When to break up a component into multiple components

Did you know that you could write any React Application as a single React
Component? There’s absolutely nothing technically stopping React from putting
your entire application into one giant component. Your function would be HUGE.
There’d be a TON of hooks for state and side-effects, but it’s totally possible.

If you tried this though you’d face a few problems:

1. Performance would probably suffer: Every state change results in a re-render
   of the entire application.
2. Code sharing/reusability would be… not easy. At least if you made it a
   class component, which you might have to do if you wanted to use
   componentDidCatch to handle runtime errors. If you were allowed to use
   react-error-boundary so you could use hooks, then it would be
   considerably easier. But… I digress…
3. State would be a challenge: Knowing which pieces of state and event handlers
   went with what parts of JSX would make your head hurt 😬 and lead to some
   hard to track down bugs 🐜 (This is one benefit of the separation of
   concerns).
4. Testing would be 100% integration:
   Not necessarily an altogether bad thing, but it would
   be pretty tough to test edge cases and keep things isolated to the parts
   that you’re trying to test, so you would suffer big-time from over-testing
   (which is a common mistake in E2E testing).
5. Working together on the codebase with multiple engineers would just be
   terrible. Can you imagine the git diffs and merge conflicts?!
6. Using third party component libraries would be… ummm… impossible? If
   we’re writing everything as a single component third party libraries is at
   odds with that goal! And even if we allowed using third party components,
   what about HOCs like react-emotion? Not allowed!
   (Besides, you should use the css prop
   anyway 😉).
7. Encapsulating imperative abstractions/APIs in a more declarative component
   API wouldn’t be allowed either meaning that the imperative API would be
   littered throughout the lifecycle hooks of our one giant component, leading
   to harder to follow code. (Again, unless you’re using hooks, in which case
   you could group relevant hooks together, making it easier to manage, but
   still a bit of a nightmare).

These are the reasons that we write custom components. It allows us to solve
these problems.

I’ve had a question on my AMA for a while:
Best ways/patterns to split app into components.
And this is my answer: “When you experience one of the problems above, that’s
when you break your component into multiple smaller components. NOT BEFORE.”
Breaking a single component into multiple components is what’s called
“abstraction.” Abstraction is awesome, but
every abstraction comes with a cost, and you have to be
aware of that cost and the benefits before you take the plunge

Duplication is far cheaper than the wrong abstraction.
— Sandi Metz

So I don’t mind if the JSX I return in my component function gets really long.
Remember that JSX is just a bunch of JavaScript expressions using the
declarative APIs given by components. Not a whole lot can go wrong with code
like that and it’s much easier to keep that code as it is than breaking out
things into a bunch of smaller components and start
Prop Drilling everywhere.

Conclusion

So feel free to break up your components into smaller ones, but don’t be afraid
of a growing component until you start experiencing real problems. It’s WAY
easier to maintain it until it needs to be broken up than maintain a pre-mature
abstraction. Good luck!