外文资料原文
Software Development Concepts and Design Methodologies
During the 1960s, mainframes and higher level programming languages were applied to many problems including human resource systems, reservation systems, and manufacturing systems. Computers and software were seen as the cure all for many business issues were sometimes applied blindly. Systems sometimes failed to solve the problem for which they were designed for many reasons including:
•Inability to sufficiently understand complex problems
•Not sufficiently taking into account end-user needs, the organizational environment, and performance tradeoffs
•Inability to accurately estimate development time and operational costs
•Lack of framework for consistent and regular customer communications
At this time, the concept of structured programming, top-down design, stepwise refinement，and modularity emerged. Structured programming is still the most dominant approach to software engineering and is still evolving. These failures led to the concept of "software engineering" based upon the idea that an engineering-like discipline could be applied to software design and development.
Software design is a process where the software designer applies techniques and principles to produce a conceptual model that describes and defines a solution to a problem. In the beginning, this design process has not been well structured and the model does not always accurately represent the problem of software development. However, design methodologies have been evolving to accommodate changes in technology coupled with our increased understanding of development processes.
Whereas early design methods addressed specific aspects of the development process, current methods attempt to address the entire scope of software development. Software design methods are often classified in reference to the period in which they were introduced and the problems at that time. Driven by coding and testing problems, tools and methods were developed. Early methods focused on modularity and top-down development, and information hiding through abstraction. This led to the development of structured languages, structured analysis, and data flow analysis.
In the last decade or so, the expense involved in automation has shifted from hardware to people. Therefore, the software engineering community has been focused on object oriented (O-O) design and the concept of re-usable code in order to reduce the human cost component. Inefficient designs and development methodologies have been addressed with Computer Aided Software Engineering (CASE) tools, and fourth generation design languages. This has been done in an attempt replace the traditional waterfall life cycle process model under which most existing software has been developed.
一、 Software Design Fundamentals
Software design methods all aim to provide the software designer with a system blueprint. This blueprint usually has three aspects: data, architectural, and procedural.
•Data design refers to the data's organization, relationships, access and processing methods.
•Architectural design defines the components of the system and their relationships.
•Procedural design builds on the data and architectural design phases to describe the processing details of the system.
Even though there are numerous design methodologies, their basic concepts are very similar-All software design methods partition the problem and software into smaller pieces in order to reduce complexity. They all strive to identify data structures and functions, and provide measurements for software quality. Some of the common principles in software design include: stepwise refinement, software architecture, program structure, data structure, software procedures, modularity, abstraction, and information hiding.
二、 Modern Design Methodologies
Conventional software development practices can generally be mapped onto the traditional life-cycle phases of analysis, functional specification, design, implementation, testing, and maintenance. This thought process is inadequate for today's complex information systems. As the demand for software is growing much faster than the number of developers, adhering to conventional techniques such as the waterfall method requires too much time, too many people, and is difficult to manage. Hence, many new software development technologies have arisen. Newly developed practices and models do not attempt to separate phases of software development, such as specification and implementation, but instead focus on the concept of program transformation through stepwise refinement and iteration.
1、Object-Oriented Technology
Object-Oriented (O-O) software design technology is fundamentally different from the traditional methods described above. With traditional methods, each module is recognized a major step in the overall process and the process goes from one step to the next. On the other hand, O-O design is structured around a model of objects and the functions they perform.
O-O programming can be traced to the simulation language SIMULA, a high level language developed in the late 60's that introduced object classes as a method to encapsulate data. Later, in the 1970s, Smalltalk was introduced as a complete grapgh design and coding as detail is added to the design. This provides a common language throughout each stage in development. O-O is best applied with specifically designed O-O development tools, but it is important to remember that as a methodology is it not specific to any programming language. Many different programming languages can be used to implement 0-0 technology and design methodologies.
Instead of procedures and functions passing data back and forth, in object oriented design, the system is viewed as a collection of objects with messages passed from object to object. Each object has its own set of associated operations. Object-oriented design is based on the idea of information hiding and modularization of both data and processing. It is best used when neither data structure nor processing operations are well defined ahead of time. This is quite useful in today's business environment where requirements are always changing and not very well defined. Thus, it has become quite popular! The concept of objects performing services is a natural way of thinking for both developers and customers. This facilitates understanding the problem domain and a more natural design. In addition, there are many benefits of object-oriented development. These include:
• Inheritance capitalizes on the commonalty of attributes and services allowing code and objects to be re-used.
.Information hiding makes systems more stable by localizing changes to objects and thereby making them reusable.
.The object-oriented development process is consistent from analysis, through design, to coding.
