Testing

Broadly speaking, there are two types of test program you can run: unit tests, and integration tests.

Unit tests

Unit tests are standalone programs with names of the form <code unit under test>_test and are implemented in the ./unittest subdirectory of a package. They're written using the Boost.Test library. For complete documentation on the Boost.Test library, you can look at the Boost documentation; for substantially quicker but still useful intros, take a look here and here.

While you can run a unit test program without arguments, some arguments are very useful - in particular, those that control which test cases are run, and those that control the level of logging output. To maximize the output to screen, add the argument -l all, and to minimize it, add -l nothing. Other intermediate levels are covered in the Boost documentation. To skip a test, you can add an argument of the form --run_test=\!test_you_dont_want; note the exclamation point needs to be escaped with a backslash since the shell interprets it as a special character otherwise.

Integration tests

Integration tests look at how various units of code interact; unlike unit tests, there's no common approach to running them. As such, a list of integration tests and how to run them can be found in the list below:

• Queue testing
• dummy_test_app: A simple application that statically instantiates a DummyModule DAQModule, which simply takes the command "stuff" to echo a string
• producer_consumer_dynamic_test.json: A sample configuration file for daq_application which runs a version of the Simple readout application