- I added a utility class to gnu/testlet/java/awt/ to take care of the
location tests using colors.
- I fixed up the other classes to use this new Class.
- I also added a new test to LightweightContainer. This test represents
the testcase that initially found the problem with lightweight
containers.
- I added more comments for the delays.

2006-02-22  Lillian Angel  <langel@redhat.com>

        * gnu/testlet/TestHarness.java
        (checkColor): Removed.
        (checkRectangleOuterColors): Removed.
        (checkRectangleCornerColors): Removed.
        * gnu/testlet/java/awt/Container/LightweightContainer.java
        (testLoc): Added comment, changed to use LocationTests.
        (testWindow): Likewise.
        (testLoc1): Added new test.
        (test): Added call to testLoc1.
        * gnu/testlet/java/awt/Frame/size1.java
        (test): Added comment, changed to use LocationTests.
        * gnu/testlet/java/awt/LocationTests.java: New class.


On Wed, 2006-02-22 at 12:25 -0500, Thomas Fitzsimmons wrote:
> On Wed, 2006-02-22 at 10:29 +0000, David Gilbert wrote:
> > Hi Lillian,
> > 
> > Lillian Angel wrote:
> > 
> > >
> > >Index: gnu/testlet/java/awt/Container/LightweightContainer.java
> > >===================================================================
> > >
> > >  
> > >
> > 
> > >+    harness.checkRectangleOuterColors(r, bounds2, Color.red, true);
> > >+    harness.checkRectangleCornerColors(r, bounds2, Color.red, false);
> > >+    
> > >+    r.delay(3000);
> > >+  }
> > >  
> > >
> > Do you know what the delay is for?  Is there some other way to write the 
> > test so that the delay isn't required?  It's just that it would be nice 
> > if the Mauve test run doesn't take longer than it has to - otherwise 
> > hackers will stop running it.
> 
> The three second delays are to help the human tester see the result,
> though we should probably enable or disable them based on if Mauve is
> running in headless mode.  The new AWT utility class that we add could
> check the SHOW_GUI_TESTS environment variable and cache its value.
> 
> We require:
> 
> +    r.waitForIdle();
> +    r.delay(300);
> 
> after showing a frame, before checking pixels in it so that GTK has had
> time to draw it.  I think this is documented in one of the other AWT
> tests.
> 
> Tom
> 
>