#读书笔记 # The trilogy of Weinberg's technical thought: subverting the perfect software (1)

The reason for the test

The real (self-aware) person knows that no matter how hard they try to do a good job, it is possible to make a mistake. A true (self-aware) person also knows

It's not just that he's one of those unreliable, and the same is true of others.

In the daily development of software or the use of software products, some decisions need to be made, often these decisions are risky. We may have to "test" this way to get some information that can help us reduce risk.

At the most superficial level, people have been testing.

1. The application does not meet the constraints, such as application compatibility issues with the operating system.

2, the application has not completed my needs, such as I want a Web page tool, you give me a picture of the software is useless.

3, the application has not asked the user "user unacceptable requirements", such as the mobile phone, when installing a software, it requires you to download another

Software in order to use it.

4, the application will not be because users use it and lose confidence, such as the user in the process of using software, software crashes, and ultimately let users no longer

Use it.

5, the application will waste the user for a long time, such as the application is all right, but in the "Save" time, the software will take 1 minutes

More time to complete.

6, the application will not be because the price is too expensive to be rejected by the user.

Don't deceive yourself, we all know that software users will always find ways not to use the software they do not want.

If there is no doubt about the risk of the product, there is no reason to test it. If you have at least one question, then you will wonder: does the cost of conducting these tests outweigh the value of the answer itself? Risk is subjective because different people feel differently about the same risk.

=====================

The decision is sensual, not rational.

Because testing provides information that helps reduce risk in decisions, you might think that everyone will like and value the test. But people tend to be influenced by many psychological factors, making decisions that have nothing to do with rationality. We have all heard this: "My ideas have been shaped, do not use these facts to interfere with me." "Even we have said so.

People have an emotional inclination and don't want to find themselves making mistakes. Some managers don't want to know that their project is heading for failure. Some developers do not want others to know that there are many flaws in their code. When you try to influence these emotional factors, it's likely to backfire.

=====================

Edsger Dijkstra: "Tests can reveal the existence of defects and not indicate that they do not exist." ”

Since we cannot test all the possibilities, any actual test set is a sample of some degree: representing a part or fragment of the entire possible test set in some form. We take it for granted that it is a good representative, but this brings us to the question: "For whom is it good?" "In essence, sampling is also a psychological process, but also a perceptual process. A sample that satisfies someone may make another person feel a bit dissatisfied.

First, realize that any test set is a sampling method. Then, no matter how many resources you have, you may want to choose the most representative test set as soon as possible.

#读书笔记 # The trilogy of Weinberg's technical thought: subverting the perfect software (1)