Hacker News Reader

Comment by nailer | Hacker News Reader

Svelte. Rather than

    const [name, setName] = useState('blastonico')
    ...
    setName('nailer')

You do:

   let name = 'blastonico'
   ...
   name = 'nailer'

Also a single .svelte file includes everything you need for a component (HTML, JS, CSS), so it's easier to manage.

View on HN

`setName` makes sure that state changes are queued up for the next render, allowing you to treat all state and props as constant within the current render. If you are just assigning variables directly, then your state is changing within the same render cycle, which can lead to all sorts of nasty bugs that are difficult to replicate.

I get that assigning variables is easier to understand, but it's way, way harder to actually scale and maintain. Check out flux architecture.

Just curious. How does that work on changes between depending states within a cycle? E.g.

   [foo, setFoo] = useState(1);
   [bar, setBar] = useState();
   setFoo(10);
   setBar(foo * 2);  // Is bar 2 or 20 at this point?

It’s 2. We’re just running a js function here with nothing magical going on.

It might seem that there’s a chance that people will be confused about what bar is, but in practice all the changes are happening in handlers and bar will be a const anyway so you have no plans to mutate it directly.

Thanks for the clarification. I'm trying to understand the rationals behind React's design decision. Why state changes within the same render cycle would lead to all sorts of bugs? And why is it not a problem with other frameworks like Vue or Svelte?

Just a follow up. If foo and bar are expected to be constant in the rendering code, what's the point of the setXXX() functions? Why not just make them constant explicitly and say mutating them elsewhere like in the handlers/controllers?

If you have state changes within your render cycle, then you can't write pure functions anymore, you have a stateful class. And if you have a stateful class, then React can't control when you render, because it can't know when you update a variable. So the onus is on you to decide when your component needs to re-render. This is extremely easy to screw up. You will end up in scenarios where the UI does not match the underlying state, just because the two have gone out of sync.

React inverts the MVC paradigm and says "no, you don't get to decide when to render." So if you want to update state, React will manage that for you and queue up the render. With React, your UI and state literally cannot go out of sync (except for refs), and all the data dependencies are managed for you.

Facebook gave a great talk on Flux architecture vs MVC ten years ago that you should check out here: https://www.youtube.com/watch?v=nYkdrAPrdcw

> what's the point of the setXXX() functions? Why not just make them constant explicitly

Look at the code again: it is explicitly constant! `const [myState, setMyState] = useState(0)`. `myState` is a const, and will literally not change its value for the entire render cycle. There is no mutable state to manage, no race conditions. You just tell React what to render based on what the value is now, and React injects the current value for you. The first time, `myState` is 0 since that's the initial value. Every subsequent render, React will give you the latest value. You can pretend everything is constant, because it is.

I'm glad someone who's expert in React can answer these questions. I got several thoughts.

First of all, I don't believe React component is pure functional with the state hooks embedded inside the component with side effects. Calling UI=F(props) twice with the same props would not produce the same result given the different sets of states embedded in them.

Second, the local variable myState is const but the state is mutable; the variable myState is an alias snapshot of the state. Otherwise what's the purpose of handing out setMyState? setMyState is for mutating the state, right? It seems the need to keep myState constant is due to React's inability to detect changes to the variable to sync up the UI. It uses setXXX() to trigger the UI re-rendering. However, whether setXXX() queuing up re-rendering and causing re-rending loop is inconsequential to the programmers. It's React's implementation detail.

Third, I'm interesting in learning what kind of bugs mutable state would cause in React, so that we can design better systems down the road.

> Calling UI=F(props) twice with the same props would not produce the same result given the different sets of states embedded in them.

Correct, but you as the component writer don't care. From your perspective, you are writing a purely functional component because you have no control over the injected state. So your function is "pure" with respect to the props that are passed in and the state that is injected by React. It's actually UI = F(props, state)

React can (and will) run your component function multiple times with different values before rendering, so it is extremely important that you have no actual side effects. Otherwise, these "intermediate renders" will affect the final render, which breaks the contract that your function return the same values for the same inputs.

> the local variable myState is const but the state is mutable

It is "mutable" in the sense that it changes between invocations of your function. But no, it is immutable within a single render of your function. The fact that you can treat all state as constant is one of the massive advantages of using React. You literally don't have to think about state mutating within a render cycle. You just write a function that spits out JSX given props and state.

> Third, I'm interesting in learning what kind of bugs mutable state would cause in React

You're going to run into issues where:

* the order that your function is called somehow affects the final render. Now you need to worry about how your function is called, and with what values. Not an issue with React.

* the underlying state is out of sync with the UI, because you changed it but didn't trigger a re-render. Not possible in React (except for refs).

