> The entire VSCode workbench - editor, terminal, extensions, themes, keybindings — ported to run on a native shell.
but also
> Many workbench features are stubbed or partially implemented
So which is it? The README needs to be clearer about what is aspirational, what is done, and what is out of scope. Right now it just looks like an LLM soup.
> There’s a new kind of coding I call “hype coding”, where you fully give in to the hype, embrace exponentials, and forget that the product even exists.
Isn't this basically what was happening with most of our personal projects coming from ideas that we found very interesting and then forget that we ever thought about it?
I think it's like a lot of things - partially unnatural but it worked up enough people who hype it naturally. I know someone irl who's all into infodumping LLM jargon now and afaik has no monetary/investment incentive for it.
Honestly, at least 50% of it is natural. Personally I don't use AI, but its emergence was highly eye-opening for me.
I personally like both the process and the result in software development. Painstakingly designing things, writing them, and seeing that everything is working as it should be very smoothly, efficiently and fast. It's like building an engine by hand, tuning it, listening and feeling that it works as its best version within your capabilities. Then taking notes of the noises and inefficiencies and iterating upon them as the time allows.
Many people are not like that. They want an engine that works. Its efficiency, appearance or inherent reliability due to elegant design doesn't matter for them. If it works reasonably well, carries the builder from A to B (or more importantly makes money for the them), that's more than enough.
I personally respect this point of view, and completely understand it. What I can't respect or accept, regardless of how hard I try to, is being stoned to death or labeled as a Luddite because I and people with similar perspectives value a different set of things in their lives.
Mass produced things have their place, as well as artisanal, one-off ones. I believe we can live together, one doesn't need to kill the other.
I won't go into building of these AI systems, because I'm tired of repeating ethical and other concerns going into it, not because it's boring, but because people don't listen or care about them. Maybe I'll reiterate them when I have more time, next time.
This likely takes 500mb+ RAM, TFA probably didn't account for tauri://localhost in their calculation, which by itself takes 200mb+ RAM. Then your app process will take 100mb+ RAM, and there will be a couple of other processes besides.
Tauri is no better than electron in terms of RAM, just like people calling it "lightweight" are no better than flat earthers. Let's hope they come around.
and you know that AI wrote all of it with minimal human supervision.
side note: last few days I noticed that vscode stopped leaking memory all over the place. when left idle it was taking all the ram I had + 20gb of swap space
and recently I noticed that I have half of the ram free.
I use insiders build btw, so stable might still not have those improvements
I am not sure that their goal (<200mb ram) is really related to their approach (using a built in rendering engine for the same exact source material) at all.
I don't think it's worth it with the sheer amount of issues you will run into on Mac and Linux. There is a reason why lot of Tauri apps are being rewritten in Electron.
Awesome project. How much work is left, are you looking for contributors? I just built very basic IDE on mobile for myself. I thought about forking vscode but it felt too heavy and I only wanted a subset of features. Now I am wondering if I should use this project for the desktop version.
The worst thing about Tauri is people gaslighting about it being more efficient. No you just save on 150MB of download and install size. And you trade that for risk of incompabilities and breaking. The last thing I want for an editor I use 8h a day.
> Same architecture, native performance ... Goal: 200mb RAM usage.
But Tauri doesn't have native performance, specifically when it comes to RAM!
> Memory benchmark might be incorrect: Tauri might consume more RAM than Electron > https://github.com/tauri-apps/tauri/issues/5889
Quoting from the README:
> The entire VSCode workbench - editor, terminal, extensions, themes, keybindings — ported to run on a native shell.
but also
> Many workbench features are stubbed or partially implemented
So which is it? The README needs to be clearer about what is aspirational, what is done, and what is out of scope. Right now it just looks like an LLM soup.
The first sentence is aspirational, while the second is describing current state.
See I thought that, but then in the putatively aspirational section it says
> 5,600+ TypeScript files from VSCode's source, ported and adapted
which doesn't really make sense as a goal?
Vibe coded UI translation with a conflicting README and no screenshots? Is this what it takes to get onto HN front page?
It's completely vibe-coded with AI. That's hip. Works or not, fulfills its promises or not, that's not important.
It's vibe-coded, with AI, to Rust. That's enough. Ticks all boxes.
Just half-joking, BTW. Hype is hell of a drug.
Paraphrasing.
> There’s a new kind of coding I call “hype coding”, where you fully give in to the hype, embrace exponentials, and forget that the product even exists.
