*** Issue 56392 has been marked as a duplicate of this issue. ***

Works for me on Windows, but there have been more reports of this bug on Linux.

Stan, please look at it.

It doesn't work on Linux :
Linux RH 9 / NB4.1 (200503141900) / JDK 1.5.0_02 (fcs)

Created attachment 20835 [details]
Proposed fix

martin, can you pls test the patch on linux?
thanx

It works pefrectly. But it worked without problems without the patch as well for
me ;) So reassigning to Marian who have the problem with current implementation.
Probably WM specific (fluxbox works fine).
Could you test it Marian? Thanks.

Ok, 
I will test it, but provide me something what I don't need compile/build,
provide patch as a jar file or patched module.
Thanks in advance.

I can reproduce this bug on JDS, too. Actually it is very annoying because
everytime I maximize the window, the cursor gets broken (a linux bug in JDK).
This bug causes that cursor is changing incorrectly as I hover over the IDE.
This bug was reported several months ago and closed as a JDK bug. As I have to
maximize the IDE everytime I start it up my cursor gets broken everytime.
Wonderful indeed :-)

Ok, so what is the JDK bug number?

Found it! See #50647 and #48188. JDK bug:
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=5097604

maximization is somewhat broken (sick) in swing, since one has to show the
window before setting the window flag. if you combine it with gnome which i
consider sick too,.. well... 

in our winsys initialization we show the component and then set the window as
maximized. during that time we also set listeners to the windows, to get
notified of the window state and update our model.. maybe we get some events
that we confuse for something else? on startup or shutdown.
I know about this lovely gnome feature that it calls "minimize/iconify" on all
windows on a particular virtual desktop and then when returning, it triggers
"deiconify" event. Maybe this maximization issue could be related to something
similar. (like gnome calls "restore" when exiting app or something like that).
Is the correct frame state persisted across IDE sessions?

Roman: no, it's not a JDK bug. What you found is a copy of the NetBeans bug
#48188 in the Sun's bug parade. If we really believe there's a JDK bug there, we
need to make sure it's filed against the 'java' category (preferably with a test
case that shows that problem can be reproduced with a simple Swing app).

I have several other Swing programs running on my PC and none have the 
maximized frame problem. 
Does Netbeans use a threadsafe startup of the main frame ? 
Like: 
=> 
    // Set controls on frame 
    pack(); 
    // Threadsafe startup of main frame 
    Runnable runner = new FrameShower(this); 
    EventQueue.invokeLater(runner); 
 
    private static class FrameShower implements Runnable { 
        final JFrame frame; 
        public FrameShower(JFrame frame) { 
            this.frame = frame; 
        } 
        public void run() { 
            frame.setVisible(true); 
        } 
    } 
=> 
 

I've tried patched build and it works fine.

So Standa or Martin, please commit patch to CVS.

Thanks in advance.

fixed - integrated the patch

Checking in DefaultView.java;
/cvs/core/windows/src/org/netbeans/core/windows/view/DefaultView.java,v  <-- 
DefaultView.java
new revision: 1.49; previous revision: 1.48
done

verified in NB4.1(200503170655)

Just for completeness - this is a reply on Honza's comment about the other issue
connected with mouse cursor: I have filed a bug against JDK for mouse cursor
misbehaviour through bugster as #6242833.

maximizing the window prior showing it works fine on ms windows but it doesn't
work on linux afaik.