I think overusing AI is bad always 100% of the time and honestly what the Bun people did was insane but I think this is just a case of a language author being salty people are dropping their project.
Zig is basically C with a few more bells and whistles, of course you’re going to fall into many of the pitfalls that C has. Yelling “you’re holding it wrong” is not a convincing argument in 2026 when languages with much better design choices exist.
When Jarred announced the Rust rewrite, we were ecstatic. It seemed too good to be true. I have to admit, I didn’t think the technology was there, to pull off this stunt. But he did it, and now I’m metaphorically sipping delicious tea from a mug that says “It Tastes Like It’s Not My Problem Anymore”.
I know this could read like a salty breakup, but… I kinda think he’s being genuine here. The whole “History” section lays out a believable case for why they weren’t happy with Bun from the get go.
I’ve also never heard of any of this before today, so maybe I’m missing something.
They were maintaining their own fork of the Zig compiler, and I’m sure that the fact that it was clearly not well written caused him headaches.
They’d try and upstream stuff and he’d have to reject it, people would use Bun stuff and then submit issues to Zig that don’t apply etc.
Zig is a BDFL language and that’s fine. You don’t need languages to be designed by committee (especially when they’re young, I think the Jai approach is actually kinda neat too).
Python is great, had a BDFL, and now it’s kinda wheel spinning as the community tries to shift away from those decisions made 30 years ago, and that friction ends up causing the language to have all sorts of weird nuiances.
That sort of organic development is another reason I don’t see AI being very useful long term. Retraining once every few months for major language releases is absolutely untenable and you’ll always be a few years behind since there isn’t code for you to steal yet.
I think the simpler answer is that building your product on rapidly evolving language and toolchain that is multiple years away from 1.0 was a bad idea. Vibecoding your own answers to language semantics still under review also sounds like a shitty idea.
Andrew Kelly’s response comes out like hes relieved that Bun is gone than salty, and I would have the same reaction considering how shitty and bad anthropic is to be associated with in any way or form.
Well if you read through the article he notes how what bun devs say is inconsistent. One one hand they complain about being verbose, on the other they say all their code is LLM written. The LLM is apparently really great at translating huge amounts of code, but can’t keep track of memory allocations. They present the whole thing as a pure win rather than honestly explaining trade offs such as compile times. They never talk about how they arrived at decisions or what options they evaluated. I think the article is absolutely correct that this is a PR stunt by Anthropic. Rust might be a better choice than Zig, but it’s pretty clear the decision was driven by marketing here.
I think overusing AI is bad always 100% of the time and honestly what the Bun people did was insane but I think this is just a case of a language author being salty people are dropping their project.
Zig is basically C with a few more bells and whistles, of course you’re going to fall into many of the pitfalls that C has. Yelling “you’re holding it wrong” is not a convincing argument in 2026 when languages with much better design choices exist.
https://andrewkelley.me/post/my-thoughts-bun-rust-rewrite.html
I know this could read like a salty breakup, but… I kinda think he’s being genuine here. The whole “History” section lays out a believable case for why they weren’t happy with Bun from the get go.
I’ve also never heard of any of this before today, so maybe I’m missing something.
They were maintaining their own fork of the Zig compiler, and I’m sure that the fact that it was clearly not well written caused him headaches.
They’d try and upstream stuff and he’d have to reject it, people would use Bun stuff and then submit issues to Zig that don’t apply etc.
Zig is a BDFL language and that’s fine. You don’t need languages to be designed by committee (especially when they’re young, I think the Jai approach is actually kinda neat too).
Python is great, had a BDFL, and now it’s kinda wheel spinning as the community tries to shift away from those decisions made 30 years ago, and that friction ends up causing the language to have all sorts of weird nuiances.
That sort of organic development is another reason I don’t see AI being very useful long term. Retraining once every few months for major language releases is absolutely untenable and you’ll always be a few years behind since there isn’t code for you to steal yet.
I think the simpler answer is that building your product on rapidly evolving language and toolchain that is multiple years away from 1.0 was a bad idea. Vibecoding your own answers to language semantics still under review also sounds like a shitty idea.
Andrew Kelly’s response comes out like hes relieved that Bun is gone than salty, and I would have the same reaction considering how shitty and bad anthropic is to be associated with in any way or form.
Well if you read through the article he notes how what bun devs say is inconsistent. One one hand they complain about being verbose, on the other they say all their code is LLM written. The LLM is apparently really great at translating huge amounts of code, but can’t keep track of memory allocations. They present the whole thing as a pure win rather than honestly explaining trade offs such as compile times. They never talk about how they arrived at decisions or what options they evaluated. I think the article is absolutely correct that this is a PR stunt by Anthropic. Rust might be a better choice than Zig, but it’s pretty clear the decision was driven by marketing here.