🚀 go-pugleaf

Dave Howorth wrote:

> Once I got rid of the sizing problem, I fixed this as well. My code looks
> like this:
> 
>   jList.setModel(model);
>   jScrollPane.getViewport().invalidate();
> 
> Heaven knows why it works. What I'd really like to see is a definitive,
> coherent explanation of these incantations :-)  ...
> Though you'd think it was pretty much a no-brainer that the layout would
> need to be redone after you change the entire list contents. 

There's no way for the JList to know that it is contained
in anything.  Hence, since you set up the containment 
hierarchy (inside a JScrollpane), you have to handle the
telling the containers that they are invalid and need to 
be layed out again (and repainted).  

If they were to assume a JScrollPane/JViewport container, and
do a relayout in JList.setModel(), what would happen if someone 
assigned a parent that doesn't behave as a JScrollPane might to 
such a request?  I would guess that the Swing team is thinking
along the lines of greatest flexibility and fewest assumptions
in this regard.  Just a guess.

The simplest rule to remember with Swing components is to 
invoke "validate()" and/or "invalidate()" when changing 
properties of a component.

> But I suspect we'll see a slew of JDK 1.2 books around Xmas :-), 
> so I'm not rushing to the shops just yet, ...

I just sit in Barnes & Noble, and read like crazy.   You'll find
this kind of stuff in just about any in-depth Swing book.


Paul Furbacher