recently reworked, it's still overcomplicated imo
  - The layouting engine is tricky to use and I fight with it much more than I should. Quite often widgets do not show up at all or take too much space because I didn't use the right combination of `Length` variants for `width` and `height` of the widget (and/or its children)
  - Some interfaces seem a bit weird (I am specifically thinking about overlays)
  - There is some confusion regarding what should go in the application state and what should be held in Widget's state. The interfaces are clearly defined here, but what can be / should be done is often found out in practice (perhaps I lack experience here)

having SSR built in means that the UX is amazing - really complex pages load basically instantly, and it degrades just as easily for folks that haven't loaded the WASM. server functions were also fantastic to work with and easy to reason about. and, the hot reloading they built means that most UX changes are reloaded within a ~few seconds, meaning iterating is fast and (mostly) painless.

the native component library dioxus has (dioxus-primitives) is... sparse so i did have to build out a hundred or so basic components, but i've done that across various stacks 5-6 times now over my career so it's a fun little journey at this point. for the components they do provide, the quality bar is very good.

That’s crazy. It’s still better than the UX’s that game engine designers think work. Sure bro, show me your File Open/Save Dialog and tree view. Show me your text editor (nicely done, zed)

I still think the ideal solution for Desktop GUIs would be the Qt company developing first class Qt bindings for Node.js (or some other runtime), and allow people to build UIs using web tech with Qt components.

I wish there were more memory safe compiled languages that focused on ergonomics for cases like this.

For the small percent that's left where speed is critical, I've just been using wgpu and wgsl-bindgen directly.

I can't think of what I'd want a native UI solution for. And then having to deal with porting it to iOS, Android, Windows, Mac etc, dealing with app stores (3+?) submissions, developer fees, rejections... vs just using HTML which works on all platforms. I don't think I'd use a native UI even if there was one. Not that HTML is great by any stretch...

(I made Sunwet, this is what my stuff looks like: https://github.com/andrewbaxter/sunwet )

Ralph Levien, author of druid and xilem made some good posts about it. I'll link one here.

https://raphlinus.github.io/rust/gui/2022/07/15/next-dozen-g...

For desktop app development, if I don’t care about native controls (and what does that mean in Linux or even post-Win32, anyway?) and don’t want to deal with Electron, why not use something a bit more established like Avalonia or Flutter?

EGUI bonus: Good integration with WGPU, so you can show 3D things as part of your UI.

Complaining time: Historically, syncing winit, EGUI, WGPU, the binder between GPU and EGUI, and EGUI libs like for file dialogs has been a pain. It gives me anxiety thinking about upgrading versions. That said... the teams are sometimes shockingly fast about syncing their UIs. It is when winnit or WGPU etc make big breaking changes (Often from accumulation over time) where things get hairy!

https://github.com/microsoft/windows-rs/tree/master/crates/s...

It's a react-style API for WinUI3 apps, meaning it's not trying to mimic the Windows LAF. It's merely a binding.

It's nice to be able to use existing design systems and components, and to be able to validate in a web browser in quicker build loops before doing the full Tauri builds. I still manually QA across platforms pretty aggressively but Tauri's "cross-platform from day one" really isn't much of a stretch. The project if curious: https://github.com/zecrocks/zkv

I suspect if such a filter was applied there would be very few. Probably just the bindings to gtk, qt or appkit.

I'm building Hypen (https://hypen.space), a UI framework with a DSL that works in in Rust, TS, Go, Kotlin, Swift, and all over the place, as long as you can use WASM or binaries.

Some cool things about it:

- Renders natively on Desktop, Web (DOM and Canvas), Android and iOS.

- Streaming-first (SSR), so you can stream native apps from the server

- Custom tailwind parser so it supports your favorite shorthands

- Support for streaming apps from CF workers with 5 lines of JS/TS

- You can embed any Hypen app into another Hypen app like its an iframe, with just 1 line of code

- Has a custom "browser" for Hypen apps, both on desktop and on mobile, so you can easily check how your app looks anywhere

- Coming soon - stdlib and WASI interface to enable full WASM portability across platforms

Note: Desktop support is still a bit early and needs more crossplatform testing

I started building this years ago, first manually, now accelerating it with LLM's which are incredible for mindnumbing tasks like writing frameworks like these requires. Its still in an "early alpha" but it's getting closer to maturity and a "stable beta" by the day, hopefully fully stable 1.0 by end of the year.

