So I have some confusion regarding the async nature of setState in ReactJS.
As per React docs, you shouldn't use this.state inside setState(). But if I have a counter as a state and i want to update it on click like this:
class App extends React.Component {
state = { counter: 0 }
onClick = () => {
this.setState({counter: this.state.counter + 1})
}
render() {
return (
<div>
<div>{this.state.counter}</div>
<p onClick={this.onClick}>Click me</p>
</div>
)
}
}
This works as expected. So why is this code wrong?
UPDATE: I know that setState is async, and it accepts a callback which has previous state as an argument, but I am not sure why I should use it here? I want to refer to the old state inside setState, so why should I use the callback function in this case? Whenever this.setState() is executed, this.state inside it will always refer to the old state, and its value will be changed to the new state only after setState has finished executing, not while it is executing.
You have access to prevState from within your setState call:
this.setState((prevState) => ({
counter: prevState.counter +1
}))
That will allow you to safely increment the current state value.
The React documentation summarises why you cannot rely on this.state to be accurate during update: https://reactjs.org/docs/state-and-lifecycle.html#state-updates-may-be-asynchronous
setState accepts a function as a parameter with previous state as argument.
Refacto as follows to avoid competition problems:
onClick = () => {
this.setState(prevState => ({counter: prevState.counter + 1}))
}
I have component Inbox and having checkbox in it... But It works on third click... does not work on first and second click... setState works perfect but it does't re-render after setState
class Inbox extends PureComponent {
constructor(props){
this.state = {
checked: [true]
}
}
updateCheck(i, convId) {
const state = this.state.checked
state[i] = !state[i]
this.setState(state)
}
render() {
return (
<input type="checkbox" checked={this.state.checked[i]} onClick={() => this.updateCheck(i, conv._id)}/>
)
}
}
You are not really updating the state correctly. setting state like
this.setState(state, () => {
console.log(this.state, '787878787878778787')
})
does not update the checked state using state but adds keys with array indices to state like
{0: true, 1: false, conversationId: '', checked: [true, false]};
You are instead mutating the checked state yourself using
state[i] = !state[i]
To update the state correctly, you would write
updateCheck(i, convId) {
const checked = [...this.state.checked]
checked[i] = !checked[i]
this.setState({ checked }, () => {
console.log(this.state, '787878787878778787')
})
}
The problem in your approach arises because you mutate the original state directly, subsequent setState calls may replace the original change and hence you see that behaviour.
According to documentation
Never mutate this.state directly, as calling setState() afterwards may
replace the mutation you made. Treat this.state as if it were
immutable.
Working demo
This question already has answers here:
Why calling setState method doesn't mutate the state immediately?
(3 answers)
Closed 4 years ago.
I have simple component
class App extends Component {
handleClick() {
let banana = {message: 'banana triggered'};
this.setState({banana});
console.log(this); // banana is set in state!!!!
console.log(this.state); // state is null :O
setTimeout(() => {
console.log(this.state); // banana is set!
}, 5)
}
render() {
const {state, actions} = this.props;
return (
<div>
{this.state && this.state.banana.message} <br />
<button onClick={() => this.handleClick()}>Test</button>
{state.alert.message && <p>{state.alert.message}</p>}
<p onClick={() => actions.alert.success("This is not")}>
This is magic
</p>
</div>
)
};
}
export default connect(
state => (
{
state: {...state}
}
),
dispatch => (
{
actions: {
dispatch: dispatch,
alert: {
success: text => dispatch(alert.success(text))
}
}
}
)
)(App);
problem is what i need to add this.state && in my JSX rendering to check if this.state exists at all, i understand what in JavaScript it's normal, but is not normal in React.js? Should he react to state change instantly? Also what get me confused, is what from two console.logs, first (this) have banana set in state, and second one is empty. How?
Image below:
p.s. there is no such problem with Redux, only local component state
react's docs mention that state updates are asynchronous.
