<AWT Dev> <Swing Dev> Weird focus problem with JTextField

pavel porvatov pavel.porvatov at oracle.com
Mon Oct 8 09:49:18 PDT 2012

Hi Clemens,

Ok. I checked it on Ubuntu 12.04 and the problem does exist. I also 
observed that the problem is not reproducible 1.8.0-ea-b56. The old bug 
is frozen, so I filed a new one (#8000528) and added the following comment:

Pavel Porvatov added a comment - 2012-10-08 20:43
This is a Linux specific problem and reproducible only in browsers. 
AppletViewer works as expected. The problem is reproducible in old 
builds but not reproducible with the following configuration:
Java Plug-in
Using JRE version 1.8.0-ea-b56 Java HotSpot(TM) Client VM

After some time the bug will be visible on bugs.sun.com and you will be 
able observe bug state

Regards, Pavel

> Hi Pavel,
> As mentioned, the issue only exists on Linux.
> Thanks, Clemens
> 2012/10/8 pavel porvatov<pavel.porvatov at oracle.com>:
>> Hi Clemens,
>> I can't reproduce the problem with your test case on 7u7 as well. The
>> following info if from my java console:
>> Java Plug-in 1.6.0_35
>> Using JRE version 1.7.0_07-b32 Java HotSpot(TM) Client VM
>> I used Chrome. Firefox 10.0.7 works as well. Tested OS is: Microsoft Windows
>> [Version 6.1.7601]
>> Regards, Pavel.
>>> Hi Pavel,
>>> Thanks for taking a look at the testcase.
>>> The bug triggering the testcase seems to have been fixed between
>>> jdk8b55 and jdk8b56, although I am still able to reproduce the issue
>>> with jdk7u7. I tried to add a comment to the report, however it seems
>>> that functionality has been disabled.
>>> For me the issue showed up with Firefox, Chrome and Opera on Fedora-17
>>> 64-bit, regardless using the oracle plugin or icedtea-web. Appletview
>>> works as expected, however.
>>> Unfourtunatly, the real application which made me create this
>>> self-contained testcase still runs into the problem ...
>>> Thanks again, Clemens
>>>> On 14.09.2012 3:36, Clemens Eisserer wrote:
>>>>> Hi Pavel,
>>>>> I was able to distill the problem to small, self-contained testcase,
>>>>> which is available at:
>>>>> I've also created a short video, demonstrating the issue:
>>>>> http://youtu.be/r1JDj5BuBOM
>>>>> The JTextField opens up a JPopupMenu on<ENTER>, which can be closed
>>>>> by clicking the JButton it contains.
>>>>> However, after the popup is hidden, no editing is possible unless
>>>>> focus is toggled once between the browser and another window.
>>>>> Unfourtunatly the testcase only triggers the problem on Linux (tested
>>>>> with jdk8+oracle plugin as well as openjdk7+icedtea-web both with
>>>>> recent versions Firefox and Chrome), However, I've seen this issue
>>>>> happen on Windows too when running the application I talked in the
>>>>> first email.
>>>>> It would be great if you could have a look.
>>>>> Thank you in advance, Clemens
>>>>> 2011/7/30 Pavel Porvatov<pavel.porvatov at oracle.com>:
>>>>>> Hi Clemens,
>>>>>>> Hi,
>>>>>>> I've developed a database applet where Swing widgets are connected to
>>>>>>> RowSets using the SwingSet library (an old sourceforge project).
>>>>>>> All queries are executed on the EDT, except for some background tasks
>>>>>>> which sync with the UI using SwingUtilities.invokeLater().
>>>>>>> When running the application over high-latency connections (EDT is
>>>>>>> busy
>>>>>>> for a few seconds), sometimes after a pending operation is finished I
>>>>>>> can't
>>>>>>> edit any JTextFields. The mouse-cursor changes when I move it over the
>>>>>>> JTextField, but when I click into the JTextField the carret simply
>>>>>>> does
>>>>>>> not
>>>>>>> appear - everything else (JMenu, JButtons, ...) works as expected - so
>>>>>>> I
>>>>>>> guess event delevery is still working as expected.
>>>>>>> To make JTextField-editing work again, I have to transfer focus to
>>>>>>> another
>>>>>>> native window and back to the browser-window. A heavyweight JMenu
>>>>>>> makes
>>>>>>> it
>>>>>>> work again too, whereas a lightweight one doesn't help.
>>>>>>> Even worse, the problem does not manifest itself with low-latency
>>>>>>> connections or when running as application.
>>>>>>> I am completly puzzled :/
>>>>>>> Any idea what could be the problem, or where / what for I should start
>>>>>>> looking?
>>>>>> First of all it's not a good idea to keep the EDT thread busy for a
>>>>>> long
>>>>>> time and I think the best way is to rewrite application.  Could you
>>>>>> please
>>>>>> make a small separated application that emulates EDT blocking and shows
>>>>>> the
>>>>>> problem?
>>>>>> Regards, Pavel

More information about the awt-dev mailing list