React "Aha" Moments

Dec 21, 2016

One of my main goals whenever I’m teaching or writing technical content is to maximize “aha” moments. An “aha” moment is a moment of sudden insight or clarity; when the subject matter suddenly makes sense. We’ve all experienced them and the best teachers I know are able to understand their audience and adapt the lesson in order to maximize these moments.

Throughout the last few years I’ve taught React in just about every popular medium. Throughout that time I’ve been taking notes on what triggers these “aha” moments, specifically for learning React. About two weeks ago I came across this Reddit thread which had the same idea. So what I want to do in this post is share my collection of these moments while also adding my thoughts on some of the moments shared in that Reddit thread. Hopefully it will help React “click” for you if it hasn’t yet.

fn(d) = V. Your UI is a function of your state and props are to components what arguments are to functions.

One of the best parts of React is that you can use the same intuition that you have about JavaScript functions for when and where you should create React components. However, instead of your function taking in some arguments and returning a value, your function is going to take in some arguments and return an object representation of your UI. This idea can be summed up in the following formula, fn(d) = V. A Function takes in some Data and returns a View. This is a beautiful way to think about developing user interfaces because now your UI is just composed of different function invocations. This is how you’re already used to building applications and now you can take advantage of all of the benefits of function composition when building UIs.

In React, your entire application’s UI is built using function composition and JSX is an abstraction over those functions.

The most common reaction I see from first timers using React is “React seems cool, but I really don’t like JSX. It breaks my separation of concerns”. JSX isn’t trying to be HTML, and it’s definitely more than just a templating language. There are two important things to realize with JSX. First, JSX is an abstraction over React.createElement which is a function that returns an object representation of the DOM. I know that was wordy but the tl;dr is that whenever you write JSX, once it’s transpiled, you’ll have a JavaScript object which represents the actual DOM (or whatever View is representative of the platform you’re on, iOS, Android, etc). Then React is able to analyze that object, diff it against the previous object representation of the DOM it created, then update the DOM only where a change occurred. This has some performance upsides to it but more importantly shows that JSX really is “just JavaScript”. Second, because JSX is just JavaScript, you get all the benefits that JavaScript provides (composition, linting, debugging) but still with the declarativeness (and familiarity) of HTML.

“Components don’t necessary have to correspond to DOM nodes.”

When we first learn React we’re taught that “Components are the building blocks of React. They take in input and return some UI (descriptor)”. Does that mean that every component needs to directly return UI descriptors as we’re typically taught? What if we wanted to have a component render another component (Higher Order Component pattern)? What if we wanted a component to manage some slice of state and then instead of returning a UI descriptor, it returns a function invocation passing in the state (Render Props pattern)? What if we had a component that was in charge of managing sound rather than a visual UI, what would it return? What’s great about React is you don’t have to return typical “views” from your components. As long as what eventually gets returned is a React element, null, or false, you’re good.

You can return other components

render () {
  return <MyOtherComponent />

you can return function invocations

render () {
  return this.props.children(this.someImportantState)

or, you can return nothing

render () {
  return null

I really enjoyed Ryan Florence’s React Rally talk where he covers this principle more in depth.

“When two components need to share state, I need to lift that state up instead of trying to keep their states in sync.”

A component based architecture naturally makes sharing state more difficult. If two components rely on the same state, where should that state live? This was such a popular question that it spurred an entire ecosystem of solutions which eventually ended with Redux. Redux’s solution is to put that shared state in another location called a “store”. Components can then subscribe to any portions of the store they need and can also dispatch “actions” to update the store. React’s solution is to find the nearest parent of both of those components and have that parent manage the shared state, passing it down to the child components as needed. There are pros and cons to both approaches but it’s important to be aware that both solutions exist.

“Inheritance is unnecessary in React, and both containment and specialization can be achieved with composition.”

React has always been, for good reason, very liberal in adopting functional programming principles. A precedent move away from inheritance and towards composition was when the release of React 0.13 made it clear React wasn’t adding support for Mixins with ES6 classes. The reason for this is because mostly everything that can be accomplished with Mixins (or inheritance) can also be accomplished through composition, but with less side effects. If you’re coming to React from an inheritance heavy mindset, this new way of thinking may be difficult and probably won’t feel too natural. Luckily there are some great resources to help. Here’s one that’s not React specific that I’ve enjoyed.

“The separation of container and presentational components.”

If you think about the anatomy of a React component, it usually involves some state, potentially some lifecycle hooks, and markup via JSX. What if, instead of having all of that in one component, we separate the state and the lifecycle hooks from the markup. This leaves us with two components. The first has state, life cycle methods, and is responsible for how the component works. The second receives data via props and is responsible for how the component looks. This approach allows us to have better reusability of our presentational components since they’re no longer coupled to the data they receive. I’ve also found that it allows you (and newcomers to your project) to better understand the structure of your application. You’re able to swap out the implementation of a component without seeing or caring about the UI and vice versa - designers can tweak the UI without ever having to worry about how those presentational components are receiving data.

For more on this topic, check out Presentational and Container Components.

“If you try to keep most of your component pure, stateless things become a lot simpler to maintain.”

This is another benefit of separating your presentational components from your container components. State is the sidekick of inconsistency. By drawing the right lines of separation, you’re able to drastically improve the predictability of your application by encapsulating complexity.