📚 IGCSE Computer Science: Software Engineering Key Points | IGCSE 计算机：软件工程 考点精讲
Software engineering is the disciplined approach to designing, developing, testing, and maintaining software systems. It ensures that large-scale projects are delivered on time, within budget, and meet user requirements. In IGCSE Computer Science, understanding the stages of the software development life cycle, testing strategies, and development methodologies is essential. This article distils the key concepts you need to master for your examination.
软件工程是一种系统化的方法，用于设计、开发、测试和维护软件系统。它确保大型项目按时、在预算范围内交付并满足用户需求。在 IGCSE 计算机科学中，理解软件开发生命周期的各个阶段、测试策略和开发方法论至关重要。本文提炼了你需要掌握的考试关键概念。
1. The Software Development Life Cycle (SDLC) | 软件开发生命周期
The SDLC is a structured process for building software. The typical stages are: analysis, design, coding/development, testing, implementation, and maintenance. Each phase produces deliverables that feed into the next stage, ensuring a systematic progression from idea to final product. Some models are linear, while others allow iteration and overlap.
软件开发生命周期（SDLC）是构建软件的结构化流程。典型阶段包括：分析、设计、编码/开发、测试、实施和维护。每个阶段产出的可交付物会流入下一阶段，确保从想法到最终产品的系统化推进。有些模型是线性的，而另一些则允许迭代和重叠。
Different methodologies, such as the waterfall model and iterative/agile approaches, share these core phases but arrange them differently. In examinations, you should be able to identify the activities that belong to each stage and explain the importance of following a structured life cycle rather than coding immediately.
不同的方法论，如瀑布模型和迭代/敏捷方法，共享这些核心阶段，但组织方式不同。在考试中，你应该能够识别每个阶段的活动，并解释遵循结构化生命周期而非立即编码的重要性。
2. Analysis: Gathering Requirements | 分析阶段：需求收集
During the analysis phase, the problem is clearly defined. Analysts interview users, observe current processes, and examine existing documentation to understand exactly what the software must do. The output is a detailed requirements specification that serves as a contract between the client and developers. Functional requirements describe what the system should do; non-functional requirements cover performance, security, and usability constraints.
在分析阶段，需要明确界定问题。分析师通过访谈用户、观察现有流程和检查文档，准确理解软件必须实现的功能。产出是一份详细的需求规格说明书，作为客户与开发人员之间的合约。功能性需求描述系统应做什么；非功能性需求涵盖性能、安全性和可用性约束。
A feasibility study may also be conducted to assess whether the project is technically and economically viable. Poor analysis leads to scope creep, missed deadlines, and a product that fails to meet user needs. Memorise that analysis answers the question 'What is needed?' before any design begins.
可能还会进行可行性研究，评估项目在技术和经济上是否可行。糟糕的分析会导致范围蔓延、错过截止日期，最终产品无法满足用户需求。记住，在任何设计开始之前，分析回答的是'需要什么'这个问题。
3. Design: Architecture and Interface | 设计阶段：结构与界面
Design translates the requirements into a blueprint for construction. System design covers the overall architecture, including modular decomposition, data flow, and choice of hardware/software platform. Detailed design specifies algorithms, data structures, database schemas, and user interface layouts. Common tools include structure diagrams, flowcharts, and pseudocode.
设计将需求转化为构建的蓝图。系统设计涵盖整体架构，包括模块分解、数据流以及硬件/软件平台的选择。详细设计则指定算法、数据结构、数据库模式和用户界面布局。常用工具包括结构图、流程图和伪代码。
User interface design is critical because it directly affects usability. Designers create screen mock-ups and prototypes to validate layouts with users. The design phase also defines the testing strategy, specifying how each requirement will be verified. A well-thought-out design reduces costly changes during coding.
用户界面设计至关重要，因为它直接影响可用性。设计师创建屏幕模型和原型，与用户一起验证布局。设计阶段还定义测试策略，明确每项需求将如何验证。周全的设计能减少编码阶段代价高昂的变更。
4. Development: Coding and Implementation | 开发阶段：编码与实现
Development is where the design is turned into a working program. Programmers write code in a chosen high-level language, following the design specifications. Coding standards, version control, and regular code reviews are used to maintain quality and consistency. Unit testing often begins in parallel: developers test individual modules as they are completed.
