Python3中使用单元测试和正常做测试的区别

Question

What is actually the difference between using unit tests and normal tests?

By normal tests I mean, using the if statement for example to determine if the calculation is equal to the desired answer if it returns false, we can raise AssertionError

Answer 1

Let's use a simple piece of code as example:

def square(a):
   return a*a;

Tests that aim at finding bugs in this function in isolation are unit-tests. This is independent of how you actually implement them: If you just write an if statement like if (square(3) != 9) and raise an AssertionError as you say, then it is a unit-test. If instead you implement the same test using unittest by calling assertEqual , then it is also a unit-test.

In other words, whether or not you use a (so-called) unit-test framework is not a criterion for whether your tests are unit-test or not. In fact, despite the names of the frameworks ('unittest' in Python, 'JUnit' in Java, ...) these frameworks can be used for unit-tests as well as for other tests, like, integration-tests. The names of those frameworks are therefore a bit misleading.

So much for the original question about the difference between unit-tests and normal tests. In one of the comments you make it clearer that you actually want to know what is better: To use or not to use a test framework. And, the answer is clear: Definitely go for a test framework.

After writing only a few tests 'by hand', that is, without a test framework, you will see that there is a lot of duplication in your test code: You compare results with if - that is not so much different. But then, in the success case you write some 'passed' message, with the test name, in the failure case you write a 'failed' messages, again with test name and in this case also with some extra information about what was the actual and what was the expected result. And - did you think about the case that the code under test exits with an exception? So, you also have to wrap it with a try/catch block to make sure unexpected exceptions lead to a 'failed' result, again with some helpful diagnostic information.

And so on... All this and more is taken care of by the test frameworks for you.

Answer 2

From your question, you assume there are two kinds of test types: unit tests and normal test . In fact, there is a lot of types of tests. The unit test is one of them. You can read here other kinds of tests.

If we assume 'normal test' is 'black box test'. It is testing software without source code. For unit tests, there is another part of the software that tests the integrity of the source code itself. What does that mean? You wrote a function that sums two numbers. For the unit test, you are writing another function to check whether your first function works

# Your Function
def my_sum_function(x,y):
    return(x+y)

# Your Unit Test Function
def test_sum():
    assert my_sum_function([1, 2]) == 3, "Should be 6"

Not only for Python but also for all programming language logic behind the test is the same.

Answer 3

For quality control. If you are coding an app called "app1" with a function f1 and an other function f2 (f2 is using f1) inside you own library

Without unit test: You can make a change in function f1 test it, it will work in f2 but you will not notice that in you program "app0" you wrote 2 weeks ago it will make you app crash because in your old app "app0" you have a function f3 and f3 is also using f1....

With unit test: You can catch that type of nasty bug because the test for the function f3 will not pass anymore and you IDE will tell you that when you change the function f1 in "app1"

I hope it make sense (english is not my native language)