Google recommends to use the text fixture constructor/destructor when possible instead of SetUp()/TearDown() (https://google.github.io/googletest/faq.html#CtorVsSetUp). Assuming I do it this way, what is the use of even using a test fixture? How are the following different, and what is the advantage of the first?

TEST_F(MyFixture, MyTest) {
  ... run test, using member functions of MyFixture for common code ...
}

TEST(MySuite, MyTest) {
  MyFixture fixture; // will call ctor
  ... run test, using public functions of MyFixture for common code ...
} // will call dtor

CodePudding user response：

The advantages are visible when there are more than one TEST/TEST_F. Compare:

TEST(MyTest, shallX)
{ 
   MyTest test;
   test.setUpX();
   test.objectUnderTest.doX();
}

TEST(MyTest, shallY)
{
   OtherTest test;
   test.setUpY();
   test.objectUnderTest.doY();
}

with

TEST_F(MyTest, shallX)
{ 
   setUpX();
   objectUnderTest.doX();
}

TEST_F(MyTest, shallY)
{
   setUpY();
   objectUnderTest.doY();
}

What we can see, are:

1. DRY (don't repeat yourselves) principle is followed. You do not have to repeat creating of some test-helper object. In TEST_F - the macro creates this instance.
2. The code is safer with TEST_F. See MyTest..shallDoY -- have you spot that wrong test-helper object is used, not the one that testname is promising.

So it is better to use TEST_F if your tests require some test-helper class. If not - then use TEST.