"Framework" semantics get in the way of progress way too much with React and I'm really tired of the "you must be this tall" narative from react.dev. Tools that drastically aid you to be productive with React deserve a seat at the table. Vite is just the obvious start among many other tools. - ThreadSky

tannerlinsley.com • 32 days ago

"Framework" semantics get in the way of progress way too much with React and I'm really tired of the "you must be this tall" narative from https://react.dev. Tools that drastically aid you to be productive with React deserve a seat at the table. Vite is just the obvious start among many other tools.

Comments

kyledl.bsky.social•32 days ago

This is exactly why I've moved away from React and haven't looked back. https://svelte.dev has a place in my heart now.

kyledl.bsky.social•32 days ago

Admittedly, Svelte has some shortcomings, but it's a newer framework, so I'm cautiously optimistic about it.

ducin.dev•30 days ago

Curious to hear - what are the main shortcomings in your opinion?

thomas.rocks•31 days ago

What are the shortcomings you face?

kyledl.bsky.social•31 days ago

The only real drawback IMO is that there's less of an environment supporting Svelte as compared to something like React. However, this isn't that major for me, and it's a given since Svelte is relatively newer/emerging compared to other frameworks. I still love Svelte the most, though.

thomas.rocks•31 days ago

That makes sense. I feel its improving, and trying to do my part in that as well, but there are of course a few ways to go.

kyledl.bsky.social•31 days ago

Of course! I appreciate all that you guys do for Svelte.

baxuz.bsky.social•32 days ago

The react ecosystem has become (always has been) a bunch of tools that work in a way to bypass react.

ethangthereal.bsky.social•32 days ago

React is going to suffocate itself. Imagine if more of these TW UI libraries realize how easy it is to be agnostic with TW. Or when people realize the significance of the Angular Wiz merge. Or when Silicon Valley realizes .NET isn't all that scary after all.

kyrre.dev•31 days ago

Come to the Svelte side, we have cookies

fredkiss.dev•30 days ago

📌

codyapearce.bsky.social•28 days ago

Should create a fake framework that wraps Vite, does the same thing but fulfills their silly requirements

rtorr.com•31 days ago

I kind of miss when react felt like just a library dependency

thiagolino8.bsky.social•31 days ago

It's really strange that a library that doesn't have an official framework is so afraid to recommend the main way of creating apps nowadays
Meanwhile, even libraries that have an official framework are still strongly recommending Vite

opensauced.bsky.social•28 days ago

reposting here for visibility
I have an open PR that adds vite in the quick start guide
https://github.com/reactjs/react.dev/pull/7416

h14h.social•32 days ago

The https://react.dev context/reducer docs could have saved me a ton of time had they even just let me know jotai & zustand exist.

Being newer to modern React, the "unknown unkowns" problem has bit me multiple times. An official showcase of great 3rd party libraries would've been a godsend.

ducin.dev•30 days ago

That's because proper documentation and "learning course" are two different things.

The goal of documentation is not to cover userland solutions. 🤷

evaogbe.bsky.social•32 days ago

The React ecosystem changes too fast for official 3rd party library endorsement to make sense. Even state management libraries don't have a consensus. Xstate or react router could replace zustand. Third party recommendations are a community responsibility

dogpawhat.tech•32 days ago

I feel like there needs to be an agreement between core and community stakeholders where the community can recommend ecosystem packages and patterns that are productive.

It's kinda overdue really given how "batterys not incuded" react used to be

atraversay.bsky.social•31 days ago

Yeah I don't think it's just because "the ecosystem changes too fast". We've seen big maintainers struggle to find the the correct ways to use react primitives. People deep into this struggle, and React team never blesses one approach. How can that lead to stable ecosystem?

acemarke.dev•32 days ago

I literally tried to get a "React community tools" docs site off the ground a few years ago, with the idea that it would be a fairly maintained list of approaches, but it didn't get any momentum:

https://github.com/markerikson/react-community-tools-practices-cheatsheet/discussions/1

johnmcdowell.bsky.social•31 days ago

Hey Mark, curious if you've heard of any similar efforts to yours which eventually took off? It's a great idea. Shame it didn't pick up enough steam.

acemarke.dev•31 days ago

I think I've seen a couple vaguely related sites. I think ThisDot maybe had one.

tannerlinsley.com•32 days ago

Everything but React itself is 3rd party, so this breaks down real quick.

ethangthereal.bsky.social•32 days ago

Simple site. OSS owned. Categories and votes?

robcresswell.bsky.social•30 days ago

Votes isn't trivial. How do you measure growing popularity? What is the "most popular", the thing with 10000 votes 5 years ago, or the thing that has 1000 votes in the past month?

devongovett.bsky.social•31 days ago

If you include Vite then you also have to include Parcel, webpack, rspack, etc too. They all do roughly the same things, so you’ll need some starter template for all of them at least. I’m hopeful it doesn’t need to be a “framework” but you do need something on top of a raw bundler.

ducin.dev•30 days ago

No, you don't "have to".
Who said you/they "have to"?

devongovett.bsky.social•30 days ago

I mean, this whole discussion is about including more options. Including only Vite is hypocritical. What makes it special? That it has more downloads? If that’s the case, than webpack still wins.

wojtekmaj.pl•30 days ago

Webpack is on a slow and steady decline, while Vite is shooting for the stars. I have no doubt it will be #1 by the end of the year.

wojtekmaj.pl•30 days ago

Still, I think it should be the combination of popularity and ease of use. It has to be a viable alternative to frameworks which set up lots of stuff for you.

