Comment by Jarred
10 days ago
I work on Bun and this is my branch
This whole thread is an overreaction. 302 comments about code that does not work. We haven’t committed to rewriting. There’s a very high chance all this code gets thrown out completely.
I’m curious to see what a working version of this looks, what it feels like, how it performs and if/how hard it’d be to get it to pass Bun’s test suite and be maintainable. I’d like to be able to compare a viable Rust version and a Zig version side by side.
It is a pity that you can't make an experimental commit on an experimental branch without igniting a fire of delirium through some people who -- if they were able to put their emotional response aside for a minute and could weigh this up on the basis of merit -- would probably agree with the motivations for researching this approach.
> if/how hard it’d be to get it to pass Bun’s test suite and be maintainable
Every month brings new opportunities to completely abstract the process of porting code with agents, all using linguistics. What an exciting time.
For those looking for a similarly interesting (and interestingly similar) example, see Cloudflare's port of Next.js[0], "vinext", from a couple of months ago. It had some teething problems at the start but I'm using it in a few production projects now with minimal issues.
[0] - https://github.com/cloudflare/vinext
This is what it means to work on a popular project, unfortunately.
You also don't have a duty to read or respond to the social media flames. Just do the work you want to do.
If people get worked up about experimentation, that's their problem, not yours.
5 replies →
I am a topic starter, and I had no emotional response, was just being curious. Never expected it will land at HN #1. I specifically posted the link to the first commit and not to the whole branch, because currently the prompt is the most interesting part.
The title kinda set the tone for this post.
13 replies →
An original topic starter? I'm pretty sure that this was originally posted on X by someone else, as I commented there, and minutes after, it was copied and put here on HN with the twisted title; the original was more of a "question, surprise tone"
1 reply →
It’s annoying for the team members I suppose, but to be fair, if you’re working on a high-profile open source project, owned by one of the most hyped companies in the world, and your branches are public, it’s probably a good idea to be clear in the branch naming and supplemental files if you’re just “experimenting”.
By working in public on a popular open source project, you are communicating intent and purpose to your users and the general public through your commit messages, branch names, and documentation. You’ll save yourself a lot of grief if you act accordingly.
The fact someone who works on Bun is willing to create and even push a branch generated by a stochastic parrot is very telling of the direction the project is going.
Doesn't matter if it's "experimental", it's a dumb experiment that shouldn't exist.
Doesn't matter if it's "experimental", it's a dumb experiment that shouldn't exist.
Do you think the same about bitcoin? Where do you draw the line as to what programs are allowed to be written?
Why are you treating branches as if they are holy? This is all OSS, people work on this in their free time, git is got and people can use branches as they like to experiment and share their experiments with others. If you don't like the code, don't use it you damn leech.
Underplaying AI, overselling what an experimental branch is, and suggesting it's representative of the entire project, all while suggesting people shouldn't even consider new tools and methodologies. Where to start.
That's not a very constructive, nor accurate, way of trying to dismiss all concerns around bun that has been raised.
I think that was a very constructive comment about the unconstructive way people are shoe-horning other concerns about bun into this thread abut a specific aspect which itself turns out to be just an experiment that someone knee-jerk reacted to, despite several active threads already discussing those matters one of which only just fell off the front page.
While the concerns many have about Bun's potential future direction are valid IMO, of the posts on this thread the one you are criticising is one of the more constructive.
1 reply →
I love your work on bun. How do you feel about all the constant concerns being raised about the quality of the project lately? I understand some of them might just be typical twitter hate but some of them are real. And I think people are right to question why you are adding image processing or web views inside a javascript runtime when there are bugs affecting production that sit unaddressed. For example on of our biggest blockers right now is https://github.com/oven-sh/bun/issues/6608 which was reported in 2023, still affecting us 3 years later.
