I like function components too, but I've seen some interesting arguments against using them:

https://medium.freecodecamp.org/7-reasons-to-outlaw-reacts-f...

And I wish I had a reference for this, but I seem to remember reading that any future performance optimizations that could be applied to function components could probably also be applied to class components (at least ones that contain only a render() function and nothing else).

In any case, I'm hoping that React eventually gets to a point where developers don't have to make a choice between class and function components, and the compiler or the runtime makes that decision for us on a per-component basis.

EDIT: As an example, here's a Babel plugin that takes care of the conversion for you at compile time: https://github.com/remcohaszing/babel-plugin-transform-react.... Not sure what its heuristics are for determining what should and shouldn't be converted though, if any.

This is true in theory but the React team has not optimized them directly. React.memo is the only way to optimize them thus far.

I think he's referring to the optimizations the JIT/engine will do.

Indeed, also checkout the React.PureComponent for another example of an in-between Class and Function optimisation.

Surprisingly, pure functional components are called even when their props haven't changed so PureComponent is not "in between" but "even more" than functional components. As for regular components, you can override this behaviour by adding a componentWillReceiveProps() method (yes, you can add a method to your functional component!).

Sadly (or luckily?) with hooks they aren't stateless anymore.