Parcel has the ease of use, but lacks popularity.
Webpack - the other way around.
Vite has both.

mmmcarl.bsky.social•30 days ago

I suggest you and everybody else to put @rspack.dev and RsBuild in your radar. Contrary to vite and parcel, migrating old webpack projects to rspack is a piece of cake. I'm talking about production stuff here (personally experienced), not hello worlds.

retropragma.bsky.social•31 days ago

Better for beginners to be recommended a single setup, instead of a list of (totally valid) solutions. The more you follow the most popular and least complicated path, the more likely someone else will have hit and solved any issues along the way.

tannerlinsley.com•31 days ago

sure. Show all the options. Anything more is better than the current setup.

retropragma.bsky.social•31 days ago

And I'll preempt the kingmaking concern: The community chose the king, and React's docs are merely respecting that choice to ensure the smoothest path.

adamrackis.bsky.social•31 days ago

This. The community settled on Vite. Everything else is niche. You don’t have to list every possible option, and it’s arguably better not to.

devongovett.bsky.social•31 days ago

Obviously I strongly disagree with that!

adamrackis.bsky.social•31 days ago

It’s less an opinion and more just an observation on usage.

devongovett.bsky.social•31 days ago

The docs already list lots of meta framework options. I don’t think listing some additional bundler templates is any different. If React were to list only one it would be Next.js, but the whole point of this discussion is that other options are valid too.

retropragma.bsky.social•31 days ago

I'm not saying there's a one size fits all solution, per se. But a lot of people have the same needs, and I believe the official guidance should prioritize what I outlined above.

jimmyvanveen.com•32 days ago

“You must be this tall” is a perfect way to phrase it

evanyou.me•31 days ago

I always try to give React team benefit of the doubt that they just genuinely believe the framework-dependent patterns they are pushing for the at the best interest of all React users, but after a few years the elephant in the room just grows bigger: what if they were… actually wrong?

munawwar.bsky.social•31 days ago

The original discussion began with CRA and IMO the alternatives being suggested in this context (especially for purpose of a migration) don't need all the new react patterns.

oscarcao.bsky.social•29 days ago

There’s arguments on both sides.

For: those who just want a plug n play solution.

Against: those who want options. And one of the main reasons why React beat out Angular in popularity. There wasn’t a “best practice” for React that you had to follow by the book.

tannerlinsley.com•31 days ago

Same here. I keep using React and fighting to support it and shape it, but recently it has felt more than ever that the future of React is not in the hands of its users.

ayc0.github.io•31 days ago

I had this feeling for a long time. Like with React 19 release, they announced a ton of things that were indeed amazing, but that wasn't communicated with the community at all.

And in the mean time there are a lot RFCs opened by the commu that will never be picked up

ayc0.github.io•31 days ago

Even the whole conversation with the change in behavior of lazy components shouldn't have happened if this change was made transparent since day 1.
It had to be found by the community & fought by the whole community to drive a change

ayc0.github.io•31 days ago

IIRC the migration to hooks started a few discourses at the time, but before the release of hooks

Whereas changes now feel like they are discovered after the release ("use client", useOptimistic, parallel fetching, etc)

rossipedia.com•31 days ago

ok so not to hit the nail too on the head here, but....

what behavioral change in lazy components?

ayc0.github.io•31 days ago

This issue:

https://github.com/facebook/react/issues/29898

acemarke.dev•31 days ago

I'm assuming this refers to the "SuspenseGate" change around prefetching sibling Suspense components:

https://tkdodo.eu/blog/react-19-and-suspense-a-drama-in-3-acts

which was fixed shortly before React 19 went live:

https://github.com/facebook/react/issues/29898#issuecomment-2477449973

nemeciii.bsky.social•31 days ago

Also because it's like that, package maintenance gets dropped and causes massive migrations to other libraries.

IMHO that's the burden of React software in maintenance phase. There's no guarantee that any of the dependencies have a migration or a support path even just 5 years forward.

stoobs.dev•31 days ago

honestly react-native is what is keeping me around. If there was the same thing for vue/solid/svelte I'd move on (nativescript doesn't count).

bigmistqke.bsky.social•31 days ago

Why it doesn't count?

venatio.dev•31 days ago

Have you tried solid.js?

tannerlinsley.com•31 days ago

oh yes. It’s my favorite

ericclemmons.com•31 days ago

I feel this. Trying to get “use” right but the docs wave me away and say it’s for framework authors to support.

Same with RSC.

rossipedia.com•31 days ago

BUT WHAT DO I DO IF I WANT TO BE ONE OF THOSE FRAMEWORK AUTHORS?!?

(ノಠ益ಠ)ノ彡┻━┻

devongovett.bsky.social•31 days ago

Spend months reverse engineering it like I did 😅

ericclemmons.com•28 days ago

I’ve seen the wire format of RSC, but unsure what the client API is to parse it.

Months to reverse engineer, or days spread out over months? 👀

adamrackis.bsky.social•31 days ago

I absolutely hate that hand-wavy bullshit. Give me the gross details, please. What are they hiding?!

developit.dev•30 days ago

dlvhdr.me•30 days ago

Seriously why recommend next js

ndeyefatoudiop.bsky.social•31 days ago

What is surprising to me here is that if you can highlight next.js you can highlight vite: which is not even opinionated.
More and more devs think react = next.js which makes the journey unnecessary harder 😢

Comments

Posting Rules

Reply