Book Image

Software Test Design

By : Simon Amey
Book Image

Software Test Design

By: Simon Amey

Overview of this book

Software Test Design details best practices for testing software applications and writing comprehensive test plans. Written by an expert with over twenty years of experience in the high-tech industry, this guide will provide you with training and practical examples to improve your testing skills. Thorough testing requires a thorough understanding of the functionality under test, informed by exploratory testing and described by a detailed functional specification. This book is divided into three sections, the first of which will describe how best to complete those tasks to start testing from a solid foundation. Armed with the feature specification, functional testing verifies the visible behavior of features by identifying equivalence partitions, boundary values, and other key test conditions. This section explores techniques such as black- and white-box testing, trying error cases, finding security weaknesses, improving the user experience, and how to maintain your product in the long term. The final section describes how best to test the limits of your application. How does it behave under failure conditions and can it recover? What is the maximum load it can sustain? And how does it respond when overloaded? By the end of this book, you will know how to write detailed test plans to improve the quality of your software applications.
Table of Contents (21 chapters)
1
Part 1 – Preparing to Test
6
Part 2 – Functional Testing
13
Part 3 – Non-Functional Testing
17
Conclusion
Appendix – Example Feature Specification

Identifying worse-case scenarios

To maximize the efficiency of your testing, identify the worst-case scenarios and just test them. If they work, then all simpler cases should be successful as well. Many tests will be a subset of complex tests, so it’s fastest just to perform the complex ones. For example, if your users can have multiple copies of many different relationships, add multiple instances of every kind of relationship. If that works, then having just one relationship should work, as well as having multiple copies of just one type because those are simpler cases.

One exception is with newly written code, where you can expect to find issues. Then, if you test the worst-case scenario, you’re likely to hit a problem and possibly several simultaneously. In that case, it’s faster to gradually build up to the most complex configuration. Then you can isolate and debug issues as they arise, and you know which step triggered them. However, once each aspect of...