The Fresnel equations describe the reflection of light when incident on an interface between different optical media.
– https://en.wikipedia.org/wiki/Fresnel_equations
yarn add @artsy/fresnel
Table of Contents
- Overview
- Basic Example
- Server-side Rendering (SSR)
- Usage with Gatsby or Next
- Example Apps
- Why not conditionally render?
- API
- Pros vs Cons
- Development
When writing responsive components it's common to use media queries to adjust the display when certain conditions are met. Historically this has taken place directly in CSS/HTML:
@media screen and (max-width: 767px) {
.my-container {
width: 100%;
}
}
@media screen and (min-width: 768px) {
.my-container {
width: 50%;
}
}
<div class="my-container" />
By hooking into a breakpoint definition, @artsy/fresnel
takes this declarative
approach and brings it into the React world.
import React from "react"
import ReactDOM from "react-dom"
import { createMedia } from "@artsy/fresnel"
const { MediaContextProvider, Media } = createMedia({
breakpoints: {
sm: 0,
md: 768,
lg: 1024,
xl: 1192,
},
})
const App = () => (
<MediaContextProvider>
<Media at="sm">
<MobileApp />
</Media>
<Media at="md">
<TabletApp />
</Media>
<Media greaterThanOrEqual="lg">
<DesktopApp />
</Media>
</MediaContextProvider>
)
ReactDOM.render(<App />, document.getElementById("react"))
The first important thing to note is that when server-rendering with
@artsy/fresnel
, all breakpoints get rendered by the server. Each Media
component is wrapped by plain CSS that will only show that breakpoint if it
matches the user's current browser size. This means that the client can
accurately start rendering the HTML/CSS while it receives the markup, which is
long before the React application has booted. This improves perceived
performance for end-users.
Why not just render the one that the current device needs? We can't accurately
identify which breakpoint your device needs on the server. We could use a
library to sniff the browser user-agent, but those aren't always accurate, and
they wouldn't give us all the information we need to know when we are
server-rendering. Once client-side JS boots and React attaches, it simply washes
over the DOM and removes markup that is unneeded, via a matchMedia
call.
First, configure @artsy/fresnel
in a Media
file that can be shared across
the app:
// Media.tsx
import { createMedia } from "@artsy/fresnel"
const ExampleAppMedia = createMedia({
breakpoints: {
sm: 0,
md: 768,
lg: 1024,
xl: 1192,
},
})
// Generate CSS to be injected into the head
export const mediaStyle = ExampleAppMedia.createMediaStyle()
export const { Media, MediaContextProvider } = ExampleAppMedia
Create a new App
file which will be the launching point for our application:
// App.tsx
import React from "react"
import { Media, MediaContextProvider } from "./Media"
export const App = () => {
return (
<MediaContextProvider>
<Media at="sm">Hello mobile!</Media>
<Media greaterThan="sm">Hello desktop!</Media>
</MediaContextProvider>
)
}
Mount <App />
on the client:
// client.tsx
import React from "react"
import ReactDOM from "react-dom"
import { App } from "./App"
ReactDOM.render(<App />, document.getElementById("react"))
Then on the server, setup SSR rendering and pass mediaStyle
into a <style>
tag in the header:
// server.tsx
import React from "react"
import ReactDOMServer from "react-dom/server"
import express from "express"
import { App } from "./App"
import { mediaStyle } from "./Media"
const app = express()
app.get("/", (_req, res) => {
const html = ReactDOMServer.renderToString(<App />)
res.send(`
<html>
<head>
<title>@artsy/fresnel - SSR Example</title>
<!–– Inject the generated styles into the page head -->
<style type="text/css">${mediaStyle}</style>
</head>
<body>
<div id="react">${html}</div>
<script src='/assets/app.js'></script>
</body>
</html>
`)
})
app.listen(3000, () => {
console.warn("\nApp started at http://localhost:3000 \n")
})
And that's it! To test, disable JS and scale your browser window down to a mobile size and reload; it will correctly render the mobile layout without the need to use a user-agent or other server-side "hints".
@artsy/fresnel
works great with Gatsby or Next.js's static hybrid approach to
rendering. See the examples below for a simple implementation.
There are four examples one can explore in the /examples
folder:
While the Basic
and SSR
examples will get one pretty far, @artsy/fresnel
can do a lot more . For an exhaustive deep-dive into its features, check out the
Kitchen Sink app.
Other existing solutions take a conditionally rendered approach, such as
react-responsive
or react-media
, so where
does this approach differ?
Server side rendering!
But first, what is conditional rendering?
