Sven Farrenkopf

Maneesh Godbole wrote:
Like it or not, applets reached their "end of life" long time back.

Well, I wouldn't go that far. I agree that applets "never took off", but there is obviously still some need for RIAs that require more reliability/memory/performance than an AJAX-Framework can provide. And even if applets never reached the popularity they deserved(!) I believe they will never really die as well. Even this forum still offers "Applets" as a category that has its visitors (e.g. you and me)

But I really think somewhere on the way something went terribly wrong, now that everyones first appoach to RIA is using HTML/CSS/AJAX ... with HTML5 as the latest consequence of that development. Don't get me wrong: HTML5 is a good thing ... but still a result of a wrong market-development.

Applets will probably never reach the mobile market, but I really wish (but don't expect) some revplution would happen that provides a different, more solid (non-java?) solution for RIAs ... and no HTML-based patchwork that was developed because all the lemmings walked in the wrong direction.

Sven

When it comes to the new security requirements for applets (and web start) I think it's surprisingly quiet.I expected more Applet-developers to complain ...

https://blogs.oracle.com/java-platform-group/entry/new_security_requirements_for_rias
http://www.oracle.com/technetwork/java/javase/overview/ria-checklist-2055184.html

What really bothers me: Is there any way to use self-signed certificates anymore? The second link above says:

"Self-signed certificates are an option to known communities, but each computer will need to import the certificate into its trust store."

Well ... I DID import the self-signed certificate into the trust store and I still get Warning-Messages under 1.7.0_45. Is that a bug? I doubt it.

Sven

I made some progress ...

The auth-dialog even pops up when the applet-jar-files are loaded. They're hosted on IIS 7.

When I disable "Negotiate" (=Kerberos) as authentication-provider in IIS, everything works fine. (Yay!)

So Integrated Windows Authentication via NTLM works and the jars are loaded without any dialogs.

It just doesn't work with Negotiate/Kerberos ... does the JRE not support loading Kerberos-protected jar-files?

Or do I neet to configure the jre to use kerberos? When I enter the URL of the jar into the browser, I can download the jar without authentication.

Sven

It's been a while ... and still no progress. Some applets actually work and connect without authentication, others don't.

I found some articles on this, but no real solution.

https://forums.oracle.com/forums/thread.jspa?threadID=2263798
http://tech.panorama.com/index.php/kb/19-panorama-novaview-web-access/562-java-authentication-popup

Any ideas?

Sven

Right now we have several applets running under XP / IE8 / Java 1.5 (we skipped 1.6 completely). We're now testing those Applets with JRE 1.7 and encounter some problems.

The applets send web-requests to the server, which worked fine with 1.5.

In 1.7 a usr/pwd-dialog appears. Entering my credentials, everything works fine too - though now the password is saved and I can't reproduce the problem.

Does anybody know where those passwords are stored? Somehow I can't find that list to remove my entry. Second, I don't know why I was asked for credentials in the first place. Where's the difference to 1.5? Did 1.5 use the IE-credentials and 1.7 doesn't?

Sven

Sorry for bringing this up again, but I just tried Java 7 early access ... and the bug is still there. We can't stay on JRE 1.5 forever, so has anybody any idea for a workaround?

Sven

Ok, running it on EDT did the trick!

Thanks!

Sven

Are we talking JFrames?
Do you have a reference from one frame to the other?

Sven

I'm working with quite complex TableModels which sometimes rearrange their data on certain events. When a repaint occurs during that rearrangig, the Model might be inconsistent and getValueAt() delivers wrong values. Can I prevent the repaint while rearranging? Or are there any other suggestions?

Here's some simplified(!) code as an example - my actual TableModels are much more complex.