More information on Object Oriented Programming principles can be found in Chapter 4-Organization of Programming Languages and Programming Concepts.
2、 Prototyping
Prototyping was invented because end users participating in the development phase found it difficult to understand requirement specifications and conceptual models. However, when it first began being used in the 1980s, most conventional life cycle developers considered it expensive and time consuming.
Since that time, users and developers have used prototypes successfully as a communications tool to demonstrate system requirements. After several prototype iterations, developers have a better understanding of user requirements and users have a better idea of how the system will eventually work, look, and l.
The number of times the prototype is incrementally refined depends on how well the user requirements and understood. It also depends on the users need to add requirements or change previously stated requirements. After establishing an overall architecture and framework, the system is developed and delivered in increments. Users may experiment with and use delivered increments while others are being developed. For instance, the first prototype may be delivered that implements a certain screen with only some active menu items. While users are experimenting with this screen and menu items, other screens and menu items are concurrently being developed which later will be combined with the existing prototype as it evolves.
Once the user is satisfied that the prototype meets requirements, the prototype is transformed into the "system". This effort depends on several factors. It may include adding functionality that wasn't initially recognized as required, replacing inefficient parts of the prototype to meet performance criteria, or adapting the prototype to fit the user's hardware environment.
Prototyping can begin very early, after some preliminary requirements analysis has determined the basic functionality, scope, and environment of the proposed software. Contrary to the traditional waterfall method, in the prototyping, functional specifications are not fixed. Rather, users are encouraged to modify their requirements as they themselves begin to understand them better. This is because users often don't really know what they want until they see it on the screen. The prototyping process of demonstration, review, and refinement gets the user more involved in the development process, giving them a sense of ownership during the process and at final system delivery. However, due to the mindset of "prototype", users often find it difficult to verify that the prototype satisfies their requirements. Therefore, guidelines must be established to determine when to stop iterating and the prototype to final product.
外文资料译文
软件开发概念和设计方法
