How to reproduce:
1) Go to a Windows 95 system
2) Go to jfc/demo/SimpleExample
3) Run SimpleExample (e.g. java SimpleExample)

4) Press the Hello,World button and move mouse.
   Press the radio buttons.
5) The block of pixels vacated by mouse will be
   incorrect, OR the mouse arrow will be painted
   over.

This is for Java 1.2beta2, and swing0.7 and
swing1_0_1
(Review ID: 26997)
======================================================================

  xxxxx@xxxxx   1998-12-10:

FYI:

Date: Tue, 08 Dec 1998 19:52:24 -0500


Subject: Fwd: Bug Id  4129096 
Mime-Version: 1.0

This seemed appropriate to forward.

>Mailing-List: contact swing-  xxxxx@xxxxx  ; run by ezmlm
>Delivered-To: mailing list   xxxxx@xxxxx  
>Date: Tue, 8 Dec 1998 16:39:09 -0800 (PST)
>From: <  xxxxx@xxxxx  >
>To: Swing! <  xxxxx@xxxxx  >, Advanced Java <advanced-  xxxxx@xxxxx  >
>Subject: Bug Id  4129096 
>
>
>http://developer.java.sun.com/developer/bugParade/bugs/4129096.html
>
>	This is a serious bug in Windows and it still exists in the
>JDK1.2 release.  It basically means that if you use Swing components in
>your apps, any Windows users who use themes or non-standard cursors (a
>LARGE crowd of people) will have horrible mouse artifacts and they will
>think your application is very buggy.
>
>	Unfortunately, this bug is also closed (non-reproducible) because,
>I suspect, the tester was not using a theme in windows.
>
>	Does anyone know who to write to at Sun to re-open this bug?  Or
>is there a duplicate bug in the parade that I'm not finding?
>
>==============================================================================
>  xxxxx@xxxxx         S/W Developer             http://www.bridgeway.com
>  xxxxx@xxxxx              Webmaster         http://www.ad1440.net/~devnull
>==============================================================================




This bug occoured since 1.2 (still in 1.2.1). It is as huge as I 
thought the bug would be detected before (or, what I think up to
now in a small case, my configuration must be wrong?)

Each time, using a non-standard mouse cursor (as example, 
3d-cursor came with any Windows distribution 95, 98 and NT), all
light-weight objects in GUI overwrite the mouse and leaving
droppings or small bitmaps, that were left under the mouse.

This occours on all Java-version I tested since 1.2 with all non-
standard cursor (such as multicolored or animated) with all 
lightweight objects.

The bug isn't as important for security or such, but nasty.
(You can't use any Java-Programm with non-standard cursors).
======================================================================

No work around.

This is a very serious and frustrating bug in
real applications. I *cannot* release my java
application if this cannot be improved.

*Please* let me know if this is unique to my
system. I cannot believe that this would be
acceptable by other developers.
======================================================================




Just use the simple standard mousecursor
======================================================================

I believe this is a 2d bug.
  xxxxx@xxxxx   1998-04-15

Ahter memory lickage found in J2D was fixed by hung, the symptoma disappeared.
  xxxxx@xxxxx   1998-06-23

Re-investigated amongst with 4196958. Seems to be a duplication.
  xxxxx@xxxxx   1999-04-29

Other info: The problem only happens when the
Win95 is setup with non-default cursors such as
those supplied by the Microsoft Plus! package.
The work around is therefore to set the Win95
cursor to be the "default" cursor:
Control Panel->Desktop Themes->Pointers/Cursors.
Now Swing/Java is back to being great.

I think this is still a serious bug.  Please read the
final paragraph.
I ran into this problem with the 1.2RC2 release.
I was quite shocked by the "holes" left by my
cursor, and wondered how the entire JFC suite
could possibly be so "buggy".
Our Windows users use many varied Windows 95 Desktop Themes.
Forcing them to go to the Default Windows Desktop configuration will stand in
the way of
implementing Swing in our Intranet.

This bug still exists in JDK 1.2 RELEASE.
Try it with jdk1.2/demo/Animator for example.
I do have my cursors back to default as suggested
under Win98, and this does NOT fix it.

I would guess that original reporter was using a desktop theme in windows,
whereas the tester was not.  I am also having this exact problem on both of my
PC's (one is Win98 and the other is Win95).
Both use desktop themes.

This problem is still here in the release version!  Please reopen this bug
immediately.

