alt Hacker NewsZed editor switching graphics lib from blade to wgpu
Comments
I am confused by this without context. I have not heard of blade, but am aware that Zed built its own GUI library called GPUI. Having used Zed, this is a vote of confidence: The crate ecosystem is historically filled with libaries which try to be The future of X in rust but are disconnected from practical applications. GPUI by nature is not that; it's a UI lib built to a practical and non-trivial purpose. It sounds like Blade is a cross-API graphics engine, by one of the original gfx-HAL (former QGPU name) creators?
I have not used GPUI beyond a simple test case, but had (prior to this news?) considered it for future projects. I am proficient with, and love EGUI and WGPU. (The latter for 3D). I have written a library (`graphics` crate) which integrates the two, and I use for my own scientific applications which have both 2D and 3D applications. Overall, I'm confused by this, as I was looking forward to using GPUI in future applications and comparing it to EGUI. I have asked online in several places for someone to compare who's used both, but I believe this to be a small pool.
I was not sure of the integration between GPUI and WGPU, which can confirm EGUI and WGPU have great integration. But I only care about this because I do 3D stuff; if I were not, I would be using eframe instead of WGPU as the backend.
Unrelated, off-topic, but I'm also not sure where to ask this: Am I missing something about Zed? I have tried and failed to get into it. I really want to like it because it's so fast [responsive], but it seems to lack basic IDE functionality in Python and Rust, like moving structs/functions, catching errors dynamically, introspection and refactoring in general etc. I thought I might be missing some config, but now lean that it's more of a project-oriented text editor than true IDE in the fashion of JetBrains. But I have been unable to get a confirmation, and people discuss it as if it's an IDE or JB alternative.
Rust GUI is in a tough spot right now with critical dependencies under-staffed and lots of projects half implemented. I think the advent of LLMs has been timed perfectly to set the ecosystem back for a few more years. I wrote about it, and how it affected our development yesterday: https://tritium.legal/blog/desktop
Zed is my goto editor when I'm not vibe coding, but that is rare these days. Their integration with Claude Code, etc really helped, but Antigravity completely pulled me away. And really, since they're catering to the same basic audience, the defaults should be the same as VSCode for most stuff. VSCode but performant would be an excellent pitch for the upcoming consumer RAM deficient world.
Dunno how they plan to get wider extensibility and community support without an embedded JS backend to support the existing Code plugins. That's where the real blocker is.
Switched from Intellij (various) to Cursor because of AI integration, only using Claude Code CLI, switched to VS because Cursor became so annoying every release, pushing their agents down my throat, activating what I did deactivate every release, recently thought "Why do I even use that slow bloated thing of VS?" and switched to Zed. Very happy camper. So much faster. So much snappier. Would love Claude Code CLI integration but can live without it. Would pay for Zed as I did pay ~25y for Intellij.
An interesting side effect of moving to wgpu is that in theory with some additional work, this could allow you to run Zed in a web browser similarly to how some folks run VSCode as a remote interface to the backend running on a server.
I find it odd the rust community feels the need to reimplement tried and tested APIs in "pure safe Rust". Like no other language has better C integration, and we have had cross-platform windowing libraries since like the 90's, why does everyone reach for a brand new unstable libraries with less maintainer support?
Edit: replying to https://tritium.legal/blog/desktop, not the OP
Zed seems to be already suffering from heavy technical debt. From my perspective, as a game dev, their stack is a lot heavier than it should be.
In 2020, I started working on a (C++) game engine. Since the only decent open-source UI option was Dear ImGui (which was obviously a bad choice for consumer-facing UIs), I ended up rolling my own retained-mode UI library on top of SDL. Now, it's fully-featured enough that I rarely have to touch it. There's even a major company using it for embedded products.
I don't get why every language's community doesn't just do the same thing: roll an idiomatic UI lib on top of SDL. It was tough, but I was able to do it as a single person (who was also building an entire game engine at the same time) over the course of a couple years.
I hope this can somehow improve the font situation. Even on a 1440p monitor, the fonts in Zed are much blurrier than any other editor I've used. I Can't even use bitmap fonts like VSCode.
Clean change, all things considered. But, like, what? What's to say you won't run into a billion problems here, too?
Will this help running Zed in environments with no GPU/old GPUs? There have been some complaints about not being able to run Zed on Ivy Bridge or in VMs, even though browsers and other applications work perfectly fine
(for the linux renderer only)
A bit of a tangent, but I wish the makepad project would get more traction in rust. Their cross-platform approach is extremely ambitious.
Oh sweet! I threw out GPUI completely from one of my projects because of Blade. I needed headless with rendering to image for e2e testing and gave up on GPUI after trying to mess with Blade. It’s definitely a mess and moving to egui has only shuffled the deck chairs around.
Will this make Zed slower, since Wgpu is just a compatibility layer, adding more code?
Oh, this is nice. Latest builds stopped working on panfrost because it does not announce the sufrace capabilities or something like that. Maybe I can have it back to working on the orange pi
Is webgpu a good standard at this point? I am learning vulkan atm and 1.3 is significantly different to the previous APIs, and apparently webgpu is closer in behavior to 1.0. I am by no means an authority on the topic, I just see a lack of interest in targeting webgpu from people in game engines and scientific computing.
Seeing that the author of Blade (kvark) isn't exactly a 3D API newbie and also worked on WebGPU I really wonder if a switch to wgpu will actually have the desired long term effect. A WebGPU implementation isn't exactly slim either, especially when all is needed is just a very small 3D API wrapper specialized for text rendering.
Can someone, who knows computer graphics, explain why the old library had so many issues with flickering and black triangles or rectangles flashing on the app, and why the new library is expected to not have those same problems?
Hopefully this will get momentum scrolling working.
Zed ignoring local LLM makes me to sad to worry about the rendering. Not sure why it needs acceleration but willing to learn!
wgpu's OpenGL support is kind of broken. wgpu + Vulkan is the stable combination, unless I am mistaken
[dead]
The Zed editor seems kind of silly to me. I would rather my editor works in many possible environments maybe even one that only has a tty interface.
What advantages are people finding with this editor other then high fidelity scrolling.
Zed also stopped GPUI (their GPU accelerated Rust UI framework) development for now, sadly.
> Hey y'all, GPUI develoment is getting some major brakes put on it. We gotta focus on some business relevant work in 2026, and so I'm going to be pushing off anything that isn't directly related to Zed's use case from now on. However, Nate, former employee #1 at Zed, has started a little side repo that people can keep iterating on if they're interested: https://github.com/gpui-ce/gpui-ce. I'm also a maintainer on that one, and would like to try to help maintain it off of work hours. But I'm not sure how much I'll be able to commit to this
https://discord.com/channels/869392257814519848/144044062864...