Effective software testing: 50 specific ways to improve your testing

2024-04-19 04:38

本文主要是介绍Effective software testing: 50 specific ways to improve your testing,希望对大家解决编程问题提供一定的参考价值,需要的开发者们随着小编来一起学习吧!

版权声明:原创作品,允许转载,转载时请务必以超链接形式标明文章原始出版、作者信息和本声明。否则将追究法律责任。 http://blog.csdn.net/topmvp - topmvp

Many of the examples are in the realm of the obvious, but not always clearly implemented. For example, number 17 is "Verify that the system supports testability" and number 31is "Know the different types of testing-support tools." In the case of 17, the word system is being used to describe large projects and the topic deals with verifying that it is possible to test the integrated system by tracking the source of errors and how the components interact with each other. Tracking down errors that arise when systems are integrated is very hard, and in general it is necessary to make the plans on how to track down such errors when the architecture for the system is being designed. Therefore, the point, which seems obvious, is in fact a very critical one.

Determining what software testing tools are available is another very critical step in the effective, efficient testing of software. The enormous number of different pathways through modern software makes it impossible to use even the largest army of testers to check them all. Therefore, the only way to provide reasonable coverage of the options is to let other software examine it. Tools such as memory and other system resource checkers, source code scanners that look for "obvious" bugs, programs that generate test data sets either randomly or clustered about most likely values, and test generators can sometimes provide valuable assistance in cleaning code. When working in C/C++, I found a code scanner to be extremely helpful, and wrote a simple one some time ago. It searched for instances of a single "=" in Boolean expressions, semicolons immediately after if, while and for loops, and places where the delete command was used on pointers that might point to an array. It was simple, but it did find some bugs that quite likely would have taken me hours to find.

The testing principles are organized into ten categories:
* Requirements phase.
* Test planning.
* The testing team.
* The system architecture.
* Test design and documentation.
* Unit testing.
* Automated testing tools.
* Automated testing: Selected best practices.
* Nonfunctional testing.
* Managing test execution.

As you can see, the concept of testing is included from the very first stages of the project. This is essential, as it is very possible to incorporate something into the design that will make effective testing difficult or even impossible. Experienced software testers should be included in the planning from the first day of construction.


http://rapidshare.com/files/52157148/0201794292.zip

这篇关于Effective software testing: 50 specific ways to improve your testing的文章就介绍到这儿,希望我们推荐的文章对编程师们有所帮助!



http://www.chinasem.cn/article/916643

相关文章

【附答案】C/C++ 最常见50道面试题

文章目录 面试题 1:深入探讨变量的声明与定义的区别面试题 2:编写比较“零值”的`if`语句面试题 3:深入理解`sizeof`与`strlen`的差异面试题 4:解析C与C++中`static`关键字的不同用途面试题 5:比较C语言的`malloc`与C++的`new`面试题 6:实现一个“标准”的`MIN`宏面试题 7:指针是否可以是`volatile`面试题 8:探讨`a`和`&a`

day-50 求出最长好子序列 I

思路 二维dp,dp[i][h]表示nums[i] 结尾,且有不超过 h 个下标满足条件的最长好子序列的长度(0<=h<=k),二维数组dp初始值全为1 解题过程 状态转换方程: 1.nums[i]==nums[j],dp[i,h]=Math.max(dp[i,h],dp[j,h]+1) 2.nums[i]!=nums[j],dp[i,h]=Math.max(dp[i,h],dp[j,h-1

50. Pow(x,n)

题目: 解答: 主要是求 n &gt; 0 n &gt; 0 n>0 的情况的计算,其他时候,可以通过转换得到。 而 n &gt; 0 n &gt; 0 n>0 的情况下, ​ n = a 0 2 0 + a 1 2 1 + a 2 2 2 … a m 2 m n = a_0 2^0 + a_1 2^1 + a_2 2^2 \ldots a_m 2^m n=a0​20+a1​21

50个实用的jquery案例

1. 如何创建嵌套的过滤器: //允许你减少集合中的匹配元素的过滤器,   //只剩下那些与给定的选择器匹配的部分。在这种情况下,   //查询删除了任何没(:not)有(:has)   //包含class为“selected”(.selected)的子节点。   .filter(":not(:has(.selected))")  2. 如何重用元素搜索 var allI

关于javaSocket中 Software caused connection abort: recv failed问题

在学习Socket中今天突然遇到了下面这样的问题 原来是网路连接出了问题,因为我测试的是远程连接所以是在学校的局域网下,结果很不稳定,开始还以为怎么了一会连上了一会又出现问题然后把IP地址改为本机的127.0.0.1之后就没有 出现过了.

【深度学习 卷积】利用ResNet-50模型实现高效GPU图片预测

本文介绍了如何使用训练好的ResNet-50模型进行图片预测。通过详细阐述模型原理、训练过程及预测步骤,帮助读者掌握基于深度学习的图片识别技术。 一、引言 近年来,深度学习技术在计算机视觉领域取得了显著成果,特别是卷积神经网络(CNN)在图像识别、分类等方面表现出色。ResNet-50作为一种经典的CNN模型,以其强大的特征提取能力和较高的预测准确率,在众多领域得到了广泛应用。本文将介绍如何使

myeclipse中没有software update选项

方法一:你在myeclipse中点Windows-->Preferences 在弹出的对话框的左侧栏,点General-->点Capabilities,然后把Classic Update前的框选上,OK了。 方法二:也可以通过Help-->MyEclipse Configuration Center-->Software-->Add Site的方式进行插件安装。。

王立铭脑科学50讲后续10,自己从课程中提起自己所需的知识,安放到自己的知识体系中。

王立铭脑科学50讲后续10,自己从课程中提起自己所需的知识,安放到自己的知识体系中。 1、大脑功能是由先天、后天两方便共同决定,先天有一堆带电量的开关,后天通过学习不断电量这些开关。 这部分知识放在整个体系中,在需求和动机之内,每个人还有自己不同程度的动机偏好,围绕着动机,产生的一些列奖惩机制、预设的情绪处理算法、语言能力、预测未来的能力等,每个人都有不同程度的差别,同时提醒自己,像价值观、人

NVIDIA RTX 50系列大爆料:功耗飙升600W,性能直逼RTX 4090 1.?倍,你准备好了吗?

在科技圈的万众瞩目下,知名硬件爆料大神Kopite7kimi再次为我们揭开了NVIDIA下一代GeForce RTX系列——“Blackwell”阵容的神秘面纱。这次,关于新显卡的功耗信息不再是模糊的概念,而是实实在在的数字,让人不禁对即将到来的性能飞跃充满期待。 功耗飙升,性能也要飞? 据Kopite7kimi确认,RTX 50系列中的旗舰型号RTX 5090将拥有惊人的功耗表现,其最高功

王立铭脑科学50讲后续6,自己从课程中提起自己所需的知识,安放到自己的知识体系中。

王立铭脑科学50讲后续6,自己从课程中提起自己所需的知识,安放到自己的知识体系中。 1、对比计算机,脑部有不同,脑有目的,脑有情感 这个知识点怎么放? 暂时将脑有目的,与繁衍结合起来,跟奖惩机制放一起,放在预期差后面的链条上 人有目的:繁衍->计算预期差->通过奖惩机制,调节多巴胺和激素->让动物围绕着激素,展开活动。 情感怎么放,情感暂且放到第四层,是人类第一级的整个输入信息的综合信息