The purpose of this short chapter is to, with a "learning by example" approach, give the newcomer a chance to get started quickly writing and executing some first simple tests. The chapter will introduce some of the basics, but leave most explanations and details for the later chapters in this User's Guide. Hopefully though, after this chapter, you will be inspired and unintimidated enough to go on and get into the nitty-gritty that follows in this rather heavy User's Guide! If you're not much into "learning by example" and prefer to get into more technical detail right away, go ahead and skip to the next chapter. Again, the basics presented here will be covered in detail in later chapters.

This chapter also tries to demonstrate how dead simple it actually is to write a very basic (yet for many module testing purposes, often sufficiently complex) test suite, and execute its test cases. This is not necessarily obvious when you read the rest of the chapters in the User's Guide.

A quick note before we start: In order to understand what's discussed and examplified here, it is recommended that you first read through the opening Common Test Basics chapter.

Execution of test cases is handled this way:

Successful vs unsuccessful test case execution.

For each test case that Common Test is told to execute, it spawns a dedicated process on which the test case function in question starts running. (In parallel to the test case process, an idle waiting timer process is started which is linked to the test case process. If the timer process runs out of waiting time, it sends an exit signal to terminate the test case process and this is what's called a timetrap).

In scenario 1, the test case process terminates normally after case A has finished executing its test code without detecting any errors. The test case function simply returns a value and Common Test logs the test case as successful.

In scenario 2, an error is detected during test case execution which causes the test case B function to generate an exception. This causes the test case process to exit with reason other than normal, and as a result, Common Test will log this as an unsuccessful test case.

As you can understand from the illustration above, Common Test requires that a test case generates a runtime error to indicate failure (e.g. by causing a bad match error or by calling exit/1, preferrably through the ct:fail/1,2 help function). A succesful execution is indicated by means of a normal return from the test case function.

As you've seen in the basics chapter, the test suite module implements callback functions (mandatory or optional) for various purposes, e.g: Init/end configuration function for the test suite Init/end configuration function for a test case Init/end configuration function for a test case group Test cases The configuration functions are optional and if you don't need them for your test, a test suite with one simple test case could look like this:

      -module(my1st_SUITE).
      -compile(export_all).

      all() ->
          [mod_exists].

      mod_exists(_) ->
          {module,mymod} = code:load_file(mymod).

In this example we check that the mymod module exists (i.e. can be successfully loaded by the code server). If the operation fails, we will get a bad match error which terminates the test case.

If we need to perform configuration operations in order to run our test, we implement configuration functions in our suite. The result from a configuration function is configuration data, or simply Config. This is a list of key-value tuples which get passed from the configuration function to the test cases (possibly through configuration functions on "lower level"). The data flow looks like this:

Config data flow in the suite.

Here's an example of a test suite which uses configuration functions to open and close a log file for the test cases (an operation that would be unnecessary and irrelevant to perform by each test case):

      -module(check_log_SUITE).
      -export([all/0, init_per_suite/1, end_per_suite/1]).
      -export([check_restart_result/1, check_no_errors/1]).
      
      -define(value(Key,Config), proplists:get_value(Key,Config)).

      all() -> [check_restart_result, check_no_errors].

      init_per_suite(InitConfigData) ->
          [{logref,open_log()} | InitConfigData].

      end_per_suite(ConfigData) ->
          close_log(?value(logref, ConfigData)).

      check_restart_result(ConfigData) ->
          TestData = read_log(restart, ?value(logref, ConfigData)),
          {match,_Line} = search_for("restart successful", TestData).
      
      check_no_errors(ConfigData) ->
          TestData = read_log(all, ?value(logref, ConfigData)),
          case search_for("error", TestData) of
              {match,Line} -> ct:fail({error_found_in_log,Line});
              nomatch -> ok
          end.

In this example we have test cases that verify, by parsing a log file, that our SUT has performed a successful restart and that no unexpected errors have been printed.

To execute the test cases in the test suite above, we could type this on the Unix/Linux command line (assuming for this example that the suite module is in the current working directory):

      $ ct_run -dir .

or

    $ ct_run -suite check_log_SUITE

If we want to use the Erlang shell to run our test, we could evaluate this call:

      1> ct:run_test([{dir, "."}]).

or

      1> ct:run_test([{suite, "check_log_SUITE"}]).

The result from running our test is printed in log files in HTML format (stored in unique log directories on different level). This illustration shows the log file structure:

HTML log file structure.

You will find detailed information about the basics introduced here in this chapter in the following chapters in the User's Guide, as well as presentations of many more useful features. Have fun!