开发阶段是将设计转化为可运行程序的过程。程序员按照设计规格，用选定的高级语言编写代码。编码规范、版本控制和定期代码审查用于维持质量和一致性。单元测试通常同步开始：开发人员完成了各模块后立即进行测试。
In modern projects, development is frequently iterative. A minimal viable product (MVP) may be built first, then enhanced through successive sprints. This allows early feedback and reduces the risk of building the wrong product. Regardless of the methodology, the goal of this stage is to produce a fully functional program ready for systematic testing.
在现代项目中，开发通常是迭代式的。可能先构建一个最小可行产品（MVP），然后通过后续冲刺逐步增强功能。这允许早期反馈，降低构建错误产品的风险。不管采用何种方法论，此阶段的目标是产生一个功能完整的程序，准备进行系统化测试。
5. Testing: Ensuring Quality | 测试阶段：保证质量
Testing is a systematic process to find errors and verify that the software meets its specifications. The main levels are unit testing (individual components), integration testing (combined modules), system testing (the whole application), and acceptance testing (user sign-off). Each level serves a different purpose, and they are typically executed in that order.
测试是发现错误并验证软件是否符合规格的系统化过程。主要层级包括：单元测试（单个组件）、集成测试（组合模块）、系统测试（整个应用程序）和验收测试（用户签字认可）。每个层级服务于不同目的，通常按此顺序执行。
Alpha testing is carried out by an internal team in a controlled environment, while beta testing involves real users in their own environment before the official release. Automated testing tools can rerun test suites quickly, which is essential for iterative development. Remember: testing can show the presence of bugs, but not their absence.
Alpha 测试由内部团队在受控环境中进行，而 Beta 测试则在正式发布前让真实用户在自己的环境中使用。自动化测试工具可以快速重新运行测试套件，这对迭代开发至关重要。请记住：测试只能证明错误的存在，而不能证明错误的不存在。
6. Test Data: Normal, Abnormal, and Boundary | 测试数据：正常、异常、边界
Choosing appropriate test data is vital for thorough testing. Three types are examined:
| Type 类型 | Description 描述 | Example 示例 |
|---|---|---|
| Normal (valid) 正常（有效） | Data that the system would typically process without errors. | Age = 25 (expected range 0–120) |
| Abnormal (invalid/erroneous) 异常（无效/错误） | Data that the system should reject gracefully, producing an error message. | Age = 'twenty' or Age = -5 |
| Boundary (extreme) 边界（极值） | Data at the very edge of the acceptable range, plus one step beyond. | Age = 0, 120 (boundary), also 121 (just outside) |
is especially important because logic errors frequently occur at range limits (e.g., using < instead of <=). A full test plan includes all three types for each input field or module.
正常数据确认程序在预期条件下工作。异常数据测试验证程序的健壮性。边界数据尤其重要，因为逻辑错误常发生在范围极限（如使用了 < 而非 <=）。一份完整的测试计划应针对每个输入字段或模块包含所有三种数据。
7. Implementation and Deployment | 实施与部署
Implementation is the process of installing the software and making it operational. This stage includes data conversion (migrating existing data into the new system), staff training, and producing user documentation. The changeover can be direct (big bang), parallel running, phased, or pilot – each with its own trade-off between risk and cost.
实施是安装软件并使其投入运行的过程。此阶段包括数据转换（将现有数据迁移到新系统）、员工培训和编写用户文档。切换方式可以是直接切换（大爆炸式）、并行运行、分阶段或试点——每种方式在风险与成本之间都有各自的权衡。
Direct changeover is quick but risky: if the new system fails, there is no fallback. Parallel running keeps the old system operational alongside the new for a period, reducing risk but increasing cost due to duplication of effort. Phased and pilot approaches introduce the system gradually, allowing issues to be resolved before full rollout.
直接切换速度快但风险高：如果新系统失败，将无回退方案。并行运行让旧系统与新系统同时运行一段时间，降低风险但因重复劳动而增加成本。分阶段与试点方法逐步引入系统，使得在全量推广之前可以解决问题。
8. Maintenance Phase | 维护阶段
Once the system is live, maintenance ensures it continues to meet user needs. Three categories of maintenance are defined:
- Corrective maintenance: fixing bugs that slipped through testing.
