Description of problem: With more than one keyboard installed and any other than the first keyboard active, in the attached java program, all mouse activities are reported against one of the windows. Version-Release number of selected component (if applicable): java-1.7.0-openjdk-1.7.0.5-2.2.1.fc17.9.x86_64 How reproducible: Always Steps to Reproduce: 1. dconf write /org/gnome/libgnomekbd/keyboard/layouts "['us', 'gb', 'cm']" 2. select "English(UK)" keyboard 3. javac Gnome3Test.java 4. java Gnome3Test 5. Click and drag in 'A' window 6. Click and drag in 'B' window 7. Click and drag in 'A' window 8. Click and drag in 'B' window 9. Place mouse outside 'A' or 'B' window Actual results: A{ released moved pressed entered exited dragged } B{ entered exited } A{ released moved pressed entered exited dragged } B{ entered exited } Expected results: A{ released moved pressed entered exited dragged } B{ released moved pressed entered exited dragged } A{ released moved pressed entered exited dragged } B{ released moved pressed entered exited dragged } Additional info:
Created attachment 608182 [details] Small java program demonstarting focus problem
This seems to happen also with Gnome 2. What I see is that when I can move the A letter but the B letter never moves, is this what you also see?
yes (if not clicking in B window,then only B moves). It also happens over remote X, so my guess is that it stumbles when the key pressed is not in the first four columns of the X-keysym table.
I tried with an older build, and I can confirm this bug.
(In reply to comment #3) > yes (if not clicking in B window,then only B moves). > > It also happens over remote X, so my guess is that it stumbles when the key > pressed is not in the first four columns of the X-keysym table. Yes, quite likely, and is definitely desktop agnostic, since I can also reproduce in KDE.
We found the issue, I started a discussion on the awt-dev mailing list for the possible fix: http://mail.openjdk.java.net/pipermail/awt-dev/2012-November/003828.html If you need a fix before it ships into the official repositories (which may require some time), you can apply this patch: http://cr.openjdk.java.net/~neugens/853079/webrev.01/
Bug has been assigned the following upstream ID: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8000285 The fix has been pushed: http://hg.openjdk.java.net/jdk8/awt/jdk/rev/a78cb3c5d434
Sorry, wrong link, the bug ID is the following: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8005018
Thanks for all the work you all have done!
Sure, no problem! :)