Description of problem: I start azureus automatically as part of my gnome-session and have it minimize to the system tray. There are two problems that arise: 1) After logging out of gnome, whenever I relogin the next start of Azureus will throw a warning message up about Azureus not having been shutdown tidily. It tells me to check ~/.azureus/logs which I will attach. If I had exited azureus before logging out, I would not get this message 2) Lately when this message comes up, I can't make it go away unless I exit azureus. I can click the "hide" button, but the message stays. Azureus works fine, but the message will not go away and furthermore always stays on top of other windows. Version-Release number of selected component (if applicable): azureus-2.4.0.3-0.20060328cvs_4.fc5 How reproducible: on every login Steps to Reproduce: 1. start azureus under gnome 2. logout from gnome without exiting azureus first 3. relogin to gnome and start azureus 4. try and "hide" the message Actual results: warning message displayed that cannot be removed Expected results: Azureus should not show this warning message and if it must it should allow me to remove it. Additional info:
Created attachment 128080 [details] Debug log from a couple of days of using azureus
(In reply to comment #0) > 1) After logging out of gnome, whenever I relogin the next start of Azureus will > throw a warning message up about Azureus not having been shutdown tidily. It > tells me to check ~/.azureus/logs which I will attach. If I had exited azureus > before logging out, I would not get this message It's possible this is due to bugzilla #187686. This has been fixed and we're just waiting for a libgcj update. > 2) Lately when this message comes up, I can't make it go away unless I exit > azureus. I can click the "hide" button, but the message stays. Azureus works > fine, but the message will not go away and furthermore always stays on top of > other windows. Interesting! Thanks for the bug report. I'll try to reproduce these problems.
(In reply to comment #1) > Created an attachment (id=128080) [edit] > Debug log from a couple of days of using azureus This log file indicates that you're running with both java-gcj-compat and Sun java. Do these problems show up with both VMs? We need to eliminate that variable.
I tried azureus using both VMs. They both exhibit the exact same behavior with respect to the message appearing and with it not being responsive to the "hide" button.
(In reply to comment #4) > I tried azureus using both VMs. They both exhibit the exact same behavior with > respect to the message appearing and with it not being responsive to the "hide" > button. Ok, thanks. I was able to reproduce this with java-gcj-compat. It's the first time I've ever seen it.
One of the azureus hackers tells me that this is fixed in azureus cvs. I haven't looked for the fix yet.
Fully reproducible under KDE/x86_64/java-gcj-compat. [gilboa@gilboa-work-dev ~]$ rpm -qa | grep gcj java-1.4.2-gcj-compat-devel-1.4.2.0-40jpp_83rh.x86_64 libgcj-4.1.0-3.i386 libgcj-4.1.0-3.x86_64 java-1.4.2-gcj-compat-1.4.2.0-40jpp_83rh.x86_64 java-1.4.2-gcj-compat-javadoc-1.4.2.0-40jpp_83rh.x86_64 libgcj-devel-4.1.0-3.x86_64 [gilboa@gilboa-work-dev ~]$ rpm -qa azureus azureus-2.4.0.3-0.20060328cvs_4.fc5.x86_64
This is definitely an upstream problem. Unforuntately sourceforge.net has been broken for the past few weeks (anonymous cvs repositories aren't getting updated). So, for now, I have no access to the upstream fix (if one really does exist). Hopefully sourceforge.net will get their act together soon.
sourceforge.net is finally fixed. I've testing the latest azureus code and the problem persists. Then I inserted some debugging code to print something when the button gets pressed, and we're not getting any output. At this point I suspect a problem with SWT. I've CC'd overholt and will try to create a reproducer. Andrew - did SWT get updated in FC5 since the FC5 release? Or gtk+? If so, I can try to revert to older versions and see what happens.
Created attachment 128950 [details] Test case
I've attached a test case to this bug report. It's a simple SWT example with a button. I'm no SWT expert. Is it supposed to work? I'm not getting any console output when I press the button.
It turns out that the button isn't pressable if you create the shell in my example with SWT.ON_TOP. It works fine without SWT.ON_TOP. But we want SWT.ON_TOP...
Something is definitely going on in SWT-land. I found the following example code that creates an SWT.ON_TOP window with a close button, and the button doesn't work here either: http://www.java2s.com/Code/Java/SWT-JFace-Eclipse/SWTShellcreateanonrectangularwindow.htm I removed the swt .jar.so file, just to be sure it wasn't a gcj miscompilation bug, and was still able to reproduce the problem. I'll file a bug against libswt3-gtk2.
Rolling gtk2 back to gtk2-2.8.15-1 solves this problem.
This is fixed in libswt3-gtk2-3.1.2-1jpp_15fc.