但这是非常常见的用法，因为这样做可以保证在逻辑执行完毕后再执行，是保证时间间隔的常用手段。

However, this is a very common usage, as it ensures that the execution will occur after the logic has been completed, which is a common means of guaranteeing time intervals.

但这是非常常见的用法，因为这样做可以保证在逻辑执行完毕后再执行，是保证时间间隔的常用手段。

However, this is a very common usage, as it ensures that the execution will occur after the logic has been completed, which is a common means of guaranteeing time intervals.

I think this is not "common" usage, you are calling test recursively, after first timeout, the second running test calling useTimeoutFn out side of setup scope, making these useTimeoutFn can't clear on dispose. maybe you want use useIntervalFn

这是前端轮询请求后端的常用方式，可以有效防止雪崩现象，我必须保证服务器响应之后的几秒再次发生心跳。如果没有响应一直发送会导致大量消息阻塞服务端，带来了不必要的开销，这对后端服务来讲是非常不友好的。然而使用 setTimeout 的递归调用可以有效防止这种情况发生，最起码可以做到自己不去攻击自己。当然，如果技术上有难度可以在文档中说明不可以这样使用，这样使用会导致无法回收，当有类似需求的时候可以自己去实现。

This is a common way for front-end polling to request back-end, which can effectively prevent avalanche phenomenon. I must ensure that the heartbeat occurs again within a few seconds after the server responds. If there is no response, sending continuously will cause a large number of messages to block the server, resulting in unnecessary overhead, which is very unfriendly to backend services. However, using the recursive call 'setTimeout' can effectively prevent this situation from happening, at least by avoiding attacking oneself. Of course, if there are technical difficulties, it can be stated in the documentation that such use is not allowed, as it will result in the inability to recycle. When there is a similar requirement, it can be implemented by oneself.

这是前端轮询请求后端的常用方式，可以有效防止雪崩现象，我必须保证服务器响应之后的几秒再次发生心跳。如果没有响应一直发送会导致大量消息阻塞服务端，带来了不必要的开销，这对后端服务来讲是非常不友好的。然而使用 setTimeout 的递归调用可以有效防止这种情况发生，最起码可以做到自己不去攻击自己。当然，如果技术上有难度可以在文档中说明不可以这样使用，这样使用会导致无法回收，当有类似需求的时候可以自己去实现。

This is a common way for front-end polling to request back-end, which can effectively prevent avalanche phenomenon. I must ensure that the heartbeat occurs again within a few seconds after the server responds. If there is no response, sending continuously will cause a large number of messages to block the server, resulting in unnecessary overhead, which is very unfriendly to backend services. However, using the recursive call 'setTimeout' can effectively prevent this situation from happening, at least by avoiding attacking oneself. Of course, if there are technical difficulties, it can be stated in the documentation that such use is not allowed, as it will result in the inability to recycle. When there is a similar requirement, it can be implemented by oneself.

All vue3 composables function in component setup should be sync, this is pre restriction, not lack of doc. https://cn.vuejs.org/guide/reusability/composables#usage-restrictions
I think your need is better to use:

const { pause, resume, isActive } = useIntervalFn(() => {
  /* your function */
}, 1000, { immediate: false, immediateCallback: true })

这是前端轮询请求后端的常用方式，可以有效防止雪崩现象，我必须保证服务器响应之后的几秒再次发生心跳。如果没有响应一直发送会导致大量消息阻塞服务端，带来了不必要的开销，这对后端服务来讲是非常不友好的。然而使用 setTimeout 的递归调用可以有效防止这种情况发生，最起码可以做到自己不去攻击自己。当然，如果技术上有难度可以在文档中说明不可以这样使用，这样使用会导致无法回收，当有类似需求的时候可以自己去实现。

This is a common way for front-end polling to request back-end, which can effectively prevent avalanche phenomenon. I must ensure that the heartbeat occurs again within a few seconds after the server responds. If there is no response, sending continuously will cause a large number of messages to block the server, resulting in unnecessary overhead, which is very unfriendly to backend services. However, using the recursive call 'setTimeout' can effectively prevent this situation from happening, at least by avoiding attacking oneself. Of course, if there are technical difficulties, it can be stated in the documentation that such use is not allowed, as it will result in the inability to recycle. When there is a similar requirement, it can be implemented by oneself.

oh, I got it, maybe this:

const keepAliveResult = ref()
async function keepAlive() {
  keepAliveResult.value = await // your request to server
}
const { start } = useTimeoutFn(keepAlive, 1000)
watch(keepAliveResult, start)