Replies: 8 comments 8 replies
-
If we trust the benchmark bun have a faster FFI than nodejs (~x2.4) . The project start just 1 year ago. So It's sound promising, but i'm afraid about the stability in real life. What do you think @alexhultman, did you work with them for the uws integration ? |
Beta Was this translation helpful? Give feedback.
-
My quick Benchmark : uws.js
bun.ts
uws.js with ssl: bun.js with ssl: uws.js without ssl: bun.js without ssl: |
Beta Was this translation helpful? Give feedback.
-
This should've probably been posted in uWebSockets.js |
Beta Was this translation helpful? Give feedback.
-
Looks pretty slick I guess, but this will be an uphill battle for years to come |
Beta Was this translation helpful? Give feedback.
-
It's pretty obvious Deno is nothing but a cash grab profiting from Node.js's success - they did strict CLA from start, they raised millions of silicon valley funding and they sell their Deno deploy as their paid product (all they missed doing with nodejs while it blew up). Good for Ryan Dahl, but for all who want something actually fast and efficient it sucks just like nodejs. So whatever this bun thing is, if they can show these perf numbers which almost reach uWS.js levels I would say that's a very good foundation and I would like to see Deno/nodejs raise their perf to try and match it |
Beta Was this translation helpful? Give feedback.
-
Honestly, without overhyping this, bun is pretty much the ideal outcome of uWS - he seems to reimplement nodejs, but without being slow as a snail. That's the, in my opinion, only logical conclusion to draw based on uWS's perf numbers - I would have done this myself if I had the time. The fact he's using nodejs compatible interfaces, but implementing them faster is going to help him get users very fast and is very clever. What people who cannot put two and two together miss, is that uWS only shows one thing: that nodejs is a trainwreck in terms of performance. People can't use their brain and think "well if uWS is so much faster, that must mean nodejs is slow as shit". This guy made the connection and took the message. Whatever he does, it will be at least 5x better than nodejs is currently doing. So yeah in essence, I think this is very interesting and very much needed. |
Beta Was this translation helpful? Give feedback.
-
Tried it, it rocks. Deno being "written with the fastest server tech available" sure performs like shit in comparison: alexhultman@Alexs-MacBook-Air uSockets % ./http_load_test 40 localhost 3000 |
Beta Was this translation helpful? Give feedback.
-
I don't think it is based on uWS, looks more like he experimented with it but ultimately made his own implementation. Still, it's way faster than Deno and Node.js and that's what counts. Problem is funding and being burned out - this is a massive project covering a lot of area, and sooner or later you will get burned out and/or get a paying job instead, because - Open source is kind of, unsustainable, since there's little money and the time you spend on it can be waaaaaay more personally rewarding by just getting a good job instead. So question is how long this project can survive - the code looks pretty experimental and will need much polishing. The guy is going to be declared a rock star for sure but then what? His own best interest will be to just knock on Google's door and get himself a high paying job instead. When you get 20k GitHub stars in a week you overload dopamine receptors and then when everything calms down you stand there with a massive heap of unfinished work and no payment, depression sets in. So yeah, I love it but realistically it will be uphill from now |
Beta Was this translation helpful? Give feedback.
-
What do think about bun? (Server based on uws)
https://bun.sh/
Beta Was this translation helpful? Give feedback.
All reactions