当前位置:网站首页>Teacher Liang's small class
Teacher Liang's small class
2020-11-08 23:53:00 【Official account Song Hua said】

Among many design patterns , I think the template method pattern is easy to understand , It's also a skill worth studying in depth . The definition is as follows , In a multi-step business framework , Most of the steps are fixed , And it can be used in a variety of business scenarios , Variable steps are left to subclasses to implement independently , So we separate stability from change .
Using this model is a little bit careless , There will be some strange problems . The first is that multiple abstract methods modify the same variable , There is a strong correlation between methods , The second problem is that there are too many abstract methods defined . The latter is exactly what I want to talk to you about today , Follow my step and look down .
One 、 Why does this happen ?
I think so , The product manager told us from a business point of view , What steps does this function involve , What needs to be done at each step , Then we copy the concept and turn it into code , Considering that some steps involve multiple implementations , So there are a lot of extension methods left .
soon , New businesses need to reuse these processes , Steps required 1\2\3, But it doesn't take steps 4\5, I can't even understand the steps 4\5 What does it do , I don't know what conditions will be called , The original design is not simple enough .
How to understand the simple here ? Simplicity is from a human point of view . If , The definer and implementer come from two different teams , After the implementer has read the document, the definer needs to explain , To know how to implement abstract methods is right , So it's complex . This is my real feeling , Before that, I was doing capacity building for China Taiwan general purpose , When external empowerment is performed through extension points , The most important thing to think about is how to make the third party understand and start quickly .
To sum up , Copy requirements documents mechanically , Let the execution order be fully reflected in the code structure , It will lead to too many abstract methods in the template method pattern .
Two 、 What to do if something happens ?
What to do if there are too many abstract methods ? In fact, we can use the interface adapter mode to patch .
The actual practice is like this , Create an abstract class Wrapper, Implement all the methods , Method implementation does not require specific business implications . When we write a concrete implementation class , Inherit Wrapper class , Rewrite the methods it cares about , So you don't have to implement steps that you don't understand .
Using interface adapters for compatibility may not be enough , Consider the situation , Abstract methods in templates explicitly need to have return values , How to deal with it ?
It seems that we have to continue patching . such as , Define some default values , Discard these values in the business process . perhaps , Method default implementation runtime throw exception , Indicates that the operation is not supported , It has to be implemented by a third party . Neither of them is good . The first one has special logic , It's expected to be despised by many people . The second is a little more elegant , The disadvantage is that the debugging process can only find the exception .
To sum up , Using the interface adapter pattern , Let the subclass only care about the methods it needs , So the old code 、 Dirty code enlivens .
3、 ... and 、 How to avoid too many abstract methods ?
There are too many abstract methods in the template method pattern , The side shows that the process steps are complicated , Not simple enough . therefore , Another description of the problem is , How to avoid exposing too much detail when designing template method .
My advice is , Organize the business logic , Extract the same kind of behavior , Or merge methods that share a lot of information , Use a general term to summarize the link , Hide the details , Then load it in combination , Don't tile the code directly .
For example . In a request link , The intermediate method directly controls the cache clearing through the cache client 、 Cache value settings , Other links read cache values . This coding style , It will cache Key、 Cache components are exposed to third parties , This leads to difficulties in subsequent modifications . Normally, a new business class should be created , Specifically with the cache Key Dealing with .
To sum up , The key to avoiding too many abstract methods is how to avoid exposing too many details .
This article is from WeChat official account. - Songhua said (songhuasay).
If there is any infringement , Please contact the [email protected] Delete .
Participation of this paper “OSC Source creation plan ”, You are welcome to join us , share .
版权声明
本文为[Official account Song Hua said]所创,转载请带上原文链接,感谢
边栏推荐
- How to reduce the resource consumption of istio agent through sidecar custom resource
- The interface testing tool eolinker makes post request
- Concurrent linked queue: a non blocking unbounded thread safe queue
- 如何让脚本同时兼容Python2和Python3?
- LeetCode-15:三数之和
- 文件拷贝的实现
- JVM真香系列:轻松理解class文件到虚拟机(下)
- Octave基本语法
- 理论与实践相结合彻底理解CORS
- Five phases of API life cycle
猜你喜欢

CMS garbage collector

Introduction to nmon

Realization of file copy

Copy the picture

Python features and building environment

分库分表的几种常见玩法及如何解决跨库查询等问题

Iterm2 configuration and beautification

How does semaphore, a thread synchronization tool that uses an up counter, look like?

JVM真香系列:轻松理解class文件到虚拟机(下)

基于链表的有界阻塞队列 —— LinkedBlockingQueue
随机推荐
What are the basic requirements for big data posts?
API生命周期的5个阶段
Concurrent linked queue: a non blocking unbounded thread safe queue
How to reduce the resource consumption of istio agent through sidecar custom resource
非阻塞的无界线程安全队列 —— ConcurrentLinkedQueue
Linked blocking queue based on linked list
小议缓冲区溢出
Programmers should know the URI, a comprehensive understanding of the article
How to deploy pytorch lightning model to production
Web上的分享(Share)API
Aprelu: cross border application, adaptive relu | IEEE tie 2020 for machine fault detection
JVM真香系列:轻松理解class文件到虚拟机(下)
CMS garbage collector
服务器性能监控神器nmon使用介绍
使用递增计数器的线程同步工具 —— 信号量,它的原理是什么样子的?
How does semaphore, a thread synchronization tool that uses an up counter, look like?
Linked list
通过canvas获取视频第一帧封面图
C/C++编程笔记:指针篇!从内存理解指针,让你完全搞懂指针
android开发中提示:requires permission android.permission write_settings解决方法