Community
Participate
Working Groups
On GTK, Eclipse 3.3 M3 (also reproducable in HEAD) Right click in the package explorer, New > Other... The dialog fills whe entire width of the screen. This regression appears to have been caused by the patch to bug 19462, and it affects all WizardDialogs opened without first creating the Shell. Here is a stack trace when the incorrect size was set: Shell(Control).setBounds(Rectangle) line: 521 WizardDialog.setShellSize(int, int) line: 1259 WizardDialog.update() line: 1183 WizardDialog.showStartingPage() line: 1125 WizardDialog.createContents(Composite) line: 516 WizardDialog(Window).create() line: 426 WizardDialog(Dialog).create() line: 1111 NewWizardAction.run() line: 176 NewWizardAction(Action).runWithEvent(Event) line: 499 ActionContributionItem.handleWidgetSelection(Event, boolean) line: 545 ActionContributionItem.access$2(ActionContributionItem, Event, boolean) line: 490 ActionContributionItem$5.handleEvent(Event) line: 402 EventTable.sendEvent(Event) line: 66 MenuItem(Widget).sendEvent(Event) line: 1095 Display.runDeferredEvents() line: 3226 Display.readAndDispatch() line: 2898 Workbench.runEventLoop(Window$IExceptionHandler, Display) line: 1945 Workbench.runUI() line: 1909 Workbench.createAndRunWorkbench(Display, WorkbenchAdvisor) line: 425 PlatformUI.createAndRunWorkbench(Display, WorkbenchAdvisor) line: 149 IDEApplication.run(Object) line: 95 PlatformActivator$1.run(Object) line: 78 EclipseAppLauncher.runApplication(Object) line: 104 EclipseAppLauncher.start(Object) line: 74 EclipseStarter.run(Object) line: 348 EclipseStarter.run(String[], Runnable) line: 165 NativeMethodAccessorImpl.invoke0(Method, Object, Object[]) line: not available [native method] NativeMethodAccessorImpl.invoke(Object, Object[]) line: 39 DelegatingMethodAccessorImpl.invoke(Object, Object[]) line: 25 Method.invoke(Object, Object[]) line: 324 Main.invokeFramework(String[], URL[]) line: 341 Main.basicRun(String[]) line: 285 Main.run(String[]) line: 987 Main.main(String[]) line: 962
Created attachment 54923 [details] Patch includes the changes I showed to Karice This patch demonstrates the use of GridData.exclude in order to adjust the preferred size of the shell rather than forcing an immediate resize. I've confirmed that this fixes bug 166544, but I don't know if it causes a regression in bug 19462. This is mainly just to demonstrate the pattern.
Note that this might be hard to see on windows since Windows chooses different default shell sizes. However, you should be able to confirm that the listener on line 486 of Window.java (the one that sets resizeHasOccurred = true) never gets called when opening a WizardDialog.
This one was my doing, so I will take it.
Stefan, I tried this using M4 on Linuk/GTK (fresh workspace) and the dialog is not full screen width. Does this happen in a fresh workspace, or after a particular wizard was finished? Is it possible this only occurs with dual monitors (which I don't have)?
Re: comment 4 This has to do with the window manager's default size of uninitialized shells. Since this is determined by the window manager, it may have to do with your choice of window manager, screen resolution, default font, and contents of Xorg.conf (yes: it is probably affected by multimonitor). Because so much can affect the window manager's decision about window placement, it would be a lot easier for you to verify this by putting a breakpoint in Window.java as I suggested in comment 2 than by actually reproducing the huge window. (Reproducing it would require figuring out which window manager or X server property causes it to happen, which would take some trial and error and is probably unnecessary since I'm offering access to a machine that can reproduce it). FYI, I'm using KWin as my window manager, and I can attach copies of my Xorg.conf and version numbers for the relevant packages if you really want to build your own system that can reproduce it.
If you can verify that the patch I attached doesn't cause a regression in bug 166544, you might consider using it.
I can confirm that this happens on ubuntu and redhat, both using KDE.
Tod: Karice suggested that you review the attached patch. Reassigning.
This patch actually turns off the progress bar completely. We will need to revisit it before it can be released.
Would it be possible to just revert the change the originally caused this bug? This bug is much more severe than bug 166544, which was really an enhancement.
Sorry - bug 19462 is the one that caused this problem.
The patch in Bug 19462 has the rollback updated
Fixed by the rollback of Bug 19462
*** Bug 174545 has been marked as a duplicate of this bug. ***
Verified in I20070430-0010
*** Bug 180745 has been marked as a duplicate of this bug. ***
