Name: dbT83986 Date: 12/14/98


There is a problem with swing distributed with 1.2 installation. Mouse movements over components destroys the displayed
graphics in a way that as if the actual sprite ( area below the mouse pointer ) is being copied on a wrong place as the mouse
moves.

===================================
REVIEW NOTE 12/13/98 - User submitted additional information; still no source, no duplicates; posting as
new bug with low priority.

I am sorry I didn't follow the procedure, but I quite guessed you'd have
similar reporting already.

I have 2 computers:

1. Acer Extensa 390 Notebook ( P166MMX/32RAM and Chips-and-tech video card -
DSTN display )
2. Intel Pentium II/300 Mhz/64 RAM - and what's most interesting Cirrus
Logic ( CLGD5420 ) old bios release video card
My initial AGP video card burned so I'm using that until replacement

The reason I am telling you my video configuration is ( to my understanding
the source of the problem ). I also run Win95.

I quite figured what is happening, when the object ( such as button )
changes state ( clicked ) it is repainted and mouse pointer is hidden and
after repainting displayed, only it doesn't display again and when I move
the mouse the sprite ( under the mouse ) is repainted ( copied ) to a wrong
place.

Curiously, this behavior is only on my notebook, unlike desktop where I have
different problem ( of the same source ). On my desktop when I start most of
swing apps ( on 256 color ) I have 16-color like display ( only it's not 16
color but rastered ).

Now bare with me, here is the solution and actual pinpointing of the
problem:

When I minimize such rastered application and then restore, my computer
blocks-totally, I have to restart it ( ctrl-alt-del ) however I realized
where is the problem:

On m notebook, I start the ( swing ) application in 256 colors display mode
( and I still have copying of sprites ) and then I minimize it, go to my
windows settings and change color depth to 24bit, apply and when I
restore/maximize application, everything is working fine and dandy.

In conclusion I figured that there may be some mishandling of bit-panes
because apparently different color depths react differently. However it
doesn't help starting app in 24bit mode because the problem is the same.

Only actual minimize-(reset the display to better c-depth)-restore helps. It
won't work if we reset to lower depth.

Unfortunately I can't supply code because it wouldn't tell you nothing but
let me just say, the problem appears in all swing examples but on my desktop
those that actually use tab-pages object cause it to block faster.

There are no error messages or anything it either smears or blocks.
Curiously I didn't have the same problem with jdk1.1.7A and swing 1.0.3. I
would just patiently wait for it to be fixed while working with 1.1.7A but I
need some features of 1.2 therefore I am a bit anxious...

Regards,
Dragomir Milivojevic
(Review ID: 47409)
======================================================================

Another customer reporting the sanme bug... The customer writes,
I hav experienced the problem and have done a bit of research and found that the problem exists in Java 1.2.2. We currently see the problem on Toshiba laptops, although research shows that others have seen it on other laptops and one desktop, so a driver/video chip problem would not seem to be the culprit. Although one source seemed to indicate his desktop also locked up, we haven't seen that.

Almost any Swing application (or applet, I've seen the Java 1.2 plug
also do this) will demonstrate the behavior; but the key is it has to be run on
the specific hardware - you will not see this on a Sparc or most desktop PCs.

I reiterate that our requested priority is low (I am not requesting
escalation), we do not currently support laptops or PCs, but this does have a
demonstration impact, since laptops are portable. I suppose my main purpose is
to let someone know that it is still a current problem and can only be
reproduced on specific hardware, which the bug report did not seem to recognize.

sv38254

12/10/99 update


Here's the test case that demonstates the problem on Toshiba Tecra 730 CDT laptop, the configuration specific as below:

Pentium processor
48 MB RAM
2 GB disk drive
Chips & Technologies 65550 PCI display adapter
ViewSonic 21 PS monitor (bug also occurs on the Tecra's laptop screen)
Windows98 (bug also occurs on Windows95 but not WindowsNT)

The test case <<Scrtest.java>>

Note that nearly every Java 1.2 application/applet using Swing exhibits the bug (haven't tried with pure AWT) and may happen with other laptops as well.

import java.awt.*;
import java.awt.event.*;
import javax.swing.*;

public class Scrtest extends JFrame {
private JTextField txt;

Scrtest() {
super();

setSize(200,140);
getContentPane().setLayout(new GridLayout(3,1));

JLabel lab = new JLabel("Java tester");
getContentPane().add(lab);

txt = new JTextField(20);
getContentPane().add(txt);

JButton but = new JButton("Finished");
getContentPane().add(but);

but.addActionListener(new ActionListener() {
public void actionPerformed(ActionEvent e) {
System.out.println("Text is " + txt.getText() );
System.exit(0);
}
});

setVisible(true);
}

public static void main(String args[]) {
new Scrtest();
}
}



sandhya.vora@Eng 1999-12-10