I thought I'd let you in on an observation we made, regarding where to place
context.assertIsSatisfied().

We are quite happily using jMock 2.4 and our tests typically look very
basic, like:

  @Test
  public void testOnNewCall() {
    // Expectations
    InitialState state = new InitialState(mockCdgContext);

    context.checking(new Expectations() {
          {
            one(mockCdgContext).setState((NewState)
with(instanceOf(NewState.class)));
          }
       });

    // Execute
    state.onNewCallEvent(event);

    // Verify
    context.assertIsSatisfied();
  }

We set up quite a lot of tests and everything worked fine. In a fit of
"over-do", we re-factored out the context.assertIsSatisfied() to a tear-down
method, like:

  @After
  public void tearDown() {
    context.assertIsSatisfied();
  }

Things still worked just dandy and we felt cool.

A few days later things got nasty. We added some more tests, but no longer
got any useful error messages in Eclipse. It most often just reported that
"java.lang.AssertionError: not all expectations were satisfied" with the
corresponding stack trace pointing out what expectations that were never
invoked, but no clue to why. After a few hours of muttering and profanities,
we noted that we didn't have the same problem in some old classes where we
hadn't implemented the tear-down approach.

We put context.assertIsSatisfied() back into the test method and voilà! We
got error messages like
"java.lang.AssertionError: unexpected invocation:
Scenario.createNewState()", clearly pointing out the problem.

I don't know if this is a well known no-no, a bug or just stupid thinking on
our behalf. Just hoping to save you guys some hours of frustration.

Cheers,

/JMK

--
Sent from the jMock - User mailing list archive at Nabble.com.


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

  http://xircles.codehaus.org/manage_email