Motivation

Consider the following synchronous JavaScript function to read a file and parse it as JSON. It is simple and easy to read, but you wouldn't want to use it in most applications as it is blocking. This means that while you are reading the file from disk (a slow operation) nothing else can happen.

function readJSONSync(filename) {
return JSON.parse(fs.readFileSync(filename, 'utf8'));
}

To make our application performant and responsive, we need to make all the operations that involve IO be asynchronous. The simplest way to do this would be to use a callback. However, a naive implementation will probably go wrong:

function readJSON(filename, callback){
fs.readFile(filename, 'utf8', function (err, res){
if (err) return callback(err);
callback(null, JSON.parse(res));
});
}
  • The extra callback parameter confuses our idea of what is input and what is the return value.
  • It doesn't work at all with control flow primitives.
  • It doesn't handle errors thrown by JSON.parse

We need to handle errors thrown by JSON.parse but we also need to be careful not to handle errors thrown by the callback function. By the time we've done all of this our code is a mess of error handling:

function readJSON(filename, callback){
fs.readFile(filename, 'utf8', function (err, res){
if (err) return callback(err);
try {
res = JSON.parse(res);
} catch (ex) {
return callback(ex);
}
callback(null, res);
});
}

Despite all this mess of error handling code, we are still left with the problem of the extra callback parameter hanging around. Promises help you naturally handle errors, and write cleaner code by not having callback parameters, and without modifying the underlying architecture (i.e. you can implement them in pure JavaScript and use them to wrap existing asynchronous operations).

What is a promise?

The core idea behind promises is that a promise represents the result of an asynchronous operation. A promise is in one of three different states:

  • pending - The initial state of a promise.
  • fulfilled - The state of a promise representing a successful operation.
  • rejected - The state of a promise representing a failed operation.

Once a promise is fulfilled or rejected, it is immutable (i.e. it can never change again).

Constructing a promise

Once all of the APIs return promises, it should be relatively rare that you need to construct one by hand. In the meantime, we need a way to polyfill existing APIs. For example:

function readFile(filename, enc){
return new Promise(function (fulfill, reject){
fs.readFile(filename, enc, function (err, res){
if (err) reject(err);
else fulfill(res);
});
});
}

We use new Promise to construct the promise. We give the constructor a factory function which does the actual work. This function is called immediately with two arguments. The first argument fulfills the promise and the second argument rejects the promise. Once the operation has completed, we call the appropriate function.

Awaiting a promise

In order to use a promise, we must somehow be able to wait for it to be fulfilled or rejected. The way to do this is using promise.done (see warning at the end of this section if attempting to run these samples).

With this in mind, it's easy to re-write our earlier readJSON function to use promises:

function readJSON(filename){
return new Promise(function (fulfill, reject){
readFile(filename, 'utf8').done(function (res){
try {
fulfill(JSON.parse(res));
} catch (ex) {
reject(ex);
}
}, reject);
});
}

This still has lots of error handling code (we'll see how we can improve on that in the next section) but it's a lot less error prone to write, and we no longer have a strange extra parameter.

Non Standard

Note that promise.done (used in the examples in this section) has not been standardised. It is supported by most major promise libraries though, and is useful both as a teaching aid and in production code. I recommend using it along with the following polyfill (minified / unminified): undefined

Transformation / Chaining

Following our example through, what we really want to do is transform the promise via another operation. In our case, this second operation is synchronous, but it might just as easily have been an asynchronous operation. Fortunately, promises have a (fully standardised, except jQuery) method for transforming promises and chaining operations.

Put simply, .then is to .done as .map is to .forEach. To put that another way, use .then whenever you're going to do something with the result (even if that's just waiting for it to finish) and use .done whenever you aren't planning on doing anything with the result.

Now we can re-write our original example as simply:

function readJSON(filename){
return readFile(filename, 'utf8').then(function (res){
return JSON.parse(res)
})
}

Since JSON.parse is just a function, we could re-write this as:

function readJSON(filename){
return readFile(filename, 'utf8').then(JSON.parse);
}

This is very close to the simple synchronous example we started out with.

Implementations / Polyfills

p Promises are useful both in node.js and the browser

jQuery

This feels like a good time to warn you that what jQuery calls a promise is in fact totally different to what everyone else calls a promise. jQuery's promises have a poorly thought out API that will likely just confuse you. Fortunately, instead of using jQuery's strange version of a promise, you can just convert it to a really simple standardised promise:

var jQueryPromise = $.ajax('/data.json');
var realPromise = Promise.resolve(jQueryPromise);
//now just use `realPromise` however you like.

Browser

Promises are currently only supported by a pretty small selection of browsers (see kangax compatibility tables). The good news is that they're extremely easy to polyfill (minified / unminified):

<script src="https://www.promisejs.org/polyfills/promise-7.0.4.min.js"></script>

None of the browsers currently support Promise.prototype.done so if you want to use that feature, and you are not including the polyfill above, you must at least include this polyfill (minified / unminified):

<script src="https://www.promisejs.org/polyfills/promise-done-7.0.4.min.js"></script>

Node.js

It's generally not seen as good practice to polyfill things in node.js. Instead, you're better off just requiring the library wherever you need it.

To install promise run:

npm install promise --save

Then you can load it into a local variable using require

var Promise = require('promise');

The "promise" library also provides a couple of really useful extensions for interacting with node.js

var readFile = Promise.denodeify(require('fs').readFile);
// now `readFile` will return a promise rather than
// expecting a callback function readJSON(filename, callback){
// If a callback is provided, call it with error as the
// first argument and result as the second argument,
// then return `undefined`. If no callback is provided,
// just return the promise.
return readFile(filename, 'utf8')
.then(JSON.parse)
.nodeify(callback);
}

Further Reading

Promise deeply lookup的更多相关文章

  1. Promise & Deferred Objects in JavaScript Pt.2: in Practice

    原文:http://blog.mediumequalsmessage.com/promise-deferred-objects-in-javascript-pt2-practical-use Intr ...

