Why even attempt to replace node?

[nbro]

I don't think that starting the readme with the premise that this is a drop-in replacement for node is a good idea. I think google made the same mistake with Dart but soon realised the mistake and opted for another strategy, and Rust with C++. Nobody is going to replace C++ or C. Everything that is actually important is written in these languges, from the actual libraries to train all these AI tools, to everything else, like your browser. Nobody is going to replace Javascript or C++ any time soon. Nobody is going to replace node.js any time soon. It's used all over the place, on the cloud, by Google in its servers, etc. It's crazy to think that bun is going to replace node. I like Rust, but I've had enough of this trend to lose time to rewrite everything in Rust because you can't write C++ code. Ok, now you've been acquired by Athropic and maybe you might replace Node and then soon that nice MIT license will end up where? You should use Rust for new tools, new problems, not lose time to replace well-established tools just because you don't like memory allocation, that is, basically you don't like computers.

[RohanDaCoder]

If you dislike the idea of bun being a better or improved alternative of node, then just don't use it. Node is still up and running and isnt going anywhere. You can think of Bun as a alternative if you dont want to replace node, its a choice.

[nbro]

Sure, I am not going to use a new technology just because it's hyped. Node works pretty well for most cases. If I have a constraint to ever use this bun, I'll use it, but I don't use stuff just because it's hyped or trendy or has an improvement that is not really significant for most cases. It's like using Rust because of the borrowing system, instead of using C++, where you also have smart pointers, and stuff like this, but you hate the raw pointers, Rust also allows you to run unsafe code. Nothing is perfect, and the fact that these tools being highly modified by AI is not actually a good thing, but ok maybe it will turn out to be useful one day. But it's rare that one technology completely kills another. One new programming language never really killed another good programming language with a good community. People still write Fortran, for example. Speed is not always the most important metric. Actually, sometimes it's actually the wrong metric. Being able to generate a lot of code fast doesn't mean anything if that code is often wrong.

[RohanDaCoder]

Bun isn't really a whole new language, it can natively run typescript. There is just some new APIs like the Bun object available globally. But it is still typescript. You could say the engine has changed but the car is still the car itself. It matches with the browser ecosystem. There are many browsers with different engines, but the web is still the web. Some are better in different aspects. Our work is to pick the best one according to our needs.

[realityworks]

I don't think that starting the readme with the premise that this is a drop-in replacement for node is a good idea. I think google made the same mistake with Dart but soon realised the mistake and opted for another strategy, and Rust with C++. Nobody is going to replace C++ or C. Everything that is actually important is written in these languges, from the actual libraries to train all these AI tools, to everything else, like your browser. Nobody is going to replace Javascript or C++ any time soon. Nobody is going to replace node.js any time soon. It's used all over the place, on the cloud, by Google in its servers, etc. It's crazy to think that bun is going to replace node. I like Rust, but I've had enough of this trend to lose time to rewrite everything in Rust because you can't write C++ code. Ok, now you've been acquired by Athropic and maybe you might replace Node and then soon that nice MIT license will end up where? You should use Rust for new tools, new problems, not lose time to replace well-established tools just because you don't like memory allocation, that is, basically you don't like computers.

The kicker is the amount of money and shares Anthropic spent on Bun... For what purpose?

[nbro]

Yes, exactly. It's just to show off. Instead of spending the resources and money on novel problems and making their tech more accessible and less expensive for everyone, no, they invest on rewriting tools that already work pretty well, just because they have the models to rewrite the code by brute-force and then they say they created a new version control system, which basically copies ideas from everyone, because the AI doesn't really generate anything that we couldn't infer from the data. What a waste of resources. These resources could have been used for something important. Instead, they use models to brute-force solutions and waste huge amounts of resources, which causes a very negative impact on nature and the environment. Why spend so many resources to create a solution that already exists just to have a tiny small improvement that is beneficial to only 0.01% of the cases?

[realityworks]

5. Yes, exactly. It's just to show off. Instead of spending the resources and money on novel problems and making their tech more accessible and less expensive for everyone, no, they invest on rewriting tools that already work pretty well, just because they have the models to rewrite the code by brute-force and then they say they created a new version control system, which basically copies ideas from everyone, because the AI doesn't really generate anything that we couldn't infer from the data. What a waste of resources. These resources could have been used for something important. Instead, they use models to brute-force solutions and waste huge amounts of resources, which causes a very negative impact on nature and the environment. Why spend so many resources to create a solution that already exists just to have a tiny small improvement that is beneficial to only 0.01% of the cases?