In order to act based on the change of the state, react setState function provides a callback which you can use as follows:
this.setState({banana}, () => {
console.log(this.state);
});
In regards to your comment, the value of the state didn't actually exist when it was printed. the value was calculated only after you clicked the expand arrow in the console see this for more deatils
According to react docs, setState() is asynchronous, and multiple calls during the same cycle may be batched together.
If you check the updated state value, you can add a callback method
this.setState({ banana }, ()=> {
// console.log(this.state);
// Here's the updated state
});
In your case, the first console.log(this) doesn't set the banana. See your code in Sandbox. It looks like first two console logs don't show any state as the initial state is null and after the timeout when the asynchronous call has finished it set the state with banana.
I need to use current props and previous props value in my React component.
So i did it like this
state = {
current: null,
previous: null,
};
componentWillReceiveProps(nextProps) {
if (nextProps.amount !== this.state.current) {
this.setState({previous: this.state.current, current: nextProps.amount});
}
}
...
render() {
const {previous, current} = this.state;
return (
...
<CountUp className="counter" start={previous} end={current} duration={1}/>
...
)
}
It works fine, but is it good React practise to do it like this? Are there others "good" ways to do it?
As of v16.2.0, componentWillReceiveProps is the right place to update state, based on prop changes and since you want to use both current state and previous state in render, you need to maintain, two different state variables as you are doing
However, when you update the state based on previous state, use functional setState method
Check this answer for more details
When to use functional setState
componentWillReceiveProps(nextProps) {
if (nextProps.amount !== this.state.current) {
this.setState(prevState => ({ previous: prevState.current, current: nextProps.amount }));
}
}
According to the latest RFC to React
State derived from props/state
The purpose of this pattern is to calculate some values derived from props for use during render.
Typically componentWillReceiveProps is used for this, although if the calculation is fast enough it could just be done in render.:
From v16.3.0 onwards, you would make use of
static getDerivedStateFromProps(nextProps, prevState) {
if (
!prevState ||
prevState.current !== nextProps.amount
) {
return {
previous: prevState.current,
current: nextProps.amount
};
}
}
I'd like to update this answer for anyone else who comes here from Google. As of v16.8.6, componentWillReceiveProps has been marked as legacy, and is not recommended to use. Instead you should use componentDidUpdate and update your state based on the new props and previous props/previous state.
componentDidUpdate(prevProps, prevState) {
if (this.props.someVal !== prevState.someVal) {
this.setState({ previous: prevState.someVal, current: this.props.someVal });
}
}
Obviously, whether you check the previous state or the previous props is up to your discretion/situation. You can implement componentDidUpdate with or without the parameters, but if you want prevState you must declare prevProps.
React Update Lifecycle
componentDidUpdate()
You can use an arrow function in your setState object.
Like this :
this.setState((prevState) => {
return { yourState: prevState.yourState }
})
prevState is a default name but you can replace the name as you want
I have the following state:
this.setState({ selected: { id: 1, name: 'Foobar' } });
Then I update the state:
this.setState({ selected: { name: 'Barfoo' }});
Since setState is suppose to merge I would expect it to be:
{ selected: { id: 1, name: 'Barfoo' } };
But instead it eats the id and the state is:
{ selected: { name: 'Barfoo' } };
Is this expected behavior and what's the solution to update only one property of a nested state object?
I think setState() doesn't do recursive merge.
You can use the value of the current state this.state.selected to construct a new state and then call setState() on that:
var newSelected = _.extend({}, this.state.selected);
newSelected.name = 'Barfoo';
this.setState({ selected: newSelected });
I've used function _.extend() function (from underscore.js library) here to prevent modification to the existing selected part of the state by creating a shallow copy of it.
Another solution would be to write setStateRecursively() which does recursive merge on a new state and then calls replaceState() with it:
setStateRecursively: function(stateUpdate, callback) {
var newState = mergeStateRecursively(this.state, stateUpdate);
this.replaceState(newState, callback);
}
Immutability helpers were recently added to React.addons, so with that, you can now do something like:
var newState = React.addons.update(this.state, {
selected: {
name: { $set: 'Barfoo' }
}
});
this.setState(newState);
Immutability helpers documentation.
