java - web端百度网盘的一个操作为什么要分两次请求服务器, 有什么好处吗

Question

以 文件重命名 为例:

• 当完成重命名操作提交会到这个地址
  https://pan.baidu.com/api/filemanager

返回如下结果

{
  "errno": 0,
  "info": [],
  "request_id": 88137407060055336,
  "taskid": 307843054247316
}

可以联想到在server端建立了一个task, 并返回了taskid让客户端后续取状态来更新ui

• 客户端轮训服务器的接口 https://pan.baidu.com/share/taskquery 来获取状态, 1秒一次请求, 服务器端返回结果如下: 分几种情况我总结了一下

#进行中的返回值
{
  "errno": 0,
  "request_id": 88137707954758994,
  "task_errno": 0,
  "status": "pending"
}

#进行中
{
  "errno": 0,
  "request_id": 88137707954758994,
  "task_errno": 0,
  "status": "running"
}


#操作成功的返回值
{
  "errno": 0,
  "request_id": 88138584419582326,
  "task_errno": 0,
  "status": "success",
  "list": [
    {
      "from": "/test1/我的照片",
      "to": "/test1/我的照片2"
    }
  ],
  "total": 1
}

当 status 为success时候, 则轮询结束, 更新UI元素

问题: 直接访问重命名接口不行吗? 为什么要这么设计, 好处是什么?

Answer 1

You have made it very clear. Is there anything else you don’t understand?

The first time is to initiate a name change application to the server.
The server will start the task.
The subsequent polling is to check whether the task is completed, and the front-end will perform corresponding operations after completion. In case of failure, the front-end will also perform rollback operations.

Answer 2

Guess the reason: In extreme cases, the operation may take a long time and cannot be returned immediately. When the operation is completed, the socket link may have been disconnected and the final result cannot be obtained. Designed as a task queue, customers can be guaranteed Get the final result.

Answer 3

In addition to the above considerations, there may be another very important reason, and that is concurrency pressure. Made asynchronous, it can solve concurrency very well