DEV Community

Cover image for What are React Server Components (RSCs)?
Corbin Crutchley for This is Learning

Posted on • Originally published at unicorn-utterances.com

What are React Server Components (RSCs)?

In our last article in the series, we talked about how React is able to pre-generate HTML from JSX on the server (or developer's machine) prior to being shipped to the end-user. This process is called "SSR" and can be mission-critical in getting your applications as performant as possible.

I originally wrote that article in early 2020. At this point in React's development lifecycle, there was an inherent problem to using SSR; it would lead to duplicate effort between the server and the client.

See, up to that point Next.js and other React SSR solutions had one way of doings things:

1) Render the VDOM on the server
2) Generate HTML from the server's VDOM
3) Ship HTML and all of the React code to the client
4) Re-generate the VDOM from scratch on the client
5) Wipe away the old DOM and re-render all components from the new client's VDOM instance

The developer ships SSR and framework code to the server, which produces HTML. This HTML/CSS is then sent to the user machine where it re-initializes on the client's browser

This process is called "Hydration" and while it worked the way it did before, it introduced a new performance problem. Hydration could be needlessly expensive if most of your content coming from the server was going to be static anyway. This was a huge problem that the React team had to solve.

Later in the same year of my article (December 2020), they had the answer: React Server Components.

What is a React Server Component (RSC)?

In short; React Server components allow you to designate which React components render on the server and which components re-initialize on the client. This is done using a special syntax in your components and allows for special server-only data loading patterns when implemented properly.

For example, take the following:

function App() {
    return <>
        {/* Render parts on the client */}
        <Header/>
        {/* Render all of it on the client */}
        <Dashboard/>
        {/* Render all of it on the server */}
        <Footer/>
    </>
}
Enter fullscreen mode Exit fullscreen mode

When executed, it might look something like the following process:

The developer authors JSX with distinct client and server components. These components are ALL rendered on the server, but only the client components are re-rendered on the client

Here, we can see that <ProfilePicture>, <Dashboard/>, and all of their children will be re-initialized on the client. Meanwhile the <Footer> and <Header> components will not re-initialize on the client.

Keep in mind, client-components will still pre-generate HTML on the server by default. The difference here is that the client re-initialization is now informed by the VDOM constructed on the server, allowing for drastically reduced required execution.

What is "use client"?

In React Server Components, "use client" is a string at the top of the file or function that indicates that React should also render the component on the client. Rendering on the client side allows you to use state and rerender reactively. Without "use client", the component will be treated as a server component.

Let's use this "use client" syntax to build out the example from the image in the previous section, distinguishing which type of component is which along the way:

// page.jsx
import {ProfilePicture, Dashboard} from "./client-components"

export function App() {
    return <>
        <Header/>
        <Dashboard/>
        <Footer/>
    </>
}

function Footer() {
    // ...
}

function Header() {
    return (
        <>
            <SearchBar/>
            <ProfilePicture/>
        </>
    )
}


function SearchBar() {
    // ...
}
Enter fullscreen mode Exit fullscreen mode
"use client"
// client-components.jsx
export function Dashboard() {
    return (
        <Chart/>
    )
}


function Chart() {
    // ...
}

export function profilePicture() {
    // ...
}
Enter fullscreen mode Exit fullscreen mode

Limitations of Server Components

Because a server component runs entirely on the server, there are a few limitations you should be aware of:

  • No usage of React Hooks that contain client-side state (useState, useReducer, etc)

  • <Context.Provider> usage

  • No usage of browser APIs (localstorage, querySelector, etc)

  • You cannot pass the following property values from a server component to a client component:

    • React Elements/JSX (component={<Component/>} or component={Component})
    • Functions (unless it's a Server Action - more on that in a future article)
    • Classes
    • Instances of Custom Classes
    • Custom Symbols
  • Cannot be called inside a Client component. IE:

  const ClientComponent = () => {
    return <ServerComponent/>
  }
Enter fullscreen mode Exit fullscreen mode

Is not allowed but:

  const ClientComponent = ({children}) => {
    return <div>{children}</div>
  }

  const App = () => {
    return <ClientComponent><ServerComponent/></ClientComponent>
  }
Enter fullscreen mode Exit fullscreen mode

Is allowed.

Conclusion

React Server Components have been a huge topic of discussion lately. So much so that while the fixes to hydration are useful, you may be left wondering:

Is that all there is to RSCs?

Luckily, it's not!

See, by providing a primitive for server behavior in React, the team has unlocked a great deal of potential in regards to data loading.

But that's a story for next time. Next up? React's Suspense APIs!

Top comments (1)

Collapse
 
efpage profile image
Eckehard • Edited

"No usage of..." looks like a veritable minefield for developers

Image description

Means: you cannot simply use SSR with React, but need to build special apps for RSC. Does this make any sense?