Since many of the answers use the current state as a basis for merging in new data, I wanted to point out that this can break. State changes are queued, and do not immediately modify a component's state object. Referencing state data before the queue has been processed will therefore give you stale data that does not reflect the pending changes you made in setState. From the docs:
setState() does not immediately mutate this.state but creates a pending state transition. Accessing this.state after calling this method can potentially return the existing value.
This means using "current" state as a reference in subsequent calls to setState is unreliable. For example:
First call to setState, queuing a change to state object
Second call to setState. Your state uses nested objects, so you want to perform a merge. Before calling setState, you get current state object. This object does not reflect queued changes made in first call to setState, above, because it's still the original state, which should now be considered "stale".
Perform merge. Result is original "stale" state plus new data you just set, changes from initial setState call are not reflected. Your setState call queues this second change.
React processes queue. First setState call is processed, updating state. Second setState call is processed, updating state. The second setState's object has now replaced the first, and since the data you had when making that call was stale, the modified stale data from this second call has clobbered the changes made in the first call, which are lost.
When queue is empty, React determines whether to render etc. At this point you will render the changes made in the second setState call, and it will be as though the first setState call never happened.
If you need to use the current state (e.g. to merge data into a nested object), setState alternatively accepts a function as an argument instead of an object; the function is called after any previous updates to state, and passes the state as an argument -- so this can be used to make atomic changes guaranteed to respect previous changes.
I didn't want to install another library so here's yet another solution.
Instead of:
this.setState({ selected: { name: 'Barfoo' }});
Do this instead:
var newSelected = Object.assign({}, this.state.selected);
newSelected.name = 'Barfoo';
this.setState({ selected: newSelected });
Or, thanks to #icc97 in the comments, even more succinctly but arguably less readable:
this.setState({ selected: Object.assign({}, this.state.selected, { name: "Barfoo" }) });
Also, to be clear, this answer doesn't violate any of the concerns that #bgannonpl mentioned above.
Preserving the previous state based on #bgannonpl answer:
Lodash example:
this.setState((previousState) => _.merge({}, previousState, { selected: { name: "Barfood"} }));
To check that it's worked properly, you can use the second parameter function callback:
this.setState((previousState) => _.merge({}, previousState, { selected: { name: "Barfood"} }), () => alert(this.state.selected));
I used merge because extend discards the other properties otherwise.
React Immutability example:
import update from "react-addons-update";
this.setState((previousState) => update(previousState, {
selected:
{
name: {$set: "Barfood"}
}
});
As of right now,
If the next state depends on the previous state, we recommend using
the updater function form, instead:
according to documentation https://reactjs.org/docs/react-component.html#setstate, using:
this.setState((prevState) => {
return {quantity: prevState.quantity + 1};
});
My solution for this kind of situation is to use, like another answer pointed out, the Immutability helpers.
Since setting the state in depth is a common situation, I've created the folowing mixin:
var SeStateInDepthMixin = {
setStateInDepth: function(updatePath) {
this.setState(React.addons.update(this.state, updatePath););
}
};
This mixin is included in most of my components and I generally do not use setState directly anymore.
With this mixin, all you need to do in order to achieve the desired effect is to call the function setStateinDepth in the following way:
setStateInDepth({ selected: { name: { $set: 'Barfoo' }}})
For more information:
On how mixins work in React, see the official documentation
On the syntax of the parameter passed to setStateinDepth see the Immutability Helpers documentation.
I am using es6 classes, and I ended up with several complex objects on my top state and was trying to make my main component more modular, so i created a simple class wrapper to keep the state on the top component but allow for more local logic.
