在向设备发送数据时，UNIX的原子性read（）/ write（）(Atomicity of UNIX read()/write() when sending data to device)

 你想创建一个类有几个原因： 
 
 
 
 强大的打字功能利用编译器确保正确性。 
 
 
 每个封装数据的类都像一个合同 。 你可以想象通过检查课堂是如何使用的。 
 
 
 在你之后强迫这个人阅读它是如何工作的。 读取类属性并将其实用程序映射为更简单。 
 
 
 这是一个糟糕的规划和工程标志。 您正在创建数据块而不是解决特定问题的结构化数据集。 认为泥坑与乐高积木。 
 
 
 这份名单始终无限。 我认为这里的共识是避免它。 在极其罕见的情况下，这很有用。 对于大多数人来说，坚持契约和编码抽象而不是实现细节。 

There are several reasons why you want to create a class:
 
 
 
Strong typing leverages the compiler to ensure correctness.
 
 
Every class that encapsulates data is like a contract. You can imagine how's used by examining the class.
 
 
You force the guy after you to read how it works. It is simpler to read class properties and image its utility.
 
 
It is a sign a bad planning and engineering. You are creating blobs of data instead of structured data sets that solve a specific problem. Think mud pits versus Lego blocks.
 
 
The list goes on ad infinitum. I think the consensus here is to avoid it. There are extreme rare cases where this is useful. For most, stick to contracts and coding to abstractions not implementation details.