* the internal state of the component ends up in an invalid state because of the ordering of set-state calls and renders. Again, not possible in React because all set-states are batched together and applied consistently to the next render cycle. There's no such thing as a "partial render".

> Correct, but you as the component writer don't care. From your perspective, you are writing a purely functional component because you have no control over the injected state. So your function is "pure" with respect to the props that are passed in and the state that is injected by React. It's actually UI = F(props, state)

This is like calling the constructor of an OO class a pure function. The class constructor takes in parameters (props) and the created instances don't interfere with each other's states. With all due respect, this stretching of the definition of pure function is beyond recognition.

> * the order that your function is called somehow affects the final render. Now you need to worry about how your function is called, and with what values. Not an issue with React.

The order of the function invocations calling setXX() matters anyway in React or non-React libraries. The queued up setXX() calls pending state updates in React has an order. There's no difference in handling state update ordering between React and others.

> * the underlying state is out of sync with the UI, because you changed it but didn't trigger a re-render. Not possible in React (except for refs).

> * the internal state of the component ends up in an invalid state because of the ordering of set-state calls and renders. Again, not possible in React because all set-states are batched together and applied consistently to the next render cycle. There's no such thing as a "partial render".

Both of these are because React's inability to detect changes to state updates in a comprehensive and fine grained way.

> This is like calling the constructor of an OO class a pure function.

No, because again I must stress that you cannot control the injected state. You merely ask React to set state in the next render. The next call to your function may not have the new state! React can and will call your component function multiple times with different values of state, so you must assume it could be anything. This is why component functions must be written without any side effects, because there is no way for you to know which invocation React will actually persist to the DOM. For this reason, you really are writing a pure function with respect to props and state. It's a requirement in React!

> There's no difference in handling state update ordering between React and others.

The huge difference is that because state is fixed within a render cycle, it is not possible to read an intermediate value. You can't get in a situation where one handler updates a state value, and another (or the same) handler reads that new value within the same render cycle. That's the sort of order-dependent nonsense that can make a component very buggy and fall into an inconsistent state.

> Both of these are because React's inability to detect changes to state updates in a comprehensive and fine grained way.

That's because React is built in a way to provide a set of very specific guarantees when it comes to managing state and rendering it. You are free to come up with "simpler" solutions, but you are going to have to sacrifice some of these guarantees for that simplicity. React component functions have no side effects, and are defined completely declaratively. That is extremely powerful with a lot of advantages, such as being trivially composable. It is extremely easy to extract functionality out of a component and turn it into a reusable hook, because everything is already stateless.

If you ditch all that just so you can use variables, you lose all of those benefits. You'll have to deal with truly stateful components, ones that are path dependent and difficult to reproduce for debugging. By contrast, you can reproduce any React component's UI by passing in the same props and state. This is only possible because state is external to the component and out of your control.

Thank you for the thoughtful replies. It helps a lot in understanding the kinds of problems React was facing.

With all due respect,

> For this reason, you really are writing a pure function with respect to props and state.

A pure function by definition is one which returns the same output for the same input. The output (html) of the React functional component with the same props can be different based on the changed state. It is not a pure function. It's really no different from an OO object with internal state.

> There's no difference in handling state update ordering between React and others.

> > it is not possible to read an intermediate value.

There is no intermediate value. All value changes are in the final form as there is no multi-threads reading and mutating the state. The block of code mutating the value is executed fully before another block of code can read it.

As for multiple handlers reading and rendering the state and some might miss the change by another handler, again it's a problem of React's inability to detect change for each handler. If every handler can detect change by any other code (other handlers included), there's no problem.

> That's because React is built in a way to provide a set of very specific guarantees when it comes to managing state and rendering it.

React requires that set of guarantees because its design decision - its use of diff to detect changes. It's a non-issue in other frameworks and solutions.

> A pure function by definition is one which returns the same output for the same input.

Yes, and the input includes state. React manages it and sets it before your function runs; your function then reads those values when it calls the hooks, and it is read-only. If I adjusted the React syntax to pass in state as a function argument instead, wouldn't you agree that it is pure? It's the same! Just because it's not passed in via arguments does not mean it is not an input. And a pure function is still pure even if the caller is stateful.

Consider the "rules of hooks" in React: you cannot conditionally use a hook, and hooks must be used in the same order every time. That's exactly the same as regular arguments, making them functionally identical but with better dev ergonomics. That's on purpose: state is an input, but having it passed via function arguments would suck to use.

Question: have you written any React components? This is a basic rule that is one of the first things you learn: components should be pure functions with respect to props and state. They're not even allowed to set state as a side-effect. If you're still convinced it's not pure, can you tell me how a component function's output can change given the same props and state?

