Lots to unpack here. Modern JS going just great.
This situation is avoidable
https://dev.to/thejaredwilcurt/bun-hype-how-we-learned-nothing-from-yarn-2n3j
Lots to unpack here. Modern JS going just great.
This situation is avoidable
https://dev.to/thejaredwilcurt/bun-hype-how-we-learned-nothing-from-yarn-2n3j
@brianleroux I really liked the fork of io.js what we got in the end was so much better node v4 was way beyond 0.12 the previous release. Iām also not convinced and npm wouldāve gotten as good as it did without yarn. The tribalism and the incompatibilities are troubling. But the years it took npm to catch up are probably why yarn never shut down. Deno on the other hand has made node better, but Iām not tapped into its trajectory. Itās absolutely not like io.js.
Bun seems like a toy tbh
@brianleroux @reconbot io.js worked because it was a true oss fork.
My deep concern is these new projects are pointing to examples like io.js as why we should trust them, but weāre seeing private companies with VC funding being spun up behind them. The pressures VC funding produces is a worrying new element that makes some of the lock in aspects of bun more concerning than they otherwise would be. If thereās money to be had, why isnāt it funding existing oss rather than greenfield projects?
@reconbot @brianleroux NPM Inc was internally in a certain kind of shape when Yarn came out. Its release shook things up in a very serious way and caused me and others a LOT of stress, and some of the most regrettable overworking I've done in my entire career, but it did directly lead to NPM@5 and subsequent perf improvements.
I think the thing that Yarn really did for us (us being the NPM team) is that it taught us that what people actually give a shit about is performance, not things being bug-free, or featureful, or cross-platform. As long as you can say your thing goes really really fast, people basically go ga-ga over it and forget all the other things that matter. They will put up with absolute trash in order to get those speed gainz (and, frankly, Yarn was a broken piece of shit when it first launched and you can't convince me otherwise). Heck, even Yarn's "ergonomics" being an improvement seems like bullshit to me.
All people actually wanted was a faster package manager, and as soon as they got that, they stopped caring about Yarn, except for the folks who had already made the switch and didn't really have a reason to bother switching back.
@reconbot @brianleroux anyway https://orogene.dev is faster than bun sometimes, and uses way less memory, and is actually built by someone who knows what they're doing.
Don't use it, though.
@reconbot @brianleroux itās the worst burnout Iāve experienced in my entire career and the most stressful several months where I felt there was an existential threat to my job if I didnāt work 12-15 hour days to get npm5 out.
Never again
@zkat @reconbot @brianleroux I'm still sore about the whole "yarnpkg registry aliasing the public registry at the CDN level" thing.
The teapot outage happened because of that proxy: our CDN had to add an exception to allow proxying between the yarnpkg domain and registry.npmjs, since they hosted both. Our CDN recommended we add strict host header checks to our worker. To my (dis-)credit, I flubbed this.
One yarn maintainer was publicly unkind about this in a way that still gets me angry.
@zkat @reconbot @isntitvacant it's stable, and been so for a long time now, so I'm happy. We just started dabbling with workspaces and its seems useful sometimes.
Ultimately for me the core features have been great (read: boring) for like decade now which is pretty epic foundation to built on.
@brianleroux everyone posting 'javascript' jobs needs to read this, cause from where I'm sitting (jarb hunting), TypeScript is alive and well.
Also affirming that I was right to ignore yarn and I'll be right to ignore bun (except for the two cute ones here at home)
@brianleroux in a recent project I let people talk me into using Bun (0.7.x) (which obviously led me to patch libraries to support Bun, subscribe to still open issues even now in oven/bun, and write workaround scripts) when they are not as well-versed in the culsterfuck that is the history of JS ecosystems.
I need to do better. :(
> But Bun isn't doing that. It isn't taking something hard, and making a simpler abstraction for it. It's taking ESBuild and abstracting it just like how Vite is. It's not an improvement, it's basically the same thing.
š š š š š
My bitching about like 9 out of 10 "cool new JS things" is mostly that.
Like, I just had someone ask last week "what do you think about Bun? it can start a http server in one line of code!"
I like to imagine there was at least some disbelief when I said in vanilla Node with no dependencies it takes two.
@brianleroux OK I'm in like broad agreement, but also there's no reason to believe npm would have gained those features or gotten so much faster if yarn were not around, and plenty of reason to believe the opposite.
I can buy the conclusion, but also I think the author should have considered that idea.
Bun has got some traction because it is attacking real problems that people are suffering from - I hope the same thing happens and the mainline improves after seeing that.