This problem should not be closed as it is unacceptable to force users to use
standard cursors when running applications! Definitely a MAJOR bug

I agreed with dion, chevy and "dev/null". It appears under 
Windows98 but not under WinNT. When I switch 
display resolution when program is run, this bug is 
dissapeared, but after new run -- again exists.

According to bug 4183744, this problem stems from ATI Mach64 cards (which all
of my compters have).  Perhaps this is not a Swing problem, but an ATI problem?

I will add myself to the list of complainants.
I find this to be a WIN95/WIN98 problem - the
applet I have created works fine on WNT 4.
My client has worked fairly closely with me on
this issue. We have reproduced this on four different
systems.
I have seen other comments that display resolutions
and color settings affect display problems. I'll
get back after I have done some tests of these
options.

3/21/99 - As promised: I was able to correct my
specific problems by reducing the harward acceleration
setting under ControlPanel\System\Performance\Graphics.
This may help others as well.
While this is clearly a system tuning issue, it
may be advisable for Javasoft to get into the
business of developing hardware compatibility
lists, putting the onus on the hardware vendor to
correct the deficiencies. As indicated by others
above, this kind of problem is a significant drag
on the delivery of solutions to the field, or even
for general use (very few web surfers are going
to be savvy enough to adjust the graphics acceleration
to deal with Java 2D performance, and those that will
would probably just as well not bother). This is the
kind of thing that can kill the market for small
developers trying to deliver to non-technical markets
(home or small business).

I was able to fix my problems by reducing the
video accelrator to "none" as Dr. DPC mentioned.
I believe that this needs to be addressed
further !!

I had the same problem on my Dell notebook but I was not using any themes or a
custom cursor so I was going crazy until I noticed the "Mouse Pointer
Trail" setting which comes enabled as default.  I disabled mouse pointer
trail by unchecking the "Show pointer trail" in the Mouse control
panel and the problem just disappeared without any need to reboot.
So the checklist for the customers until a fix comes out is:
1. Use Windows standard theme.
2. Use Windows standard cursor.
3. Disable mouse pointer trail.
Don Park
Docuverse

It is not an ATI problem, it happens on my S3-based  card under Win95 and on my
laptop which has a Cirrus Logic card (and Win98). On the laptop it  goes away
when I remove pointer trails, but on my stationary system the problem persists
no matter what I do to my pointers, even with all hardware acceleration
removed.

I have the same problem and I've tried
everything listed here.  It is annoying and should
be fixed ASAP if anyone is going to use SWING
components.
Running on Windows 98, I have the problem, running
on Windows NT or Linux, no problems at all.  I use
the jdk provided by Sun for Windows and the jdk
provided by www.blackdown.org for Linux.

It is utterly outrageous that Sun would release Swing with
this glaring bug. It quite simply means that no commercial
applications that could run under Win 95/98 can be developed
because users can not be asked to adjust system settings and
the mouse mess left is too ugly to be acceptable. I've been
waiting two years for Sun to get Java to a point that I can
use it commercially as I'm really excited about it's
potential but things seem to be getting worse (though Swing
is great in theory). It's also amazing that menu hot keys
don't work under Solaris. They work fine under Windows.
Another really major bug is the apparent inability to focus
a component such as the BCB SetFocus() command. This is an
essential for many applications. None of the methods offered
by the JDK actually can custom focus a component on start up
unless I missed something. I'd be grateful for some fixes
for these problems. I wonder if Sun developers ever look at
these pages?

I also have this problem.  I run Windows98, have a Voodoo 
Banshee based video card, and have tried both 1.2 and 
1.3RC1 JVMs.  The problem persists.  changing screen 
resolutions sometimes help, haven't tried putting my mouse 
cursor to normal, but I agree that seems to be more of a 
hassle to the average user then running a program.

I'm running Win95 with NO THEMES and NO SPECIAL MOUSE POINTER and I am running into this 
problem.  If I try reducing graphics acceleration, my system is unable to boot up, so that is not a
viable workaround for me.  I'm using a 2MB STB Trio card, which, while somewhat old, should
certainly be able to display a simple GUI! Heck, it displays Windows apps just fine!
With bugs such as this, I certainly can't justify using Java to write apps.  Write-once-run-anywhere-
as-long-as-anywhere-isn't-windows?  Please reopen and investigate further...

This is still a problem under JDK 1.2.2, using Win95 with 
custom cursors.  Please re-open and fix the problem.

The work around does "band-aid" the problem though.

Here's a workaround that seems to work for me.  Start the 
virtual machine with the following option:

-Dsun.java2d.noddraw=true

example:

java -Dsun.java2d.noddraw=true myclass