下面是我使用的一个自定义钩子，我认为它比使用usePrevious更直观。

import { useRef, useEffect } from 'react'

// useTransition :: Array a => (a -> Void, a) -> Void
//                              |_______|  |
//                                  |      |
//                              callback  deps
//
// The useTransition hook is similar to the useEffect hook. It requires
// a callback function and an array of dependencies. Unlike the useEffect
// hook, the callback function is only called when the dependencies change.
// Hence, it's not called when the component mounts because there is no change
// in the dependencies. The callback function is supplied the previous array of
// dependencies which it can use to perform transition-based effects.
const useTransition = (callback, deps) => {
  const func = useRef(null)

  useEffect(() => {
    func.current = callback
  }, [callback])

  const args = useRef(null)

  useEffect(() => {
    if (args.current !== null) func.current(...args.current)
    args.current = deps
  }, deps)
}

您可以像下面那样使用useTransition。

useTransition((prevRate, prevSendAmount, prevReceiveAmount) => {
  if (sendAmount !== prevSendAmount || rate !== prevRate && sendAmount > 0) {
    const newReceiveAmount = sendAmount * rate
    // do something
  } else {
    const newSendAmount = receiveAmount / rate
    // do something
  }
}, [rate, sendAmount, receiveAmount])

希望这能有所帮助。

下面是Aadit M Shah的答案的Typescript版本。

我把它从useTransition重命名为usePrevious，因为useTransition已经存在于React中。

import { useEffect, useRef, useState } from 'react';

const usePrevious = <T extends any[],>(callback: (prev: T) => void, deps: T): void => {
  const callbackRef = useRef<null | ((prev: T) => void)>(null);

  useEffect(() => {
    callbackRef.current = callback;
  }, [callback]);

  const depsRef = useRef<null | T>(null);

  const [initial, setInitial] = useState(true);

  useEffect(() => {
    if (!initial && depsRef.current !== null && callbackRef.current !== null) {
      callbackRef.current(depsRef.current);
    }

    depsRef.current = deps;
    setInitial(false);
  }, deps);
}

export default usePrevious;

用法:

  usePrevious<[boolean]>(([prevIsOpen]) => {
    console.log('prev', prevIsOpen);
    console.log('now', isOpen);
  }, [isOpen])

选项1 -当值改变时运行useEffect

const Component = (props) => {

  useEffect(() => {
    console.log("val1 has changed");
  }, [val1]);

  return <div>...</div>;
};

Demo

选项2 - useHasChanged钩子

比较当前值和以前值是一种常见的模式，它证明了它自己的自定义钩子隐藏了实现细节。

const Component = (props) => {
  const hasVal1Changed = useHasChanged(val1)

  useEffect(() => {
    if (hasVal1Changed ) {
      console.log("val1 has changed");
    }
  });

  return <div>...</div>;
};

const useHasChanged= (val: any) => {
    const prevVal = usePrevious(val)
    return prevVal !== val
}

const usePrevious = (value) => {
    const ref = useRef();
    useEffect(() => {
      ref.current = value;
    });
    return ref.current;
}

Demo

由于状态与功能组件中的组件实例不是紧密耦合的，因此如果不先保存之前的状态，例如使用useRef，则无法在useEffect中到达之前的状态。这也意味着状态更新可能在错误的位置被错误地实现，因为之前的状态在setState更新函数中可用。

这是useReducer的一个很好的用例，它提供了类似redux的存储，并允许实现各自的模式。状态更新是显式执行的，因此不需要确定更新的是哪个状态属性;这在已调度的行动中已经很清楚了。

下面是一个例子:

function reducer({ sendAmount, receiveAmount, rate }, action) {
  switch (action.type) {
    case "sendAmount":
      sendAmount = action.payload;
      return {
        sendAmount,
        receiveAmount: sendAmount * rate,
        rate
      };
    case "receiveAmount":
      receiveAmount = action.payload;
      return {
        sendAmount: receiveAmount / rate,
        receiveAmount,
        rate
      };
    case "rate":
      rate = action.payload;
      return {
        sendAmount: receiveAmount ? receiveAmount / rate : sendAmount,
        receiveAmount: sendAmount ? sendAmount * rate : receiveAmount,
        rate
      };
    default:
      throw new Error();
  }
}

function handleChange(e) {
  const { name, value } = e.target;
  dispatch({
    type: name,
    payload: value
  });
}

...
const [state, dispatch] = useReducer(reducer, {
  rate: 2,
  sendAmount: 0,
  receiveAmount: 0
});
...

在你的例子中(简单对象):

useEffect(()=>{
  // your logic
}, [rate, sendAmount, receiveAmount])

在其他情况下(复杂对象)

const {cityInfo} = props;
useEffect(()=>{
  // some logic
}, [cityInfo.cityId])

我刚刚发布了react-delta，它解决了这种情况。在我看来，useEffect有太多的责任。

责任

它使用Object.is比较其依赖数组中的所有值 它基于#1的结果运行effect/cleanup回调

分解责任

react-delta将useEffect的职责分解为几个更小的钩子。

责任# 1

usePrevious(值) useLatest(值) useDelta(价值,选项) useDeltaArray (valueArray选项) useDeltaObject (valueObject选项) 一些(deltaArray) 每一个(deltaArray)

责任# 2

布尔useConditionalEffect(回调)

根据我的经验，这种方法比useEffect/useRef解决方案更灵活、干净和简洁。