We all know it. Big tech is not here to solve problems, it's here to make money. Let me just list a few things.

1. Social Media causing increased depression and suicide in teens.
2. Smart phone addiction causing a disconnection of people from each other.
3. Marketing which makes us all feel poor because we don't have the latest gadget.
4. Narratives that "AI" will replace all white collar jobs.
5. VR headsets that cause eye strain and nausea are the future of human communication and work.

Since 1990, we can keep in contact with one another more often, and buy things online, most of which we don't need.

What else?

It's the system that was born out of "Everybody lives by selling something".

My solution, go touch some grass and let everyone get on with what they want to get on with.

No one will save you, except yourself, so turn off that phone, get some sun and forget about LLMs. :)

We will be sent to war soon enough...

[amondnet]

I don't think the Dart and Rust vs C++ comparisons fit here.

Dart tried to replace JavaScript as a language. The plan was to ship a Dart VM in Chrome, and it failed because adopting Dart meant rewriting your code. Rust vs C++ is also a case of a new language competing with an established one.

Bun isn't a new language. Like Node, it's a runtime for the same language. Node describes itself as "an asynchronous event-driven JavaScript runtime" (https://nodejs.org/en/about), and Bun's docs describe its runtime the same way (https://bun.com/docs). Bun runs the same JS/TS code and npm packages, and it targets the Node APIs.

Much of the JS toolchain has already moved to Rust: SWC, Rspack, oxlint, Rolldown. None of them asked anyone to rewrite their JavaScript. Python went through the same shift with uv and ruff. Bun applies that pattern to the runtime, as Deno did before it.

When Bun's docs call it a "drop-in replacement for Node.js", they mean the runtime. The phrase promises compatibility. It doesn't claim that Node is going away.

Alternative runtimes are already common. My favorite is workerd, Cloudflare's runtime. It's written in C++ and offers Node compatibility behind the nodejs_compat flag, and Node is doing fine alongside it. Other languages have had the same arrangement for years. OpenJ9 runs the same bytecode as HotSpot, PyPy runs the same Python code as CPython, and JRuby runs the same Ruby code as CRuby. HotSpot, CPython, and CRuby are still the defaults.

[nbro]

You probably didn't read what I wrote to think I don't understand what a runtime is.

If you highlight that Bun is a drop-in replacement for node and make claims that it's better, you're suggesting that people should stop using node and node should die, i.e. you should replace node with bun.

But why replace node if it works well in 99.9% of the cases?

• Is it because you have nothing better to do in life than rewriting what already works and call it original?
• Is it because you think Rust is secure when it has security holes in the type system, etc?

My point is that we should stop rewriting stuff that works well. We should solve problems for which we don't have a good solution yet, not recreate solutions and waste time, effort, money. It happens all over the place. For example, people attempt to create OSes in Rust, to do what? To replace Linux, Windows or MacOs? It's just a waste of time. I think rewriting slow tools in C++, Rust or C, makes sense, if there's the need for a significant speedup. But what speed do you want more from node? It works pretty well. Never had any issue with node so far. So, write new tools in your pet language Rust, which soon will be owned by corporations like Anthropic, just because it's trendy, but don't waste time on rewriting tools that work well already. How many software projects died because people simply didn't accept that they were doing exactly the same with no significant advantage? So many.

[ADolkun]

"Not anytime soon" is never the reason not to push new things that has the potential to be better...

[nbro]

Ok, but at what cost? You spend thousands of hours and dollars on a new product that is better in 0.01% of the cases. Is it really worth it? Instead of trying to solve problems that do not yet have a solution, people keep resolving the same problems, just to show that they built something, as they can't build anything original. It's easy to start a new programming language, a new runtime, a new compiler, when there are already like 10 references you can copy from. It's even easier if you let the AI analyse everything and get the best of each, while it consumes a stupid amount of energy and resources and pollutes the world just to be able to get a significant advantage in 0.01% of the cases, which don't affect basically anyone. It's just something you write on your CV to feel like you've achieved something. But in reality nowadays you can't even claim you really built anything, because you made the AI do everything. You just wrote a command like a baby commands his mother. Do this, do that. Then the baby is happy. There's absolutely no merit in writing commands in English. Everyone can do it. So, what now? You're using a tool, like a smartphone, to listen to music, call your mom because you cry, and so on. Everyone does it. There's no merit. You will not get a nobel prize for prompting. Well, given how stupid society is now, one day maybe people wil get nobel prizes for the best prompt.