Re: FYI: MetalComboBoxUI test fixes

[David Gilbert <david dot gilbert at object-refinery dot com>]

Hi Roman,

Roman Kennke wrote:

> Hi David,
>
>  
>
> > > > This patch brings these tests in line with JDK 1.5, and cleans up any 
> > > > custom look and feel settings at the end of each test:
> > > >  
> > > >
> > > >        
> > > thanks for fixing up all those tests for JDK1.5. Awesome work.
> > >
> > > /Roman
> > >
> > >
> > >      
> > No problem.  I'm about done for now, there still remain a few failures, 
> > see this report:
> >
> > http://www.object-refinery.com/classpath/mauve/swing-jdk15/
> >    
>
> Would it be possible to create such reports on a regular basis? Maybe we
> could install it on builder.classpath.org or something?
>  
Yes, I'm working towards that.  I'm attempting to create a version of 
the report that shows the JDK1.5 and GNU Classpath results side by side, 
split into four sections:

(1)  Pass on both runtimes;
(2)  Fail on GNU Classpath, pass on JDK1.5 (we have bugs to fix);
(3)  Pass on GNU Classpath, fail on JDK1.5 (Sun has bugs to fix - 
ideally I'd like to get the Sun bug ID tagged on these fails, but we 
should also double check that the test is not faulty);
(4)  Fail on both - in this case, we should look closely at the tests;

> > Some of these I believe are real bugs in the JDK, for example the 
> > failure in javax.swing.text.GapContent and the 5 failures in 
> > DefaultMutableTreeNode.
> >    
>
> I have looked into the GapContent failure and I am quite certain that
> this is no bug. I have sent a separate email for that one. I will also
> look at the other failures now. We should really look very hard at such
> failures, often times this is only a misinterpretation of the specs and
> the JDK is actually right.
>  
I think the GapContent failure really is a bug in the JDK (I replied to 
your other e-mail).  And I did look hard at the DefaultMutableTreeNode 
tests and I'm confident the tests match the specs for those 5 failures too.

> >  The JTable failures I wasn't so sure about, the 
> > API spec is too incomplete - do you know any other sources of info for this?
> >    
>
> I normally refer to the OReilly Swing book, both editions (they are
> quite different in some areas, while the 2nd edition covers 1.4, it
> seems they cut out lots of the details that can be found in the 1st
> edition). The Sun press JFC books are also very valueable. Anthony and
> Lillian lately found that the Swing connection also has loads of
> interesting articles that complement the API specs:
>
> http://java.sun.com/products/jfc/tsc/articles/index.html
>  
Thanks.  I have the O'Reilly Swing book (must be first edition because 
it doesn't have anything about 1.4).

Regards,

Dave