-
Bug
-
Resolution: Duplicate
-
P1
-
None
-
1.1.6
-
sparc
-
solaris_2.5.1
This is a 1.1.6F regression bug from Oracle.
There is a regression bug that first appeared
in JDK 1.1.6F. It did not exist in 1.1.6E.
We are seeing repaint problems with even simple
applets: dragging windows on top of other
windows leaves behind a trail of repaint garbage.
The problem seems to have been caused by a
performance enhancement made between 1.1.6E and
1.1.6F.
From JDK 1.1.6F putback report (jdk116F_0128.html):
Putback Comments from Jan 27, 1998 @ 12.47.15
Bug: 4106598
Synopsis: AWT clipping performance on Windows can be significantly
improved.
Comments: Removed the LogicalSelectClipRgn Win32 API call from
AwtGraphics::Rele
aseDC(). It is not necessary to reset the clip region since each thread
calls G
etDC or ValidateDC to set the clip region. Removing this
call provides substantial improvement when clipping is enabled.
Diffing the code shows that a clip region call has
been commented out:
Before (JDK 1.1.6E): \src\win32\sun\windows\awt_Graphics.h, Lns. 195-205
INLINE void ReleaseDC() {
if (m_pSharedDC != NULL) {
if (m_pSharedDC->GetComponent() != NULL) {
// Bug #4055623 - Performance enhancements
if (m_bClipping) {
LogicalSelectClipRgn(m_hDC, NULL);
}
}
m_pSharedDC->Unlock();
}
}
After (JDK 1.1.6F): \src\win32\sun\windows\awt_Graphics.h, Lns. 195-205
INLINE void ReleaseDC() {
if (m_pSharedDC != NULL) {
//if (m_pSharedDC->GetComponent() != NULL) {
// Bug #4055623 - Performance enhancements
// if (m_bClipping) {
// LogicalSelectClipRgn(m_hDC, NULL);
// }
// } //If loop taken away for bug # 4106598
m_pSharedDC->Unlock();
}
}
Please put this code back into the next version of
JDK 1.1.6.
Test case (2 files):
*** BadRepaint.html ***
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<BODY>
<APPLET CODE="BadRepaint.class"
WIDTH=400
HEIGHT=400></APPLET>
</BODY>
</HTML>
*** BadRepaint.java ***
import java.applet.*;
import java.awt.*;
public class BadRepaint extends Applet
{
private Frame _mFrame1;
public BadRepaint()
{
}
public void start()
{
_mFrame1 = new Frame("Frame 1");
_mFrame1.setBounds(100, 100, 300, 200);
_mFrame1.setVisible(true);
}
}
There is a regression bug that first appeared
in JDK 1.1.6F. It did not exist in 1.1.6E.
We are seeing repaint problems with even simple
applets: dragging windows on top of other
windows leaves behind a trail of repaint garbage.
The problem seems to have been caused by a
performance enhancement made between 1.1.6E and
1.1.6F.
From JDK 1.1.6F putback report (jdk116F_0128.html):
Putback Comments from Jan 27, 1998 @ 12.47.15
Bug: 4106598
Synopsis: AWT clipping performance on Windows can be significantly
improved.
Comments: Removed the LogicalSelectClipRgn Win32 API call from
AwtGraphics::Rele
aseDC(). It is not necessary to reset the clip region since each thread
calls G
etDC or ValidateDC to set the clip region. Removing this
call provides substantial improvement when clipping is enabled.
Diffing the code shows that a clip region call has
been commented out:
Before (JDK 1.1.6E): \src\win32\sun\windows\awt_Graphics.h, Lns. 195-205
INLINE void ReleaseDC() {
if (m_pSharedDC != NULL) {
if (m_pSharedDC->GetComponent() != NULL) {
// Bug #4055623 - Performance enhancements
if (m_bClipping) {
LogicalSelectClipRgn(m_hDC, NULL);
}
}
m_pSharedDC->Unlock();
}
}
After (JDK 1.1.6F): \src\win32\sun\windows\awt_Graphics.h, Lns. 195-205
INLINE void ReleaseDC() {
if (m_pSharedDC != NULL) {
//if (m_pSharedDC->GetComponent() != NULL) {
// Bug #4055623 - Performance enhancements
// if (m_bClipping) {
// LogicalSelectClipRgn(m_hDC, NULL);
// }
// } //If loop taken away for bug # 4106598
m_pSharedDC->Unlock();
}
}
Please put this code back into the next version of
JDK 1.1.6.
Test case (2 files):
*** BadRepaint.html ***
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<BODY>
<APPLET CODE="BadRepaint.class"
WIDTH=400
HEIGHT=400></APPLET>
</BODY>
</HTML>
*** BadRepaint.java ***
import java.applet.*;
import java.awt.*;
public class BadRepaint extends Applet
{
private Frame _mFrame1;
public BadRepaint()
{
}
public void start()
{
_mFrame1 = new Frame("Frame 1");
_mFrame1.setBounds(100, 100, 300, 200);
_mFrame1.setVisible(true);
}
}
- duplicates
-
-
- Closed
-