Re: [PATCH] fixes to swing button implementation

[graydon hoare <graydon at redhat dot com>]

Sascha Brawer wrote:

> But probably, it would best tp use
> javax.swing.event.EventListenerList here (which, by coincidence, became
> functional this afternoon...). 

I'm using this now.

> Weird. From looking at the J2SE 1.4.2 API specification, I would expect:
>
>   protected int stateMask; // e.g. ARMED | SELECTED
>   protected int mnemonic;
>   protected String actionCommand;

yes, I've updated the model (and component, and UI, etc.) to work more 
in line with the J2SE docs. if you want to look at what I wound up 
committing, I'd certainly appreciate further commentary by mail.

> It seems strange that these colors are hard-coded. Shouldn't they come
> from the UIManager?

perhaps, I left them there because they were there when I first opened 
up this class. if it needs to be delegated I'll change it.

> A final comment: Please, please, please write documentation and Mauve
> test cases. With javax.swing.DefaultBoundedRangeModel, I've just made the
> experience myself how immensely useful this is for finding bugs, and for
> making sure that changes don't break things.

ok, the version I committed to gcj and just merged to classpath has 
extensive javadocs and is reformatted using jalopy. I have a simple 
mauve test which works but doesn't exercize every aspect of the button; 
I'll extend it across the next few days and fix up any sense in which my 
code doesn't conform with the sun stuff.

-graydon