One of the many features provided by GTest is the ability to generate test at runtime. One useful application of this is that you can execute data-driven testing by obtaining a list of test file names, then generating a test for each of them. It’s actually very simple to do this in GTest – they call these value-parameterized tests (see documentation). Here’s a snippet of code to demonstrate how little overhead is involved:
#include <gtest/gtest.h>
using namespace testing;
// placeholder - could be used for test setup/cleanup
class MyTest : public ::testing::TestWithParam<std::string>
{
};
TEST_P( MyTest, IntegrationTest )
{
std::string test_file_name = GetParam();
// open file and run the integration test
}
std::vector<std::string> test_file_names();
INSTANTIATE_TEST_CASE_P( MyLibraryName, MyTest, test::ValuesIn( test_file_names() ) );
On one hand, this approach isn’t really in the spirit of unit test – data-driven tests have a habit of quickly escalating to integration tests between libraries. However, such integration tests also have their place in a testing strategy, and this is a neat way to accomplish it with minimal overhead.
In The Drop, Harry Bosch is working for the cold cases department and suddenly finds he is called on to investigate two cases at once. The first is an apparent suicide of the son of a leading politician – with all the accompanying fallout dealing with a long-term enemy of the police force could bring. The other is a puzzling murder in which the suspect matching DNA from the scene would have been a child at the time of the crime. In this book, Harry meets Hannah Stone, the therapist of one of the suspects. It also features his daughter, Maddie.
musingstudio 