- Adaptive maintenance: modifying the system to work in a changed environment (e.g., new operating system or hardware).
- Perfective maintenance: adding new features or improving performance based on user requests.
系统上线后，维护确保它继续满足用户需求。维护分三类：
- 纠正性维护：修复测试阶段遗漏的错误。
- 适应性维护：修改系统以适应变化的环境（如新操作系统或硬件）。
- 完善性维护：根据用户请求添加新功能或改进性能。
Maintenance is the longest and most expensive phase of the SDLC. Good design, thorough documentation, and modular coding reduce maintenance effort. In the exam, you may be asked to categorise a given change or justify why maintenance accounts for the majority of total cost.
维护是 SDLC 中时间最长、成本最高的阶段。良好的设计、详尽的文档和模块化编码能减少维护工作量。在考试中，你可能会被要求对某个变更进行分类，或解释为什么维护占总成本的大部分。
9. Development Methodologies: Waterfall vs. Iterative | 开发方法论：瀑布与迭代
The waterfall model is a linear sequential approach: each phase must be completed before the next begins. It is easy to manage and works well when requirements are stable and well understood. However, it is inflexible; if a mistake is discovered late, returning to an earlier phase is expensive.
瀑布模型是一种线性顺序方法：每个阶段必须在下一阶段开始之前完成。它易于管理，在需求稳定且理解透彻时效果良好。然而，它缺乏灵活性；如果在后期发现错误，回到早期阶段代价高昂。
Iterative (and agile) models break the project into small cycles (iterations or sprints), with each delivering a working increment of the software. This allows user feedback to refine requirements along the way. Iterative development is more adaptable to change and reduces the risk of building a product that no longer meets user needs, but it requires close collaboration and may be harder to plan.
迭代（及敏捷）模型将项目分解为小的周期（迭代或冲刺），每个周期交付一个可运行的软件增量。这使得用户反馈可以在过程中不断完善需求。迭代开发更能适应变化，降低构建出不再满足用户需求的产品的风险，但需要紧密协作，并且可能更难规划。
10. Prototyping and User Involvement | 原型与用户参与
A prototype is an early, simplified version of the software used to demonstrate ideas and gather feedback. Prototypes can be throwaway (built quickly, discarded after clarification) or evolutionary (refined into the final product). They help bridge the gap between user expectations and developer understanding.
原型是软件的早期简化版本，用于展示想法和收集反馈。原型可以是抛弃式的（快速构建，澄清后丢弃）或演化式的（逐步完善为最终产品）。它们有助于弥合用户期望与开发者理解之间的鸿沟。
User involvement throughout the SDLC improves the likelihood of success. Users participate in analysis, review designs, test prototypes, and conduct acceptance testing. Engaging users reduces misunderstandings and increases satisfaction, but it requires time and effective communication channels.
用户全程参与 SDLC 能提高成功率。用户参与分析、评审设计、测试原型并进行验收测试。让用户参与能减少误解，提高满意度，但这需要时间和有效的沟通渠道。
11. Documentation and CASE Tools | 文档与CASE工具
Good documentation is vital at every stage. Technical documentation describes the internal structure of the system (algorithms, data structures, code comments) and is intended for developers. User documentation (manuals, help files, tutorials) explains how to operate the software for end users. Both must be clear, accurate, and updated alongside the software.
每个阶段良好的文档都至关重要。技术文档描述系统的内部结构（算法、数据结构、代码注释），面向开发人员。用户文档（手册、帮助文件、教程）解释如何操作软件，面向最终用户。两者都必须清晰、准确，并与软件同步更新。
Computer-Aided Software Engineering (CASE) tools automate many SDLC activities. They include diagramming tools for design, code generators, version control systems, and automated testing suites. CASE tools improve productivity, enforce standards, and reduce human error, though they require training and investment.
计算机辅助软件工程（CASE）工具能自动化 SDLC 的许多活动。它们包括用于设计的绘图工具、代码生成器、版本控制系统和自动化测试套件。CASE 工具提高生产力，强制执行标准，减少人为错误，尽管它们需要培训与投入。
Published by TutorHao | Computer Science Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图