I agree your way doing this works perfectly and you can even wrap it in a custom hook to reuse it.
Having said that (admitting there are cases where you don't want to run a stuff on mount but still on all re-renders !), is this hack improving our code ?
According to React guys, hooks allows us to separate concerns and avoid us to write stuff like :
componentDidMount(){// concern 1// concern 2// concern 3}componentDidUpdate(prevProps){if(prevProps.xxx!==this.props.xxx){// concern 2}if(prevProps.yyy!==this.props.yyy){// concern 3}// concern 4 (on each re-render because we can :) )}
which is imperative, redundant and difficult to simplify. Effects are made to be used atomically, to maintain good readability :
If the effect concernX have to be trigged only on updates right after any user action or async stuff, IMO you better should rely on your component scope (props, states, sometimes ref or context, ...). Ex :
const[selectedColor,setSelectedColor]=useState('blank');// or null or whateveruseEffect(()=>{if(selectedColor!=='blank'){// selectedCorolized unicorns popping}},[...]);
For readability purposes you can even put this in a custom hook and reuse the concern anywhere you want.
If the effect concernX have to be trigged only on updates right after any user action or async stuff, IMO you better should rely on your component scope (props, states, sometimes ref or context, ...). Ex:
So, if I'm understanding right what you're saying and especially this example, what you propose is to use something that already exists and that we're already monitoring to decide whether an effect hook should execute or not (and play with how that something has been initialised), rather than creating a specific reference to know whether the component has been mounted or not. Is that so?
Exactly, rely on data values ans data changes instead of component lifecycle. It seems to be the way hooks are designed.
Actually all your props should be observed and changes propagated in your component render flow, it's a really different approach than class components. Dan Abramov made a really detailed blog post about that here : overreacted.io/a-complete-guide-to...
Just an apetizer :
It’s only after I stopped looking at the useEffect Hook through the prism of the familiar class lifecycle methods that everything came together for me.
I'm still struggling with this approach. I have a really basic Class Component that calls an action creator (which is passed as a prop) if a particular prop is true and if another prop has changed, but only if that second prop isn't being changed for the first time.
The suggestion in the article by @savagepixie
got me around my issue, but I can't see how I could use @devdufutur
's suggestion (which sounds sensible but unachievable in my use case!):
So, would you care to provide a quick example of how to make
useEffect
not execute after the first render but execute on subsequent renders?Or are you suggesting that one should (and indeed can) always find an alternative way that doesn't require to skip the
useEffect
on the first render?I agree your way doing this works perfectly and you can even wrap it in a custom hook to reuse it.
Having said that (admitting there are cases where you don't want to run a stuff on mount but still on all re-renders !), is this hack improving our code ?
According to React guys, hooks allows us to separate concerns and avoid us to write stuff like :
which is imperative, redundant and difficult to simplify. Effects are made to be used atomically, to maintain good readability :
If the effect concernX have to be trigged only on updates right after any user action or async stuff, IMO you better should rely on your component scope (props, states, sometimes ref or context, ...). Ex :
For readability purposes you can even put this in a custom hook and reuse the concern anywhere you want.
So, if I'm understanding right what you're saying and especially this example, what you propose is to use something that already exists and that we're already monitoring to decide whether an effect hook should execute or not (and play with how that something has been initialised), rather than creating a specific reference to know whether the component has been mounted or not. Is that so?
Exactly, rely on data values ans data changes instead of component lifecycle. It seems to be the way hooks are designed.
Actually all your props should be observed and changes propagated in your component render flow, it's a really different approach than class components. Dan Abramov made a really detailed blog post about that here : overreacted.io/a-complete-guide-to...
Just an apetizer :
It’s only after I stopped looking at the useEffect Hook through the prism of the familiar class lifecycle methods that everything came together for me.
“Unlearn what you have learned.” — Yoda
I like that approach
I'm still struggling with this approach. I have a really basic Class Component that calls an action creator (which is passed as a prop) if a particular prop is true and if another prop has changed, but only if that second prop isn't being changed for the first time.
The suggestion in the article by @savagepixie got me around my issue, but I can't see how I could use @devdufutur 's suggestion (which sounds sensible but unachievable in my use case!):
Any suggestions?
I guess your RequireLogon component could have been a custom hook...
I like that, primarily because what I had wasn't a "component" as such, but some sort of rule. This is certainly cleaner code. Thanks!