Unfortunately, I had the expectation that it should be as simple as making an HTML page. My failure to find a library or a framework to make GUI application made me learn a lot about how GUI works. I realized that making GUI for browser and for desktop are quite different problems. Browser makes easy what's difficult having a desktop oriented GUI framework - text rendering. However, the situation is fair the other around. GUI framework makes easy what's difficult in a browser - drawing arbitrary shapes. As a result, a web-frontend programmer struggles to figure out how to write some text having something like Qt, a GUI programmer tries to find the API to the bitmap in a browser.

It's fair noticed in the previous comments that a GUI framework brings a lot. That's because the problem is complex:

1. Create a window

2. Communicate with the window compositor (you do in WinAPI too btw). How to access the system tray and the child window.

3. Communicate with the operating system.

4. Handle the user input. Callback vs event streams. The user has 4 keyboards for some reason.

5. Rendering. Subpixels, shapes, different DPI. The user has 6 monitors.

6. Text rendering.

7. Widgets. Where probably the most difficult part is to make a textbox, because it involves the solutions of all previous steps.

The steps above touch only the visual part. There's also audio, accessibility, somebody wants the GUI framework to solve the networking.

After all of this research, I picked simply SDL for my project.

1. It's easy to compile.

2. It's small.

3. It relies on the subjectively common dependencies.

4. It's fairly straightforward to upgrade. Given that, you have to create a lot from scratch the part with updating is smaller comparing to a Qt-based solution.

5. It has batteries. My favorite is SDL_ttf which allowed me recently to implement selection of the text which is quite a bit through towards a textbox.

Having a project on SDL requires a lot of knowledge, but not a lot of code.

Most of the time, I just want some UI. And TUI's are easier / more portable than GUI's.

* It's a genuinely hard problem

The state of GUI on Windows is a total mess, very disappointing considering the relative importance and resources that Microsoft has. SwiftUI was a success (Apple is uniquely good at this stuff), but the transition from Cocoa is still not complete. A large part of the reason why it's so hard is that the requirements for different apps vary widely; I think we are spoiled by the relative cohesion

* The infrastructure is in disrepair

A UI is built on top of lots of infrastructural pieces, obviously including graphics and compositing, but also text, accessibility, input handling, and many other things. When you look at support for those layers across platforms, it's like picking up a rock and seeing the bugs crawl. Take compositing: Windows and Mac have excellent compositors, Android and Linux much less so. But even there the features don't line up, Mac doesn't support incremental present (they don't really need to, considering how beefy their chips are, but it's really important on low end Android).

So a lot of the challenge is working around the brokenness across the various platforms. But, good news, there is really excellent progress on all these fronts. We now have solid text layout in Rust, a clear winner on accessibility, and even poor winit is making slow, steady progress.

I'd also like to say, by contrast the web has really been investing in infrastructure, and hides much of the lower level pain from developers. I think that's a major reason Electron is winning so much.

* The computer science is not quite done

We're still in an exploratory phase to figure out the best patterns for writing Rust UI. At the core of that is reactivity, which is at heart incremental computation. Most systems are converging on a hybrid of something generally React like (coarser grain reactivity) with fine-grained signals. I'd like to think there is a Right Answer here and that it's possible to find it.

This feels to me a bit like async. The foundations for async Rust were laid 10 years ago, but it still doesn't feel fully baked. I've seen some exciting recent work about improving the confusing Pin mechanism with something more principled. We'll get there, but it takes time.

I make three predictions:

* Rust UI will continue to improve, as the infrastructure is built out, we try more things, and projects mature. But not quickly.

* If we thought we had a lot of Rust UI projects before, wait til we see what AI wreaks. I predict we'll see dozens of vibe coded Rust UI toolkits.

* In the longer term, we're going to have Rust UI anyway, as browsers are increasingly going to be rewritten in Rust, and apps will be built on Web technology. And that's not just Electron, the modular browser approach pioneered by Blitz is also promising.

I started out looking for lightweight X-Windows solutions, for the LLM and me to use. Cutting a long and frustrating story short I settled on SDL2 Rust wrapper

But what a PITA X-Windows is (e.g. need Wayland for some things, but does not work well for others), and my goal is an appliance, not an app.

So in a moment of insight I spent an afternoon getting SDL2 to draw on the Framebuffer and dispensed with X entirely

I highly recommend this approach. Life is so much easier without X-Windows