It's kind of an exploratory phase for what works sensibly with Rust's borrow checker, especially since most UI libraries/frameworks really rely on a GC.
The most popular language by far used for writing UI is JavaScript, and the go to framework this day (React) doesn't use that though.
From the various experiments that popped up over the years, it's pretty clear that the React way works pretty well for Rust, but it's also too slow to be desirable for Rust (what's the point of using Rust for UI if you're going to have web-like performance).
And then again, making a half decent UI framework is a gigantic task, there's just not a whole lot of languages with a decent UI story at all, no matter what's the paradigm of the programming language. (And if you want a
language for cross-platform UI, I'd argue that the only one that ticks the box is JS with React in Electron and React Native, and it's not even truly a single framework).
The view part would be fine, the problem is updating the state. In a language which discourages shared mutability, most of the solutions are not terribly ergonomic.
You either end up needing to:
- handle all your state via interior mutability (i.e. Arc<RefCell<_>>)
- use a reducer (i.e. the state blob is immutable during rendering, updates are deferred via events that are delivered between frames)
- or invert the relationship between state and view (i.e. immediate-mode) which comes with it's own implementation challenges (caching immediate mode views is hard)
> Unlike common object-oriented GUI frameworks, Ribir widgets do not need to inherit a base class or hold a base object. It is a pure composition model
I'm really not sure how this "composition" is any different to the usual inheritance you see in frameworks like QML *in practice*.
This may not be what you're after, but note that egui and Slint have accessibility support (at differing levels of completeness), e.g. for blind people using screen readers, while Ribir and GPUI do not.
I might still got ptsd from a job where literally all of the rust codebase was written as macros. Since then I avoid them at all costs.
I find the route, that gleam took, way more elegant with squirrel (sqlx-ish) and lustre (elm-like) being examples of what we could have instead. Avoiding language mixing is so important for proper/clean lsp-support - yet macros are a different language as i see it.
As for the rest of this: i also don't see how it's any different from iced, egui etc. but maybe I didn't take the time to check the details...
From the various experiments that popped up over the years, it's pretty clear that the React way works pretty well for Rust, but it's also too slow to be desirable for Rust (what's the point of using Rust for UI if you're going to have web-like performance).
And then again, making a half decent UI framework is a gigantic task, there's just not a whole lot of languages with a decent UI story at all, no matter what's the paradigm of the programming language. (And if you want a language for cross-platform UI, I'd argue that the only one that ticks the box is JS with React in Electron and React Native, and it's not even truly a single framework).
You either end up needing to:
- handle all your state via interior mutability (i.e. Arc<RefCell<_>>)
- use a reducer (i.e. the state blob is immutable during rendering, updates are deferred via events that are delivered between frames)
- or invert the relationship between state and view (i.e. immediate-mode) which comes with it's own implementation challenges (caching immediate mode views is hard)
This is how I implemented my last Angular project, works fine for non-trivial tasks.
I'm really not sure how this "composition" is any different to the usual inheritance you see in frameworks like QML *in practice*.
This in Ribir:
```
Column {
}```
Would be this in QML:
```
ColumnLayout {
}```
I like the idea of using macros to clean syntax; am writing some for EGUI right now to make colored text easier.
Yet both examples use macros.
I might still got ptsd from a job where literally all of the rust codebase was written as macros. Since then I avoid them at all costs.
I find the route, that gleam took, way more elegant with squirrel (sqlx-ish) and lustre (elm-like) being examples of what we could have instead. Avoiding language mixing is so important for proper/clean lsp-support - yet macros are a different language as i see it.
As for the rest of this: i also don't see how it's any different from iced, egui etc. but maybe I didn't take the time to check the details...