在20世纪60年代，大型机和高级程序语言被用来解决包括人力资源系统、专有系统和制造系统等许多问题。计算机和软件被视为解决所有商业问题的万能药，有时候甚至被盲目的应用。因为很多设计上的原因，这些系统并不是万能的。主要因素如下：
1．不能完全理解复杂的问题
2．没有充分满足终端用户的需求，组织环境和性能折中
3．没有准确估计开发时间和运行成本
4．缺乏一致，规范的客户通讯框架
这个时候，结构化的编程，自上而下设计的概念出现了。对软件工程来说，结构化编程至今仍是最重要的方法且不断发展。“软件工程”概念的出现则是基于这样的构想：一个类似工程学的学科可以应用于软件的设计和开发。
软件设计是一种方法，软件设计人员可以籍此应用技术和规则生成一种描述并定义问题解决方法的模型。最初，设计方法一直未能构建好，而且模型也不能准确地描述软件开发的问题。然而，随着我们对开发过程的深入理解，设计方法已经不断适应技术的变化了。
生命周期过程的模型下开发的，但人们开始尝试寻找这种模型的替代品。
一、 软件设计基础
软件设计方法最终的目标就是向软件设计者提供一张系统蓝图。它通常有三个方面：数据，构架和过程。
． 数据设计指的是数据的组织、关系、访问和处理方法。
．构架设计定义系统组件和它们之间的关系。
．过程设计建立在数据和构架设计阶段之上描述系统的处理细节。
尽管设计方法众多，但它们的基本概念非常相似。为了减少复杂度，几乎所有软件设计方法都把问题和软件分割成较小的部分用于标识数据结构、功能以及度量软件品质。软件设计包括以下这些普遍原则：逐步求精、软件构架、程序结构、数据结构、软件过程、模块化、抽象和信息隐藏。
二、 现代设计方法
常规的软件开发实践通常能被映射到传统的生命阶段上，包括分析、功能说明、设计、实现、测试和维护。然而对软件需求的增长比软件开发者数量增长要快，遵守常规的技术你瀑布模型）耗时太长，过多人员的参与也带来了管理上的困难，显然常规的思考过程对于今天的复杂信息系统是不够的。因此，产生了许多新的软件开发技术。最新发展出的实践和模型井不试图把软件开发分割成多个阶段（如说明和实现），而是注重于通过逐步求精和迭代把概念转换成程序。
1、面向对象的技术
面向对象的软件设计技术从根本上有别于传统的设计方法。传统方法中，每个模块被当作全局过程的一个主要步骤，一步一步地往下走；而面向对象的设计围绕着对象模型和对象所执行的功能进行结构化。
面向对象的编程可以追溯到仿真语言SIMULA。SIMULA是一种20世纪60年代后期的高级语言，引入了“对象类”作为封装数据的方法。到了20世纪70年代，Smalltalk被作为一种完全的图形用户界面（GUI）面向对象的编程环境被引入。甚至在30年以后，Smalltalk仍然是度量其他所有面向对象语言的标准。由于面向对象的概念日趋成熟，最近十年这种软件开发方法已经流行起来。同时，软件业注意的焦点己经从编码和结构化过程转移到通过设计和柔韧性来节省劳动力成本和时间。柔韧性变得十分关键，因为系统随着需求的变化而快速改变：变得更大，更复杂和更不稳定。
在面向对象中，分析和设计没有真正分开。在分析期间，系统对象及其特性和关系一起被确定。这些对象可以护，这样就给整个开发过程中的所有阶段提供了一种公用的语言。采用面向对象方法最好是使用专门设计的面向对象的开发工具，但是请一定记住它是一种方法而不是特指任何编程语言。许多不同的编程语言都可以用来实现面向对象技术和设计方法。
和过程、功能往返传递数据的方式不同，在面向对象的设计中，系统被看成一个由很多互相传递消息的对象组成的集合，每个对象都有它自己关联操作的集合。面向对象的设计基本构想是把数据和过程进行信息隐藏和模块化，它最适用于数据结构或者过程操作没有被提前的定义好的情况。这对于今天的商业环境中相当有用，毕竟需求总是不断改变而不能很好的定义。这也是面向对象的设计现在相当流行的重要原因。对象执行服务的概念是一种开发者和客户都很自然的思考方法，这有利于理解问题的范围，也是一种更加自然的设计。此外，面向对象的开发还有许多优点。
．通过属性和服务的结合使用，继承可以重用代码和对象。
．信息隐藏通过局限对象的变化使系统更加稳定，从而使对象可以重用
．面向对象的开发过程从分析、设计到编码都是一致的。
2、原型法
原型法的出现是因为参于开发阶段的终端用户觉得很难理解需求说明和概念模型。而当原型法在20世纪80年代第一次被使用时，大部分常规的生命周期开发者认为它费时费力。但从那时开始，用户和开发者已经能成功地应用原型作为通讯工具来演示系统的需求。原型多次迭代后，开发者对用户的需求有了更好的理解，用户也对系统最后如何操作、看起来像什么和如何感觉都有所了解。原型法已经被证明是一种理解用户需求和问题的有效方法，它有效地消除了在以后开发过程中因错误理解导致高昂代价的重复工作。原型法对于开发用户界面特别有价值，因为对于用户界面，有时候无法使用具体的词和图表进行描述，而用户界面的开发在今天的开发环境中又格外关键，它在整个系统开发费用中份额仍在增加。
（1） 丢弃型原型
原型有两类，丢弃型原型和演进型原型。丢弃型原型被设计成用来校验或者驱动系统需求，也可以用于评估性能或决定设计是否可行。一旦获得所需的信息和完成原型的目的就把原型就丢弃掉，继续常规的软件设计。
丢弃型原型:
A 用于减少需求风险，便于理解用户需求；
B 不作为最终的系统，在使用以后就被抛弃具体原因有:
①原型省掉了系统许多特性；
②为演示用而快速完成，其结构或许不必最佳且难于维护。
在使用丢弃型原型时，分析家通常考虑把原型作为设计或说明语言设计阶段的输入，而不是作为产品或者系统本身。以用于传统系统
（2）．演进型原型
与有意设计成丢弃型原型不同，演进型原型向终端用户交付一个可用的系统在完全了解用户需求后，即可开始演进型原型的开发它被当作过程模型的焦点，条的屏幕原型，当用户正在体验屏幕和菜单条时，其他屏幕和菜单条正同时被开发，然后再与他们相关的现有原型结合在一起。
一旦用户对原型所能满足的需求感到满意，原型就转变成“系统”这取决于若干因素，包括增加最初无法识别需求的功能，为满足性能标准而替换原型中效率低的部分，或者改动原型以适应用户的硬件环境。
可以很早就开始建立原型，经过一些预先需求分析以后决定出其基本的功能、范围和建议的软件环境。与传统的瀑布法相反，原型法中的功能说明并不固定。原型法甚至鼓励用户在更好的理解自己需求后修改他们的需求，因为用户经常直到在屏幕上看到设计结果才知道他们自己真正需要的是什么。原型法的示范、回顾和精炼过程都注重了用户对开发过程的参与性，无论在开发期间还是最终系统提交都给予他们一种所有权的感觉。然而，由于“原型”是预期的，用户也发现难于验证原型是否满足他们的需求。因此，必须建立一个指南，决定什么时候停止选代，并转换原型为终端产品。