When you start getting hate, you’ve made it. Up until then you’re a hypothetical that people like. Maybe they’ve built a side project with you or read the docs. You only get hate when people have used your tool and butted up against limitations. We saw this with Deno too where they went from beloved potential savior to realistic, limited tool. Hate is good. It means people rely on you
Do you know which project gets the most hate? Nodejs, so in that sense, Nodejs has made it and it is widely deployed but this hate was the reason that two seperate alternatives for Node have emerged as Deno and bun.
Recently Bun's latest version had memory leaks which crashed production code from my understanding and their attitude[0] of saying OSS will have no human contribution allowed, now doing these ports of zig to rust, going back for years what the decision making of using zig was and this code basically being vibed as there is no way that they are reviewing the code while being VC funded/bought by anthropic.
These are all genuine issues which cause hate. You can say people are hating because people rely on it but the true thing is that also seems like a bait and switch and that people switched from node.js to bun (maybe even being locked inside bun), only for them to do these highly questionable decisions which is the reason why people are starting to hate on bun.
Atleast that's my interpretation right now reading this whole thread.
[0]:https://x.com/jarredsumner/status/2048434628248359284: "I expect OSS to go the opposite direction: no human contribution allowed. Slop will be a nostalgic relic of 2025 & 2026."
- Jarred Sumner
7 replies →
Okay, let's be honest. That's a feature request, not a bug report.
I'd agree but bun is supposed to be a "drop-in replacement" and is marketed as such. This breaks several packages and projects.
1 reply →
Why not offer a bounty to get this issue fixed? Are you otherwise paying any money to the bun team?
This is getting stupid. Now one can’t even make a reasonable polite question with praise without being asked if they pay.
Bun raised millions of dollars and was acquired by a commercial entity which bragged in the same blog post of reaching $1B. They’re not a guy with an eyepatch and a tin can out on the street.
Open-source developers should be compensated, but they don’t have to be. You can’t reasonably offer your work for free then complain someone isn’t paying you. If you want to be paid, charge for it.
Signed: A long time open-source developer who has dedicated years of full-time work to useful projects without compensation or raising VC money or being acquired.
3 replies →
I think the question still deserves a proper answer.
10 replies →
What's the main motivation for considering Rust?
For what it's worth, in my last experience with Bun[0] I ran into a couple of bugs where it seemed Rust could have helped, e.g. using Bun.write
[0]: https://mastrojs.github.io/blog/2025-10-29-what-struggled-wi...)
With AI agents and how good they are in doing "language translation" tasks against an identical target with a comprehensive test suite, you end up doing these things out of curiosity. The AI agent has the originals to test it's assumptions with too.
I've had surprisingly good results from getting AI agents to take a script in shell, python or typescript and have it translate it into those other programming languages, including rust versions. Or swapping from one build system to another.
Totally agreed... It enables you to try swapping out dependencies you might not otherwise even consider because of the cognitive load in trying to do so as an individual, and get it done/working in a few hours and a few days to follow in order to review.
Or take on an additional/related feature (like Redis grepping over the new array data types). Because you can be relatively sure the borders are stable and you can limit the surface/scope.
Thank you for the clarification!
While you are here, can you elaborate on the method chosen? For example, why not write a conversion script for phase A? I mean, same Anthropic model will produce it in no time, prompting it is at the same cognitive load level, but you would have a deterministic result.
Thank you, Jarred, for your work. It’s unfortunate to see so much backlash toward legitimate research. Bun is often seen by some as “the flagship project for zig” - especially among those frustrated with rust who want zig to "win over rust" for whatever reasons. At the end of the day, you should do what makes the most sense for your project and your circumstances, regardless of the language or tools involved.
Personally, I find this experiment interesting and I’m curious to see how it develops. Writing idiomatic rust requires a shift in mindset, so it’ll be worth watching how well LLMs adapt to that over time.
I can only speak for myself... but I've found at least Claude Opus to handle Rust very well, and in my own use cases WebAssembly (wasm) and FFI for interoperation with TS/JS has been pretty smooth.
