> My idea is to strip the system down as much as possible and leave only what is required for production.
> Development would be done using the full Bun runtime, while production would use its lightweight fork, Cruller. I do not have the resources of the Oven team to develop and maintain a massive general-purpose runtime, so I want to focus on specific production requirements.
I don't know about others, but I don't think I would deviate my development runtime from my production runtime so significantly. The chance for behavior that only rears its head in production is too high for my liking.
I don't understand why the git history has been pruned in this fork.
According to the first commit in this fork:
> Squashed as a single orphan commit — the original oven-sh/bun history
isn't relevant to this stripped fork and its shallow clone doesn't
push cleanly to a fresh remote.
IMHO it's always a bad decision to do that. Here, all commit authors are lost. The original history is always relevant.
Not a fork, but a new project that takes a subset of the old Bun code to create something new, meant to complement Bun.
“Cruller is not intended to replace Bun for development. It is a minimal, specialized runtime for executing production code. In any case, I do not want to throw away such a large codebase that has taken several years to build. It makes more sense to turn it into a convenient embeddable library that can be used throughout the Zig ecosystem.”
If I'm taking a project in a completely different direction, with different people, I might not care about commit history very much. But I worry more from a licensing standpoint. Unless a project has a strict copyright re-assignment policy (uncommon) where all contributors sign over their copyright (not just contribute their "patch" under the designated license), the copyrYight history is lost or at best muddled.
Now, in practical terms, one can always go back to the original repo and find it. But if I was doing this in a corporate $DAYJOB, I'd keep the history just to be sure.
Because there will be a reason why the code was changed that introduced the bug. If you fix the bug you might inadvertently break a different piece of functionality.
A git commit gives you the reason for the change and the context of what other files were changed at the same time. I’ve found that invaluable.
> A git commit [ideally] gives you the reason for the change
FTFY ;) I broadly agree, but it's important to note that, in practice, Git commits are hit and miss: they might save you a lot of time if they are well-structured and described properly, but they might also completely fail to give you anything of value (for the problem you're facing at the moment). Turning Git history into a broadly useful artifact requires a lot of discipline from everyone working on a codebase over its lifetime.
It's never totally useless, but it's often less helpful than it could be.
Recently, I turned to JJ for this exact reason: crafting a meaningful change history is much easier there, especially if you need to retrofit it after you're done with the implementation. jj's split, squash, and describe are a joy to work with, also because the interactive version lets you easily split or squash hunks (selected changes in the same file).
You can of course choose alternate strategies. It comes up a lot when doing low level systems programming because the bugs can be non-trivial to repro and root cause (particularly for C/C++ where you have segfaults and whatnot). So it’s cheaper and faster to try a bisect to find the regression introduction as an additional clue in the root cause analysis (eg ownership rules changed or there were changes that looked correct regarding threading but when you hyper focus on it you realize what the problem is). Or you’re working on a high performance codebase and there was a performance regression - it can be faster to bisect than it is to try to find the issue with profiling tools (I had this and seen issues where profiling tools would never have told you the issue).
I'm just going to answer this one candidly. Yes, you are doing something wrong. Immediately: the commit history is useful to understand the context in which a bug was introduced, because it's the best context to understand the _reason_ the mistake was maybe, which is a really good way to prevent regressions. But more generally: when an entire industry has arrived at an accepted practice, you the junior developer (sorry, 5 years is _nothing_: you need to be told this too) have a chance to pick up on it.
Have you heard "have you played guitar for 20 years, or have you played guitar 20 times for a year?" Time only helps with accumulating wisdom if you let it.
I would ignore it if I spotted it once. This is the third time, though, so I probably should point it out: 5 years is a very short time in terms of personal experience and development. While the whole industry moves very fast, it does so thanks to slow, grinding effort parallelized over hundreds of thousands of individuals. For an organization, 5 years is an era. For individuals, it's just 3-6 projects (+/- a few, depending on the context you work in).
TL;DR: "never have encountered in 5 years" is a meaningless data point. My professional career can legally drink beer since a few years back, yet I'm still regularly hitting "first times" on various things in my day job; I'm also frequently mind-blown reading about others' experiences in domains I never touched. Don't make those "5 years" into a badge to hide behind; use it as a springboard to reach higher.