The wrapper class takes a function as its constructor that sets a property on the main component state.
export default class StateWrapper {
constructor(setState, initialProps = []) {
this.setState = props => {
this.state = {...this.state, ...props}
setState(this.state)
}
this.props = initialProps
}
render() {
return(<div>render() not defined</div>)
}
component = props => {
this.props = {...this.props, ...props}
return this.render()
}
}
Then for each complex property on the top state, i create one StateWrapped class. You can set the default props in the constructor here and they will be set when the class is initialised, you can refer to the local state for values and set the local state, refer to local functions, and have it passed up the chain:
class WrappedFoo extends StateWrapper {
constructor(...props) {
super(...props)
this.state = {foo: "bar"}
}
render = () => <div onClick={this.props.onClick||this.onClick}>{this.state.foo}</div>
onClick = () => this.setState({foo: "baz"})
}
So then my top level component just needs the constructor to set each class to it's top level state property, a simple render, and any functions that communicate cross-component.
class TopComponent extends React.Component {
constructor(...props) {
super(...props)
this.foo = new WrappedFoo(
props => this.setState({
fooProps: props
})
)
this.foo2 = new WrappedFoo(
props => this.setState({
foo2Props: props
})
)
this.state = {
fooProps: this.foo.state,
foo2Props: this.foo.state,
}
}
render() {
return(
<div>
<this.foo.component onClick={this.onClickFoo} />
<this.foo2.component />
</div>
)
}
onClickFoo = () => this.foo2.setState({foo: "foo changed foo2!"})
}
Seems to work quite well for my purposes, bear in mind though you can't change the state of the properties you assign to wrapped components at the top level component as each wrapped component is tracking its own state but updating the state on the top component each time it changes.
Solution
Edit: This solution used to use spread syntax. The goal was make an object without any references to prevState, so that prevState wouldn't be modified. But in my usage, prevState appeared to be modified sometimes. So, for perfect cloning without side effects, we now convert prevState to JSON, and then back again. (Inspiration to use JSON came from MDN.)
Remember:
State updates are shallow: they only go one level deep
State shouldn't be directly mutated. Instead, use setState(prevState => stateChange)
Steps
Make a copy of the root-level property of state that you want to change
Mutate this new object
Create an update object
Return the update
Steps 3 and 4 can be combined on one line.
Example
this.setState(prevState => {
var newSelected = JSON.parse(JSON.stringify(prevState.selected)) //1
newSelected.name = 'Barfoo'; //2
var update = { selected: newSelected }; //3
return update; //4
});
Simplified example:
this.setState(prevState => {
var selected = JSON.parse(JSON.stringify(prevState.selected)) //1
selected.name = 'Barfoo'; //2
return { selected }; //3, 4
});
This follows the React guidelines nicely. Based on eicksl's answer to a similar question.
ES6 solution
We set the state initially
this.setState({ selected: { id: 1, name: 'Foobar' } });
//this.state: { selected: { id: 1, name: 'Foobar' } }
We are changeing a property on some level of the state object:
const { selected: _selected } = this.state
const selected = { ..._selected, name: 'Barfoo' }
this.setState({selected})
//this.state: { selected: { id: 1, name: 'Barfoo' } }
React state doesn't perform the recursive merge in setState while expects that there won't be in-place state member updates at the same time. You either have to copy enclosed objects/arrays yourself (with array.slice or Object.assign) or use the dedicated library.
Like this one. NestedLink directly supports handling of the compound React state.
this.linkAt( 'selected' ).at( 'name' ).set( 'Barfoo' );
Also, the link to the selected or selected.name can be passed everywhere as a single prop and modified there with set.
have you set the initial state?
I'll use some of my own code for example:
getInitialState: function () {
return {
dragPosition: {
top : 0,
left : 0
},
editValue : "",
dragging : false,
editing : false
};
}
In an app I'm working on, this is how I've been setting and using state. I believe on setState you can then just edit whatever states you want individually I've been calling it like so:
onChange: function (event) {
event.preventDefault();
event.stopPropagation();
this.setState({editValue: event.target.value});
},
Keep in mind you have to set the state within the React.createClass function that you called getInitialState
I use the tmp var to change.
changeTheme(v) {
let tmp = this.state.tableData
tmp.theme = v
this.setState({
tableData : tmp
})
}