>who want zig to "win over rust" for whatever reasons
I don't understand why this mentality is so common. Zig and Rust are both fine languages with markedly different design goals and they can coexist.
Honestly, I don't know. I think it's because of frustration, but the community attitude is part of it. I experienced first hand people frustrated with Rust moving to Zig and finding other people to pick onto Rust and finding fertile ground (especially if moderators and heads of the community let this kind of behavior continue).
....you were saying?
You can view it as an overreaction, but also as a sign that your work is significant. It impressed some, and scared others. In any case, you made something interesting.
You're replying to the original author of Bun. Given the usage of Bun, and the fact that his company (primarily him, actually) was recently acquired by Anthropic for what I'm guessing was a bajillion dollars, I think he probably already knows his work is significant and that he made something interesting.
Lol! My bad, I wasn't aware it was the original author (my fault for replying with too little reading). In any case, I think what I said still applies to his LLM experiment.
Calm and curious about your results.
I hope you get the code elegant and not only maintainable but future friendly and performant.
I'm very curious what Zig vs Rust code looks like for the same project! What are your thoughts so far?
Might be a good idea to let AI handle social media. I'm not saying you're doing it badly, just that it doesn't seem like worth the drained energy to do manually.
Can't think of a more stupid and detrimental way to use AI. Pretending to be a (particular) human on social media.
the is lovely, how admirable that you have the space to do this. its very rare that we as a community take the time to actually implement a non trivial system in X and Y and look at the differences. so much discussion around these things is based on pointless tribalism.
I'm sure recasting Bun in a new mold is going to be hugely informative about the structure of Bun itself, regardless of the outcome.
would love to read a postmortem
A research prototype. This is normal.
[dead]
[flagged]
Advice for the future: experiments should be explicitly tagged as such. The commit message "docs: add Phase-A porting guide" says nothing about the experimental and looks like a planned move to rust. That message certainly looks very official to me.
> This whole thread is an overreaction. 302 comments about code that does not work. We haven’t committed to rewriting. There’s a very high chance all this code gets thrown out completely.
Trying to pass off a blunder like this like its no big deal is an insult to your users. You made a dumb mistake. Own it, be transparent and correct the problem that started this; namely, put some form of experimental tag in the commit message. Then say you made a simple mistake, sorry, and move on. Being dismissive is a defense mechanism that can arouse suspicion, as in are you now lying about the experimental state to quench the flame war? Not that I believe that but it can certainly now become conspiracy. Again, you can avoid all that with transparency.
Or the community at large could stop acting deranged over language wars like it’s 2001.
It’s their repo, let them do what they want lol
I didn't get the impression that anyone cares about the source or destination language. I think the concern is centered around the long history of failure with large scale rewrites like this-- See Netscape 5, Perl 5, etc. Joel Spolsky wrote a legendary article about this [0]. I think the NextJS app router might be slowly joining this conversation as well.
It could get even worse if they get Second System Syndrome[1] and try to add features as they rewrite it. Considering Bun's rapid development cycle, this seems likely.
[0] https://www.joelonsoftware.com/2000/04/06/things-you-should-...
[1] https://en.wikipedia.org/wiki/Second-system_effect
Or we can stop being toxic to open source maintainers and acting like we own them or they owe us anything.
A commit message on a random branch is not an obligation. Not telling random internet users what side projects they're working on is not a blunder. It quite frankly doesn't matter what you think looks official, it doesn't give you the right to treat people like this.
It's so embarrassing to be a programmer some times, so many of my peers behaving like spoiled rotten brats.
> Or we can stop being toxic to open source maintainers and acting like we own them or they owe us anything.
The majority of the community feels this way which says something. The author's reaction is to publicly display being upset and dismissive of the communities reaction. That is just making it worse.
When you work on a project this big, more care is needed. The commit was an innocent mistake. The blunder is blowing off the communities response as overblown which it would be had the commit been tagged experimental. But it wasn't. And the author did themselves no favor blowing it off.