Concrete example: yabridge is a tool to let Linux users run DAW plugins designed for Windows. Some changes to Wine (yabridge's main dependency) made yabridge stop working in 2024, and those Wine changes are not going to be undone as they advance progress on Wine's own bug list and test suite. When those changes were made, there wasn't a clear way to change the yabridge code to work with the modified Wine functions.
At the very least, if Wine and yabridge didn't keep a history of their code, you wouldn't even be able to download and build Wine 9.21 staging, when yabridge worked.
Every few weeks, a new dev will find this problem and look into fixing it. By having a history, this dev can always look back on the working implementation. Maybe one day there will be a series of function calls in the new Wine codebase that gets the same behavior that worked before? Maybe Wine adds a flag for yabridge that reintroduces the old functionality but retains for everyone else the changes they decided were necessary. Both avenues would fix the bug but from different ends.
(I understand Wine avoiding the flag solution: It's more to maintain just to help a single project, plus it introduces a conditional jump fail in the very active event handler and windowing subsystems affecting nearly every Wine user, and the development of Wine keeps moving, which could make a flag-based workaround redundant in the near future.)
I haven't opened my DAW in a few months. Last I knew, the problem is resolved for most users but not all, and there might have been a major breakthrough in April or May that still hasn't reached end users.
I don’t - Windows keeps a compatibility database for a reason. It seems inevitable that you’d need to maintain that too to keep apps working or drop support. The performance is a non-sequiter for a few reasons. Even in the hot path the conditional is going to get speculated away since the wine subsystem is uniquely instantiated per application. If you were really paranoid you could ship multiple binaries and select which ones to use dynamically at runtime. I suspect the bigger problem is one of maintenance - Microsoft does this with a huge budget and a huge team whose sole responsibility is maintaining that comparability. A community project is going to have to make more pragmatic choices.
Diagnosing why a bug happened is interesting. Sometimes it points to a communication shortcoming that the project/organization should strive to improve on. Of course, if you're slinging code with a few buddies, it might not matter.
In the worst/most paranoid case (xz) you might also want to know who planted a backdoor and where else they made commits to.
Checking git history is one of the first things I do when I find unfamiliar code or when I'm debugging something. Git history contains a lot of context that helps speed up the rest of the process
You've truly never come across a single thing you've gone to "fix" only to have it break stuff and then explained why it needs to stay that way in a commit message?
There are several possible reasons for that, but I'll just say it's a very common use case. It's actually the one of the original use cases for SCSs. How else could you revert to a prior commit?
Simple but pretty common use case: If you're ever looking at some pre-existing code and thinking "I don't really get why it's written like this?", you can get a lot out of looking through the git blame/history.
Often "weird design" is there for historic reasons around stuff like backwards compatibility. If your project has a well managed commit log, you can find the notes of whoever implemented it, possibly even with their motivation for doing so.
At the very least, you can find out who wrote that code and ask them about it.
That actually is helpful, but it would require a git history that actually makes sense. I rarely encounter well thought out git histories. For me its often just dev ramblings or 10 word commit messages
If I can choose "badly maintained history" and "no history" I'll choose the former every time. That's even less of a problem when Claude can do a first pass for you.
There seems to be some confusion here (including in the linked discussion?) about what this is. This is not continuing the development on the original Bun (Zig) codebase.
It is extracting a subset of that codebase for deployment purposes. The full version of Bun (presumably in Rust?) will continue to be used for actual development.
So it is not a replacement for Bun, but a supplement to it.
> Cruller is not intended to replace Bun for development. It is a minimal, specialized runtime for executing production code.
> In any case, I do not want to throw away such a large codebase that has taken several years to build. It makes more sense to turn it into a convenient embeddable library that can be used throughout the Zig ecosystem.
This seems pretty sensible to me. It's nice if the Zig ecosystem has an embeddable JS runtime.
I think there are quite a few high profile examples where the fork was successful (egcs comes to mind which eventually became the official gcc, also all the BSD flavours). And even when the fork ultimately isn't successful, it sometimes at least forces the original project to adapt (e.g. ffmpeg vs libav).