import java.util.Vector; import javax.swing.JFrame; import javax.swing.JScrollPane; import javax.swing.JTable; import javax.swing.table.AbstractTableModel; public class TableTest { public static void main(String[] args) { new TableTest(); } public TableTest() { // Create a frame with a JTable in it. JFrame f=new JFrame(); f.setDefaultCloseOperation(JFrame.DISPOSE_ON_CLOSE); TestModel model=new TestModel(); f.setContentPane(new JScrollPane(new JTable(model))); f.setBounds(100,100,1000,800); // Set the frame visible and rearrange the TableModel immediatly after that // As a consequence, rearranging might take place while repainting is still in progress. // So getValueAt() might be called on an inconsistent TableModel // How do I prevent that? f.setVisible(true); model.rearrange(); f.setVisible(false); } class TestModel extends AbstractTableModel { int[] rowCounts={10,40,70,90}; int index=0; Vector<Integer> values=null;; public TestModel() { rearrange();}; public int getColumnCount() {return 10;} public int getRowCount() {return rowCounts[index];} public Object getValueAt(int rowIndex, int columnIndex) {return values.get(rowIndex);} public void rearrange() { int oldRowCount=getRowCount(); index++; if (index>=rowCounts.length) index=0; // Go to sleep for simulating a longer execution-time // When getValue() is called now (e.g. during repaint) an // ArrayIndexOutOfBoundsException will be thrown, // since the value-vector has not the proper size yet try {Thread.sleep(2000);} catch (Exception e) {e.printStackTrace();} // Now resize the vector values=new Vector<Integer>(); for (int i=0;i<getRowCount();i++) values.add(123); // Simplified Events if (oldRowCount>getRowCount()) this.fireTableRowsDeleted(0, oldRowCount-getRowCount()-1); if (oldRowCount<getRowCount()) this.fireTableRowsInserted(0, getRowCount()-oldRowCount-1); } } }

I'd appreciate a solution without synchronized methods in it (if there is any)

Sven

>

Rob Camick wrote: I believe the background is painted in ComponentUI.update(...)

Now THAT'S what I was looking for. Thanks!

Sven

Rob Prime wrote:Check out the source of javax.swing.plaf.basic.BasicLabelUI.

Have you read my first posting? I already checked - couldn't find it.

Sven

Ok, though my actual problem is solved, I want to rephrase my question:

Since Swing sources are open, I was looking for the line of code that actually painted the background of a JLabel. I wasn't looking for hacks, since (in my case) setBackground() worked just fine. But when I wanted to paint my own (multi-color) background in a subclass of JLabel, the sizing didn't work properyly. So I was looking for an explanation by comparing both sources. I still haven't found that particular line of code, though I found the explanation of what I did wrong in a different code-passage of Swing. (I used column- and row-spanning in my table which was the origin of my problem - the Rectangle had a bug in its calculation.)

But thanks anyway

Sven

I searched Swing-Sources but somehow I can't find where JLabel checks for an opaque Background and paints it. The closest thing I found was in BasicLabelUI:

/** * Paint the label text in the foreground color, if the label * is opaque then paint the entire background with the background * color. The Label text is drawn by paintEnabledText() or * paintDisabledText(). The locations of the label parts are computed * by layoutCL. * * @see #paintEnabledText * @see #paintDisabledText * @see #layoutCL */ public void paint(Graphics g, JComponent c) { JLabel label = (JLabel)c; String text = label.getText(); Icon icon = (label.isEnabled()) ? label.getIcon() : label.getDisabledIcon(); if ((icon == null) && (text == null)) { return; } FontMetrics fm = SwingUtilities2.getFontMetrics(label, g); String clippedText = layout(label, fm, c.getWidth(), c.getHeight()); if (icon != null) { icon.paintIcon(c, g, paintIconR.x, paintIconR.y); } if (text != null) { View v = (View) c.getClientProperty(BasicHTML.propertyKey); if (v != null) { v.paint(g, paintTextR); } else { int textX = paintTextR.x; int textY = paintTextR.y + fm.getAscent(); if (label.isEnabled()) { paintEnabledText(label, g, clippedText, textX, textY); } else { paintDisabledText(label, g, clippedText, textX, textY); } } } }

But there is no opaque-check and no background-drawing.

(I need the code for debugging my own. JLabel seems to give different heights when used in a JTable - since "normal" background-painting works fine in swing and it definetly needs the height to paint it, I was looking for the code)

Sven

For anyone who might have the same problem some day, I reported the bug to sun:

http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6828208

Sven

Well, I'm into Swing/Applet-Development for some years now, so not actually "new". And I didn't expect it to work on all platforms - but IE is standard in our company, so "%" saves me some scripting. That's why I asked if it was specified this way. Looking it up, I just found suggestions that you shouldn't use it - but no specification. (Which doesn't mean it doesn't exist - I just googled 10mins!)

Anyway ... pixel-sizes don't help - same effect. I'm aware the guys from sun didn't expect developers to put a JTextfield in a JMenu. I even found a posting on sun's Bug-DB (in a different context) where a supporter was totally surprised that people could have such an idea. But browsing swing-codes I can't see why it shouldn't work. Having a JTextfield in a simple JPopupMenu (without JMenu as parent) always works - no matter how big or small the applet is and whether the popup is inside or outside the applet- or browser-area. But that won't solve my problem since I need the JMenu.

Our company can't upgrade to JRE 1.6 unless I either solve this problem or write them a different user-interface (which they have to accept first).

Sven