useEffect()
can combine componentDidMount
, componentDidUpdate
, and componentWillUnmount
but is tricky.
You can decipher most of what I have to discuss here from official docs for hooks. It’s easier to see hooks at work, than to reason from text.
Pre-render lifecycle events equivalent to componentWillReceiveProps
or getDerivedStateFromProps
and componentWillMount
can just be the things we do first in the functional component before returning JSX (react-node) as the function itself is equivalent to render(…)
method of class based component.
We don’t need hooks handling Pre-render lifecycle events.
Post-render lifecycle events, those equivalent to componentDidMount
, componentDidUpdate
and componentWillUnmount
in class based component.
We need to **useEffect(…)
to handle these Post-render lifecycle events** as we can’t write the logic tied to these lifecycle events inside the main component function as these should run after the component function returns JSX (react-node) to react-dom
renderer.
This means, we have lot we can do with hooks. How?
We know useEffect(fn, […watchStates])
, takes in 2 arguments.
fn
: (required) useEffect
invokes this function to run as side-effect after every render cycle based on values being tracked for changes given by the (2) argument. The function fn
, could return another function that should be run as a cleanup before the effect function runs again or component un-mounts[…watchValues ]
: (optional) useEffect
tracks values in this array has changed from the last render cycle then only effect fn
is invoked. If this argument is not given, the effect will run with every render cycle.If we don’t pass the (2) argument all-together, the effect logic in fn
will be invoked after every render cycle.
If we pass (2) array with values the component needs to watch for changes, and invoke fn
on change, pretty self explanatory.
The trickiest part is in using an empty array []
as the (2) argument, we can restrict side-effect logic in fn
to execute only during the mounting phase as there are no changes effect hook would be watching for after subsequent render-cycles to trigger fn
again.
import React, { useState, useEffect } from "react";
export default props => {
console.log("componentWillMount");
console.log("componentWillReceiveProps", props);
const [x, setX] = useState(0);
const [y, setY] = useState(0);
const [moveCount, setMoveCount] = useState(0);
const [cross, setCross] = useState(0);
const mouseMoveHandler = event => {
setX(event.clientX);
setY(event.clientY);
};
useEffect(() => {
console.log("componentDidMount");
document.addEventListener("mousemove", mouseMoveHandler);
return () => {
console.log("componentWillUnmount");
document.removeEventListener("mousemove", mouseMoveHandler);
};
}, []); // empty-array means don't watch for any updates
useEffect(
() => {
// if (componentDidUpdate & (x or y changed))
setMoveCount(moveCount + 1);
},
[x, y]
);
useEffect(() => {
// if componentDidUpdate or componentDidMount
if (x === y) {
setCross(x);
}
});
return (
<div>
<p style={{ color: props.color }}>
Your mouse is at {x}, {y} position.
</p>
<p>Your mouse has moved {moveCount} times</p>
<p>
X and Y positions were last equal at {cross}, {cross}
</p>
</div>
);
};
The code snippet is simple and self explanatory. You can try it out on code-pen.
One important thing to note is that if you are making a state change inside a effect, ensure you exclude the state that’s changing inside from the watch array.
For example in the second effect (one that counts the mouse movements) we only trigger it on updates on x and y, by passing [x , y]
as the second argument because
This article is also aviable on my Medium publication. If you like the artile, or have any comments and suggestions, please clap or leave comments on Medium.