If the author was smart, their reply would simply have been:
Hello, To clarify, this is an experimental branch only. There are no plans to port, only experiment. I will tag the repo as such to ensure people understand its intention and avoid future misunderstandings.
Nothing difficult to understand here.
1 reply →
> We haven’t committed to rewriting. There’s a very high chance all this code gets thrown out completely.
Props for the effort man, but people have already picked up on Zig-to-Rust transition.
Poor Zig folks ...
More like poor Bun
Hoping that an AI rewrite is thrown out.
You may even be an OK programmer, but IF YOU AREN'T ABLE TO DO THE WORK I DON'T WANT TO USE IT.
Not worth your time? Not worth my time.
Most of Bun’s code is already written by LLMs. If you feel that way, it’s already been too late for a while. Furthermore, we’re talking about a million line port done in a couple of days. The question of whether it’s worth the time looks extremely different if done by hand. It would take a year.
The "too late" argument isn't gonna fly with someone like me who has both the time and energy to own a Javascript runtime. Heck, I'm quickly becoming the most prolific author of the ES spec too.
4 replies →
I think the criticism is still a valid to an extent because I don't see how this would give you a good way to evaluate Zig vs. Rust. Maybe a better approach is to migrate a particularly problematic space and bench that on its own?
It's not like OP asked for any criticism to start with, right? This whole thread is pretty good example of why saying "Fools and children should never see half-finished work" exists. ¯\_(ツ)_/¯
Since when was HN ever about asked for criticism?
2 replies →
Will you have a way to measure the ecological impact it has to make such a throw away attempt?
Not actually pointing on you or anyone in particular here to be clear. And if the answer would be "not much more than forgetting the light when leaving the toilets", certainly that would be a "go have fun" cheerleading on my part.
But otherwise we collectively have to keep in mind that the prompt that we can throw mindlessly and without perceiving any direct negative feedback are possibly not anodyne.
So if you can measure it, come back also with these numbers so we can all take that into consideration next time the thrill to run it just to see what happens rise in our mind. Thanks.
Right now it seems to say:
> Showing 1,808 changed files with 790,916 additions and 151 deletions.
Just looking at the git diff [0].
I looked at one of these rust port files [1]. Its 827 loc and apparently 7,576 tokens. So that gives you a first order guess that the full 700k additions is around 8 million output tokens. Obviously there are some tool calls, reasoning, reads of the zig version, and fixing compile errors as overhead. So I would guess maybe this is like 40 million tokens by multiplying by 5?
If we guess that is around $200 to $500 in token spend. We can probably guess that it emits around the same as buying $100 in gas? Or like 50 or so kgs of CO2?
[0] https://github.com/oven-sh/bun/compare/main...claude/phase-a...
[1] https://github.com/oven-sh/bun/blob/dacc59c62a8f93eabe6d9998...
Thanks, that's a really great answer.
It feels odd that the same message can be thus down voted and give the impulse to provide courteous response with reasoning, metrics and values.
Glory to your kindness and informarive way to react.
Less than the impact of people who can't be bothered to remember basic historical facts or directions in terms of hitting Google services dozens of times a day across the population.
Probably less than the impact of having dozens/hundreds of actual developers, each with a dedicated computer running for months/years in what it would take for a similar effort.
If you want to go live in the woods and farm/hunt for yourself, feel free. I'd suggest you stay away from the museums with paint and not glue yourself to a car mfg.
Isn't b it same to emit doubts that the resources required to find and access webpages or use a GPS is causing at scale the same ecological impact as everyone selling the world by the token. Though that might be wrong, of course, then doubts should be addressed with proper reasoning, not aggressive rejection which would be a call to run with a blind fold.
Having people working together at some goal is not not going to create the same social structures as running LLMs at the same goal. That's missing the ecosocietal forest for the digital output.
Actually, at societal level, no, people are not free to go into gather and hunt mode, that is not at scale. Sure some individual can do it on the margin, but by definition that won't make the mainstream societal impact disappear.
2 replies →