mobx - 为什么 mobx 需要 Action ，为什么不只是将所有突变批处理到 setImmediate 中？

Question

我开始学习 mobx，我不明白为什么 mobx 发明了“ Action ”实体。将所有更改批处理到 setImmediate 中的下一个刻度会更容易吗？这将自动使所有同步状态更改以与@action 现在相同的方式运行。在 Action 完成后而不是在下一个刻度内立即触发观察者是否有任何好处？

Answer 1

根据来自 https://github.com/mobxjs/mobx/issues/1523 的对话:

setImmediate 仅在 IE 和 Nodejs 中实现。 setTimeout(fn, 0) 会导致 4 ms 延迟

https://hackernoon.com/the-fundamental-principles-behind-mobx-7a725f71f3e8 :异步更新不会有这么好的堆栈跟踪来找到谁触发了该更新

React 仅批处理来自事件监听器的 setState 调用，mobx-react Couple 也是如此。这意味着 React 组件不会在突变后立即重新渲染，它只会在事件监听器回调后同步更新。

几个例子(所有 jsx 组件都应该在 @observer 内):

// Ok to do this
<input value={appState.name} onChange={e => appState.name = e.target.value}/>


// Ok too, will rerender only once
<input value={appState.name} onChange={e => {
    appState.name = e.target.value; 
    appState.foo = appState.foo + 1;
}}/>


// Not very good, will rerender twice on each click.
// Nevertheless, I do not know why someone will do this
<button onClick={() => {
    setTimeout(() => {
        appState.foo = appState.foo + 1;
        appState.bar = appState.bar + 10;
    }, 0);
}}>Click me foo={appState.foo} bar={appState.bar}</button>


// But that way it will rerender once and show console output only once too
autorun(() => console.info(`App state foo={appState.foo} bar={appState.bar}`));
<button onClick={() => {
    setTimeout(() => {
        runInAction(() => {
            appState.bar = appState.bar + 10;
            appState.foo = appState.foo + 1;
        })
    }, 0);
}}>Click me foo={appState.foo} bar={appState.bar}</button>


// That code will show console output twice, but rerender only once
autorun(() => console.info(`App state foo={appState.foo} bar={appState.bar}`));
<button onClick={() => {
    setTimeout(() => {
        ReactDOM.unstable_batchedUpdates(() => {
            appState.bar = appState.bar + 10;
            appState.foo = appState.foo + 1;
        })
    }, 0);
}}>Click me foo={appState.foo} bar={appState.bar}</button>