In the React ecosystem a common approach to writing declarative responsive
components is to use the browser’s matchMedia
api:
<Responsive>
{({ sm }) => {
if (sm) {
return <MobileApp />
} else {
return <DesktopApp />
}
}}
</Responsive>
On the client, when a given breakpoint is matched React conditionally renders a tree.
However, this approach has some limitations for what we wanted to achieve with our server-side rendering setup:
-
It's impossible to reliably know the user's current breakpoint during the server render phase since that requires a browser.
-
Setting breakpoint sizes based on user-agent sniffing is prone to errors due the inability to precisely match device capabilities to size. One mobile device might have greater pixel density than another, a mobile device may fit multiple breakpoints when taking device orientation into consideration, and on desktop clients there is no way to know at all. The best devs can do is guess the current breakpoint and populate
<Responsive>
with assumed state.
Artsy settled on what we think makes the best trade-offs. We approach this problem in the following way:
-
Render markup for all breakpoints on the server and send it down the wire.
-
The browser receives markup with proper media query styling and will immediately start rendering the expected visual result for whatever viewport width the browser is at.
-
When all JS has loaded and React starts the rehydration phase, we query the browser for what breakpoint it’s currently at and then limit the rendered components to the matching media queries. This prevents life-cycle methods from firing in hidden components and unused html being written to the DOM.
-
Additionally, we register event listeners with the browser to notify the
MediaContextProvider
when a different breakpoint is matched and then re-render the tree using the new value for theonlyMatch
prop.
Let’s compare what a component tree using matchMedia
would look like with our
approach:
Before | After |
---|---|
<Responsive>
{({ sm }) => {
if (sm) return <SmallArticleItem {...props} />
else return <LargeArticleItem {...props} />
}}
</Responsive> |
<>
<Media at="sm">
<SmallArticleItem {...props} />
</Media>
<Media greaterThan="sm">
<LargeArticleItem {...props} />
</Media>
</> |
See the server-side rendering app for a working example.
First things first. You’ll need to define the breakpoints and interaction needed for your design to produce the set of media components you can use throughout your application.
For example, consider an application that has the following breakpoints:
- A viewport width between 0 and 768 (768 not included) points, named
sm
. - A viewport width between 768 and 1024 (1024 not included) points, named
md
. - A viewport width between 1024 and 1192 (1192 not included) points, named
lg
. - A viewport width from 1192 points and above, named
xl
.
And the following interactions:
- A device that supports hovering a pointer device, named
hover
. - A device that does not support hovering a pointer device, named
notHover
.
You would then produce the set of media components like so:
// Media.tsx
const ExampleAppMedia = createMedia({
breakpoints: {
sm: 0,
md: 768,
lg: 1024,
xl: 1192,
},
interactions: {
hover: "(hover: hover)",
notHover: "(hover: none)",
landscape: "not all and (orientation: landscape)",
portrait: "not all and (orientation: portrait)",
},
})
export const { Media, MediaContextProvider, createMediaStyle } = ExampleAppMedia
As you can see, breakpoints are defined by their start offset, where the first one is expected to start at 0.
The MediaContextProvider
component influences how Media
components will be
rendered. Mount it at the root of your component tree:
import React from "react"
import { MediaContextProvider } from "./Media"
export const App = () => {
return <MediaContextProvider>...</MediaContextProvider>
}
The Media
component created for your application has a few mutually exclusive
props that make up the API you’ll use to declare your responsive layouts. These
props all operate based on the named breakpoints that were provided when you
created the media components.
import React from "react"
import { Media } from "./Media"
export const HomePage = () => {
return (
<>
<Media at="sm">Hello mobile!</Media>
<Media greaterThan="sm">Hello desktop!</Media>
</>
)
}
The examples given for each prop use breakpoint definitions as defined in the above ‘Setup’ section.
If you would like to avoid the underlying div that is generated by <Media>
and
instead use your own element, use the render-props form but be sure to not
render any children when not necessary:
export const HomePage = () => {
return (
<>
<Media at="sm">Hello mobile!</Media>
<Media greaterThan="sm">
{(mediaClassNames, renderChildren) => {
return (
<MySpecialComponent className={mediaClassNames}>
{renderChildren ? "Hello desktop!" : null}
</MySpecialComponent>
)
}}
</Media>
</>
)
}
Note: This is only used when SSR rendering
Besides the Media
and MediaContextProvider
components, there's a
createMediaStyle
function that produces the CSS styling for all possible media
queries that the Media
instance can make use of while markup is being passed
from the server to the client during hydration. If only a subset of breakpoint
keys is used those can be optional specified as a parameter to minimize the
output. Be sure to insert this within a <style>
tag
in your document’s <head>
.
