Tom MacWright

Increasingly miffed about the state of React releases

I am, relative to many, a sort of React apologist. Even though I’ve written at length about how it’s not good for every problem, I think React is a good solution to many problems. I think the React team has good intentions. Even though React is not a solution to everything, it a solution to some things. Even though React has been overused and has flaws, I don’t think the team is evil. Bask in my equanimity, etc.


The state of React releases right now is bad. There are two major competing React frameworks: Remix, funded by Shopify and Next.js, funded by Vercel. Vercel hired many members of the React team.

It has been one and a half years since the last React release, far longer than any previous release took.

Next.js is heavily using and promoting features that are in the next release. They vendor a version of the next release of React and use some trickery to make it seem like you’re using React 18.2.0 when in fact you’re using a canary release. These “canary releases” are used for incremental upgrades at Meta, too, where other React core developers work.

On the other hand, the non-Vercel and non-Facebook ecosystems don’t have these advantages. Remix suffers from an error in React that is fixed, but not released. People trying to use React 18.3.0 canary releases will have to use npm install --force or overrides in their package.json files to tie it all together.

This strategy, of using Canary releases for a year and a half and then doing some big-bang upgrade to React 19.0.0: I don’t like it. Sure, there are workarounds to use “current” Canary React. But they’re hacks, and the Canary releases are not stable and can quietly include breaking changes. All in all, it has the outward appearance of Vercel having bought an unfair year headstart by bringing part of the React team in-house.