Isn't this basically what was happening with most of our personal projects coming from ideas that we found very interesting and then forget that we ever thought about it?
> to Rust
Not even to rust, really. This is still >95% typescript in a web view...
Maybe I'm being too conspiratorial, but I just can't believe that the feverish hype around AI on HN is completely natural.
I think it's like a lot of things - partially unnatural but it worked up enough people who hype it naturally. I know someone irl who's all into infodumping LLM jargon now and afaik has no monetary/investment incentive for it.
Honestly, at least 50% of it is natural. Personally I don't use AI, but its emergence was highly eye-opening for me.
I personally like both the process and the result in software development. Painstakingly designing things, writing them, and seeing that everything is working as it should be very smoothly, efficiently and fast. It's like building an engine by hand, tuning it, listening and feeling that it works as its best version within your capabilities. Then taking notes of the noises and inefficiencies and iterating upon them as the time allows.
Many people are not like that. They want an engine that works. Its efficiency, appearance or inherent reliability due to elegant design doesn't matter for them. If it works reasonably well, carries the builder from A to B (or more importantly makes money for the them), that's more than enough.
I personally respect this point of view, and completely understand it. What I can't respect or accept, regardless of how hard I try to, is being stoned to death or labeled as a Luddite because I and people with similar perspectives value a different set of things in their lives.
Mass produced things have their place, as well as artisanal, one-off ones. I believe we can live together, one doesn't need to kill the other.
I won't go into building of these AI systems, because I'm tired of repeating ethical and other concerns going into it, not because it's boring, but because people don't listen or care about them. Maybe I'll reiterate them when I have more time, next time.
The graveyard of LLM-generated comments from new accounts on the bottom of every thread supports your hypothesis.
Don't worry, at least it has a discord server linked.
It seems so, as it also describe every single “show HN” project that I've seen on the front page for the past 5 months or so.
What a world we live in where 200mb RAM usage for a text editor is considered "lightweight".
This likely takes 500mb+ RAM, TFA probably didn't account for tauri://localhost in their calculation, which by itself takes 200mb+ RAM. Then your app process will take 100mb+ RAM, and there will be a couple of other processes besides.
Tauri is no better than electron in terms of RAM, just like people calling it "lightweight" are no better than flat earthers. Let's hope they come around.
Tauri can't help with RAM. It's still running in a browser.
> Zero Electron imports remaining in the codebase
and you know that AI wrote all of it with minimal human supervision.
side note: last few days I noticed that vscode stopped leaking memory all over the place. when left idle it was taking all the ram I had + 20gb of swap space
and recently I noticed that I have half of the ram free.
I use insiders build btw, so stable might still not have those improvements
I am not sure that their goal (<200mb ram) is really related to their approach (using a built in rendering engine for the same exact source material) at all.
> Goal: 200mb RAM usage.
We've come such a long way from: "Eighty Megs and Constantly Swapping" :D
I don't think it's worth it with the sheer amount of issues you will run into on Mac and Linux. There is a reason why lot of Tauri apps are being rewritten in Electron.
Awesome project. How much work is left, are you looking for contributors? I just built very basic IDE on mobile for myself. I thought about forking vscode but it felt too heavy and I only wanted a subset of features. Now I am wondering if I should use this project for the desktop version.
My attitude toward vibe coding is quite open — as long as the quality of the software it produces, including the documentation, is not a problem.
Unfortunately, this is yet another example of a vibe-coded project that does have those problems.
> Goal: 200mb RAM usage
This is why Zed is great but I just can't get used to the debugger experience so I end up back in VSCode.
vscode debugger has always felt lacking to me; I still find myself firing up jetbrains* just for that
Would this also support the whole plugin infrastructure?
> In Progress / Unstable:
> - Extension host is early-stage — not all extensions will work
This is the kind of thing that looks simple until you actually implement it.
The worst thing about Tauri is people gaslighting about it being more efficient. No you just save on 150MB of download and install size. And you trade that for risk of incompabilities and breaking. The last thing I want for an editor I use 8h a day.
Good. I’m glad. It’s a great editor but the whole electron backend is so bloated.
What are you glad about exactly? This seems to be another big bang rewrite with LLMs. Fun but extremely unlikely to go anywhere.
Not op but there is a lot of details about VS code including electron that could have been done better.
VS Code is one of the most performant electron apps ever written. Extensions and plugins are always the culprit of poor performance.
That can be said about any project.
And VSCode as been improving since inception, hence why it ate a large pie of the market.