It’s advisable to do this setup in its own module so that it can be easily imported throughout your application:
import { createMedia } from "@artsy/fresnel"
const ExampleAppMedia = createMedia({
breakpoints: {
sm: 0,
md: 768,
lg: 1024,
xl: 1192,
},
})
// Generate CSS to be injected into the head
export const mediaStyle = ExampleAppMedia.createMediaStyle() // optional: .createMediaStyle(['at'])
export const { Media, MediaContextProvider } = ExampleAppMedia
Rendering can be constrained to specific breakpoints/interactions by specifying a list of media queries to match. By default all will be rendered.
By default, when rendered client-side, the browser’s matchMedia
api will be used to further constrain the onlyMatch
list
to only the currently matching media queries. This is done to avoid triggering
mount related life-cycle hooks of hidden components.
Disabling this behaviour is mostly intended for debugging purposes.
Use this to declare that children should only be visible at a specific breakpoint, meaning that the viewport width is greater than or equal to the start offset of the breakpoint, but less than the next breakpoint, if one exists.
For example, children of this Media
declaration will only be visible if the
viewport width is between 0 and 768 (768 not included) points:
<Media at="sm">...</Media>
The corresponding css rule:
@media not all and (min-width: 0px) and (max-width: 767px) {
.fresnel-at-sm {
display: none !important;
}
}
Use this to declare that children should only be visible while the viewport width is less than the start offset of the specified breakpoint.
For example, children of this Media
declaration will only be visible if the
viewport width is between 0 and 1024 (1024 not included) points:
<Media lessThan="lg">...</Media>
The corresponding css rule:
@media not all and (max-width: 1023px) {
.fresnel-lessThan-lg {
display: none !important;
}
}
Use this to declare that children should only be visible while the viewport width is equal or greater than the start offset of the next breakpoint.
For example, children of this Media
declaration will only be visible if the
viewport width is equal or greater than 1024 points:
<Media greaterThan="md">...</Media>
The corresponding css rule:
@media not all and (min-width: 1024px) {
.fresnel-greaterThan-md {
display: none !important;
}
}
Use this to declare that children should only be visible while the viewport width is equal to the start offset of the specified breakpoint or greater.
For example, children of this Media
declaration will only be visible if the
viewport width is 768 points or up:
<Media greaterThanOrEqual="md">...</Media>
The corresponding css rule:
@media not all and (min-width: 768px) {
.fresnel-greaterThanOrEqual-md {
display: none !important;
}
}
Use this to declare that children should only be visible while the viewport width is equal to the start offset of the first specified breakpoint but less than the start offset of the second specified breakpoint.
For example, children of this Media
declaration will only be visible if the
viewport width is between 768 and 1192 (1192 not included) points:
<Media between={["md", "xl"]}>...</Media>
The corresponding css rule:
@media not all and (min-width: 768px) and (max-width: 1191px) {
.fresnel-between-md-xl {
display: none !important;
}
}
Pros:
- Built on top of simple, proven technology: HTML and CSS media queries.
- Users see rendered markup at the correct breakpoint for their device, even before React has been loaded.
Cons:
- If utilizing SSR rendering features, when the markup is passed down from the server to the client it includes all breakpoints, which increases the page size. (However, once the client mounts, the unused breakpoint markup is cleared from the DOM.)
- The current media query is no longer something components can access; it is
determined only by the props of the
<Media>
component they find themselves in.
That last point presents an interesting problem. How might we represent a
component that gets styled differently at different breakpoints? (Let’s imagine
a matchMedia
example.)
<Sans size={sm ? 2 : 3}>
<>
<Media at="sm">
{this.getComponent('sm')
</Media>
<Media greaterThan="sm">
{this.getComponent()
</Media>
</>
getComponent(breakpoint?: string) {
const sm = breakpoint === 'sm'
return <Sans size={sm ? 2 : 3} />
}
We're still figuring out patterns for this, so please let us know if you have suggestions.
This project uses auto-release to automatically release on every PR. Every PR should have a label that matches one of the following
- Version: Trivial
- Version: Patch
- Version: Minor
- Version: Major
Major, minor, and patch will cause a new release to be generated. Use major for breaking changes, minor for new non-breaking features, and patch for bug fixes. Trivial will not cause a release and should be used when updating documentation or non-project code.
If you don't want to release on a particular PR but the changes aren't trivial
then use the Skip Release
tag along side the appropriate version tag.