Is React Fiber ready yet?

I can't comment on this specific implementation, but it depends on your tolerance to risk.

If you're going for a backwards compatible change to some thing that is quite complex, it's not unheard of to create a suite of tests that fully describe the features of a piece of software (if they don't already exist) and then compare the 'old' and 'new' results.

This then allows you to understand if you've changed anything, and can even highlight where you've fixed bugs in the old codebase.

I've done something similar before when refactoring a reasonable sized portion of legacy code that had no unit tests (code was old + organicly changed over the years), plus a good number of configuration options that all had to be backwards compatible (work like it used to, or have a good explanation why it doesn't).

Yes, these are our existing unit tests (which are fairly comprehensive) running against our new implementation. Once we reach parity in tests we'll be close to release.

Here's more info about it https://github.com/acdlite/react-fiber-architecture

The tl;dr based on this talk[0] by acdlite is that they're looking to separate out high-priority and low priority rendering, so you don't end up (for instance) with animations that are janking because item #12,141 in your list hasn't rendered yet. They're using requestIdleCallback and requestAnimationFrame to schedule the work. Sounds like the architecture of fibers themselves is very heavily inspired by the way a call stack is implemented.

[0]: https://youtu.be/aV1271hd9ew