> The block of code mutating the value is executed fully before another block of code can read it.

And that's the problem: with arbitrary code, there is no such thing as a distinct "block of code". One function can read value A, set it to B, then call another function, which sees value B, does something as a result, and then it renders. But that's wrong: value B has not been persisted to the DOM yet, so it should not have been read by the second function. Now you're in trouble, because your component is behaving as if both A and B are true at the same time.

This is not possible with React. But to pull that off, they had to externalize state and make it read-only to your component.

> Yes, and the input includes state. ... state is an input.

This is the same argument that the member variables of an OO object are additional input to its member method and the method is "pure."

> how a component function's output can change given the same props and state?

This is the same argument that given the same values of the member variables of an OO object and the same method arguments to its member method, the method returns the same value.

All these are just twisting the meaning of words and concepts to fit the square React into a functional round hole.

> a basic rule that is one of the first things you learn: components should be pure functions with respect to props and state.

You can certainly mutate the state in a React component; otherwise, what's the purpose of [.., setXX] = useState()? If such a "rule" is so important, why don't React remove the ability?

BTW, the "rules of hooks" are another set of extra burdens. All these rules imposed on the programmers are there due to React's weak design.

> And that's the problem: with arbitrary code, there is no such thing as a distinct "block of code". One function can read value A, set it to B, then call another function, which sees value B, does something as a result, and then it renders. But that's wrong: value B has not been persisted to the DOM yet, so it should not have been read by the second function. Now you're in trouble, because your component is behaving as if both A and B are true at the same time.

If every renderer can detect every single update of a variable/state, the above is a non-issue. Function1 can set v1 to A. The renderer detects the change to v1 and renders it as A. Function2 set v1 to B; the renderer detects it and renders it to B. Function1 and Function2 can be called in any order; the rendering output is consistent with the state. The renderer might be called twice but who cares the eventual outcome is the same (certainly there're memorizing optimization to avoid re-rendering). It's really React's weak design causing all the out-of-sync rendering problems.

`bar` is 2 because `setFoo` does not update the value of `foo` until the next render

Will bar become 20 eventually? In the next cycle perhaps?

So there will be 3 cycles?

   1: foo = 1, next(foo = 10)
      bar = undef, next(bar = 2)
   2: foo = 10
      bar = 2, next(bar = 20)
   3: foo = 10
      bar = 20

Does the above sound right?

You would not write this code for a very simple reason: all `setState` calls queue up a new render. Since you're setting the state every time the component is rendered, it will loop infinitely.

Let's say you instead did the `setFoo` and `setBar` in a click handler. In that case, when you click the button it would tell React "Hey, in the next render set `foo = 10` and `bar = 2` (since `foo = 1)`. Then React goes ahead and runs your function again, and `foo` and `bar` will both be updated.

The value of `foo` does not change within a render cycle, you literally don't have to worry about state.

Bar is 2.

> If you are just assigning variables directly

We’re not just assigning variables directly, current JavaScript frameworks use a compiler.

setName() vs name= are just syntax. None of those have anything to do with how state changes. If you have control over the compiler (which they do) there's no reason it can't do the right thing (case in point, svelte).

This is not true at all. `setName` queues up a state change for the next render cycle, and has no effect on the current render. You cannot actually edit state directly, so that multiple handlers will all read the same state value regardless of state changes queued up for the next render.

    const [state, setState] = useState(0)
    const func = () => {
      console.log(state) // 0
      setState(40)
      console.log(state) // 0
    }

let name = $state('blastonico')

I know some people didn't like this change. I think it works well to help make it clear and obvious to reactive vs non reaction changes.

Vue is also nice.

Yep. Svelte's SFC was inspired by Vue. If you like Vue you'll like Svelte which is familiar but adds a compiler for simpler syntax.

I honestly don't feel like Svelte makes things simpler. And I'm saying this as a "fan" of Svelte and think its compiler is very cool.

It just feels a bit weird that there's special .svelte specific rules.

Runes kinda fix this, but it's still a bit weird.

I think Solid or Vue are much simpler (but only did small projects in them) and have a simpler mental model. I've not yet used Preact, but heard good things about it

Yeah I like Svelte a lot, but I can't help but feel Rich feels the need to be different, and not necessarily always in a good way. SvelteKit routing[1] is an example that pops out at me as being genuinely bafflingly bad to the point of comedy.

[1] https://svelte.dev/docs/kit/routing

Overriding standard language constructs with magic behaviors is a giant no-no and a recipe for disaster.

Reply on news.ycombinator.com