  2. Javascript - Promise学习笔记

    最近工作轻松了点,想起了以前总是看到的一个单词promise,于是耐心下来学习了一下.   一:Promise是什么?为什么会有这个东西? 首先说明,Promise是为了解决javascript异步编 ...

  3. 路由的Resolve机制(需要了解promise)

    angular的resovle机制,实际上是应用了promise,在进入特定的路由之前给我们一个做预处理的机会 1.在进入这个路由之前先懒加载对应的 .js $stateProvider .state ...

  4. angular2系列教程(七)Injectable、Promise、Interface、使用服务

    今天我们要讲的ng2的service这个概念,和ng1一样,service通常用于发送http请求,但其实你可以在里面封装任何你想封装的方法,有时候控制器之间的通讯也是依靠service来完成的,让我 ...

  5. 闲话Promise机制

    Promise的诞生与Javascript中异步编程息息相关,js中异步编程主要指的是setTimout/setInterval.DOM事件机制.ajax,通过传入回调函数实现控制反转.异步编程为js ...

  6. 深入理解jQuery、Angular、node中的Promise

    最初遇到Promise是在jQuery中,在jQuery1.5版本中引入了Deferred Object,这个异步队列模块用于实现异步任务和回调函数的解耦.为ajax模块.队列模块.ready事件提供 ...

  7. Promise的前世今生和妙用技巧

    浏览器事件模型和回调机制 JavaScript作为单线程运行于浏览器之中,这是每本JavaScript教科书中都会被提到的.同时出于对UI线程操作的安全性考虑,JavaScript和UI线程也处于同一 ...

  8. JavaScript进阶之路——认识和使用Promise,重构你的Js代码

    一转眼,这2015年上半年就过去了,差不多一个月没有写博客了,"罪过罪过"啊~~.进入了七月份,也就意味着我们上半年苦逼的单身生活结束了,从此刻起,我们要打起十二分的精神,开始下半 ...

  9. SQL Server-聚焦移除Bookmark Lookup、RID Lookup、Key Lookup提高SQL查询性能(六)

    前言 前面几节都是讲的基础内容,本节我们讲讲索引性能优化,当对大数据进行处理时首先想到的就是索引,一旦遇到这样的问题则手忙脚乱,各种查资料,为何平常不扎实基本功呢,我们由浅入深,简短的内容,深入的理解 ...

随机推荐

  1. ORA-04091: table xxxx is mutating, trigger/function may not see it

    今天同事让我看一个触发器为什么老是报错,当执行DML语句触发触发器后,会报ORA-04091错误:ORA-04091: table xxxx is mutating, trigger/function ...

  2. ORA-00911: invalid character --- 字符集的问题

    网上搜了一遍, 大多数是因为分号( ; ) 的问题.  而我的sql文件是没有分号的, 最后发现是sql文件编码和服务器字符集的差异造成 sql文件怎么都看不出问题,直到在UltraEdit里切换到1 ...

  3. springmvc学习资料整理

    springmvc接口开发返回XML/JSON数据 1.SpringMVC中使用@RequestBody,@ResponseBody注解实现Java对象和XML/JSON数据自动转换(上):http: ...

  4. Migrate Instance 操作详解 - 每天5分钟玩转 OpenStack(40)

    Migrate 操作的作用是将 instance 从当前的计算节点迁移到其他节点上. Migrate 不要求源和目标节点必须共享存储,当然共享存储也是可以的. Migrate 前必须满足一个条件:计算 ...

  5. Mac安装Windows 10的简明教程

    每次在Mac上安装Windows都是一件非常痛苦的事情,曾经为了装Win8把整台Mac的硬盘数据都弄丢了,最后通过龟速系统恢复模式恢复了MacOSX(50M电信光纤下载了3天才把系统下载完),相信和我 ...

  6. 【Windows编程】系列第三篇:文本字符输出

    上一篇我们展示了如何使用Windows SDK创建基本控件,本篇来讨论如何输出文本字符. 在使用Win32编程时,我们常常要输出文本到窗口上,Windows所有的文本字符或者图形输出都是通过图形设备接 ...

  7. cf之路,1,Codeforces Round #345 (Div. 2)

     cf之路,1,Codeforces Round #345 (Div. 2) ps:昨天第一次参加cf比赛,比赛之前为了熟悉下cf比赛题目的难度.所以做了round#345连试试水的深浅.....   ...

  8. SSH----小小项目的小小总结

    嘛,之前学了一下SSH框架,跟人合作写了个小东西参加比赛,(当然我是队长),真的感慨良多~,现在用这篇博客记录下来吧 1.责任心/责任感 首先要说的一点,要有责任心,当你与别人组成一个团队的时候,虽然 ...

  9. SQL高级查询:嵌套和分页

    1.嵌套子查询 --查询最近一次oop考试没有参加考试的学生 select StudentName from Student where StudentNo not in( select Studen ...

  10. [No00005B] word快速插入当前时间&怎样一次性删除文档中的全部链接

    按Alt+Shift+D键来插入系统日期 按Alt+Shift+T键则插入系统当前时间 同时,在插入的时间上右键->编辑域 一次性删除文档中的全部链接: 方法1:一劳永逸法(推荐) 因为链接大多 ...