E.g. "it's difficult to make predicitions, especially about the future" ;)
PS: of course for this specific project I don't quite understand the reason. The original Bun was largely a line-by-line port of esbuild from Go to Zig, so it's not like the original codebase was a marvel of engineering to begin with...
I don't know that there's ever been a high-profile fork of a product acquired by such a fat, mealy, and genuinely unspooling parent as Anthropic's acquisition of Bun before.
But by all means I would love to hear some examples of that.
What makes Bun Bun is all the things that got removed from this project. Node is already powerful enough and well maintained. Why would anyone use this?
> The main design decision is to treat this as a runtime, not a general-purpose Bun replacement. A minimal launcher loads a pre-built entrypoint; features that require package installation, bundling, TypeScript transformation, or bun test are intentionally outside the scope.
The author is saying explicitly they don’t want to make a Bun replacement
One thing to point out is all the author's comments seem LLM generated and so does the README and the latest commit to the branch (large explanation in a comment and then change)..
Heroic effort, sifting through that code I mean, but frankly I would have started a new one from scratch, the only thing of value is the name/popularity of the original project.
Indeed it's surprising, given the zig compiler famously causes all developers who use it to merge into a single organism with a combined nervous system.
It seems you misunderstand what happened to Bun: Anthropic has burned some investor money for a PR stunt -- our LLM can port this -- and also to punish Zig which dared to ban LLM contributions by taking away a significant project from the ecosystem. It worked, Rust didn't dare to enact a similar ban.
Anthropic tried hard to port using their LLMs but their models were not trained on enough zig corpus hence failed and they blamed it on Zig's principles
Please re-read my comment, I didn't mention anything about Rust. (IK they later switched to Rust. A year ago no LLMs were good at rust, similar to zig today, however its different story now)
Why feel bad? If you’re using Zig to write a game or a system tool why should you care if Bun uses it or not? All this drama has very little impact on most users of Zig or its developers (some financial impact, but Andrew had already foreseen that as a risk and avoided depending on it)
libghostty is more impactful than Bun IMO. It doesn’t matter to most other developers that Bun was written in Zig, but that such a a good library is written in Zig could have an impact on whether other devs use Zig to write libraries, or if they consider Zig for their app when having libghostty as a dependency
TigerBeetle is also more important than Bun since Zig is a better match for their needs it was for Bun, and TigerBeetle team seems to be a better partner for the Zig foundation. The Bun/Oven team seemed to be an annoyance rather than a synergetic partner.
But in general, pre-1.0, I think any large project that uses it is just a bonus.
You don't need to "justify" Zig's existence. It exists, and people like it, and that's enough. IMHO, I wouldn't use it, but there are valid use cases, even if it's "just" a nicer C.
Back when Bun was announced I was so excited for it. I was onboarding on a large node TypeScript project at the time and the build time and resource usage were killing me. Deno’s compatibility hacks and rough edges with node made it a non-starter.
I thought Bun’s promise of full node compatibility with better devex was a dream come true. Then “1.0” was announced and the first simple script I tried segfaulted Bun. The second threw a random error because the http server didn’t implement some stream api. I knew then that Bun was in the league of Vlang. Over promise and under delivery. Just ship “1.0” anyway for marketing and VC reasons. I haven’t touched it since. 2 people I trust that spent time looking at its code and both said to stay away.
Deno had the quality, but missed on the compat because (initially at least) it had an original and more ambitious vision. At least I’d give them that. Bun promised compat in exchange of any ambitious vision but it offers nothing. It runs on hype and hyperbole.
i dont see the point of this to be honest, if rust is actually better for bun why are you people just hating on it for no reason? a software doesnt have to be written in your favourite language for it to work.
bun was a sloppy project is zig and still sloppy in rust
> Development would be done using the full Bun runtime, while production would use its lightweight fork, Cruller. I do not have the resources of the Oven team to develop and maintain a massive general-purpose runtime, so I want to focus on specific production requirements.
I don't know about others, but I don't think I would deviate my development runtime from my production runtime so significantly. The chance for behavior that only rears its head in production is too high for my liking.
reply