In my opinion, any advancement is worthwhile, even if current consensus is that it’s over-optimization. And if it starts tying up its loose ends, it’ll no doubt benefit node developers as well. Healthy competition and all that.
I’m with you on that. I’ve built dozens and dozens of node apps both professionally and for personal projects and yeah maybe the package installs could be faster, but the overall performance of the server has also been pretty good. If node is slow for you, maybe there’s some other optimizations to be made rather than switching the next new things as a solution.
Pretty much. Their benchmarks seem to be VERY cherry picked to skew things in their favour, specially the testing framework part, where bun compares its speed to one of the slowest testing frameworks out there (jest) and claim victory.
I'm very glad that this guy actually made benchmarks instead of just reading what's on bun's site before posting a video about it.
I guess I don’t understand this obsession with speed? I work in development but I have never ran into a build that’s so slow that it impacts my work. Why is Node considered so bad that everyone wants to make is faster?
I'm with the parent poster. How often is it hindering your work?
It's at tops, a 15-30 second wait, usually done at the start of the project.
I'm a unique use ase, where I have about 10 projects (because microservices) that I'd have to npm install, and it's still not a pain to install them.
Are people running npm install dozens of times a week on the same project? Because why?
I'm not shitting on Bun. I'm actually supporting it's growth. But until it solves some real performance bottlenecks, (switching from gulp/webpack to vite changed our compiling time from 18 seconds to 2 seconds), I can't see any reason for people to change their workflow.
Build times in nodejs are not so great in even medium size projects if you make heavy usage of advanced typescript features - either yourself or through libraries like zod. So if something makes the nodejs runtime faster, it could potentially make ts compiler faster too - for which I'd be very grateful.
I guess I don’t understand this obsession with speed?
for me it hasn't been build speed but rather execution
I've run into problems with dayjs slowing down requests where I need to do a lot of processing. There are arguments to be made about replacing dayjs with datefns and how I should've been doing it differently anyway, but fact is that if the whole execution environment was twice as fast, it probably wouldn't have been much of a problem at all
because speed = your page loading fast = more clients = more satisfied clients = less environmental impact = less build time = less time downloading and lock-ing modules = more time for you ... the list goes on.
I dont understand why is it so hard to understand that ... guys ... speed ... does ... matter? Our time matters? Our anything matters? And if you dont care, I do.
Recently change node+npm+esbuild to bun runtime+package management+bundling and happy with the result.
The project is a static site built with middleman, tailwind, postcss and some frontend libraries.
It was simpler to work with for me. Node is way faster than ruby and so node speed was never an issue for me. But bun install is noticeably faster even for a small project.
Heh. Two days ago I was seeing people claim it was faster than AOT LLVM languages - think Rust or C++.
Google has spent fifteen years stuffing V8 with every JIT optimization they can muster, and it still pales in comparison to native code. No way Bun is cracking that nut.