<AWT Dev> Problem with modal Dialog

[Roman Kennke]

Hi Oleg,

> as far as I can see hideAndDisposeHandler() called either from hide()
> (if it is not called by dispose()) or
> from doDispose(), but after super.doDispose() is executed.

You are right. My original analysis was not correct. The problem is a 
little diferent, the application I have here calls hide() and dispose() 
one after the other, which then causes this situation, 
hideAndDisposeHandler() called first, followed quickly by 
Window.doDispose(). The attached test program demonstrates the problem. 
Not executing the DisposeAction immediately, but instead sending it over 
the EQ solves the problem by bringing the actions into correct order. 
This would be a good case for testing the brand new bugzilla, no? ;-)

/Roman

   So, from
> your description it is not clear
> how Window.doDispose() for the dialog can be executed after
> hideAndDisposeHandler() :(
> 
> Perhaps if you start your description from the very top of the call
> stack it would provide enough information
> to understand that.
> 
> With best regards, Oleg.
> 
> On Fri, Feb 13, 2009 at 7:54 PM, Roman Kennke <roman.kennke at aicas.com> wrote:
>> Hi there,
>>
>> I have a strange problem with modal dialogs here. I have a modal dialog,
>> which is never released, even though it is disposed. What happens is this:
>> in Dialog.hideAndDisposeHandler() the waking event is correctly posted to
>> the EQ. Then doDispose() continues in Window.doDispose(), where a
>> DisposeAction is created. Then it finds that it is on the EDT and
>> immediately executes this action. This in turn calls removeNotify(), which
>> then leads to all events that are related to the dialog to be removed from
>> the EQ, including the waking event. Thus, the Dialog is never released and
>> the thread that opened the dialog is stuck forever, since nobody will
>> notify() it. It can easily be solved in Window.doDispose(), just remove that
>> if (EventQueue.isDispatchThread()) check.
>>
>> Unfortunately I have no easy testcase to reproduce. I found this in
>> JamaicaVM w/ DirectFB/Caciocavallo. It is not unlikely that this only
>> happens because we have different thread scheduling than Hotspot, or I might
>> even think wrong and something is missing to me.
>>
>> Any ideas? Is this a bug? Am I thinking in the wrong direction?
>>
>> /Roman
>>
>> --
>> Dipl.-Inform. (FH) Roman Kennke, Software Engineer, http://kennke.org
>> aicas Allerton Interworks Computer Automated Systems GmbH
>> Haid-und-Neu-Straße 18 * D-76131 Karlsruhe * Germany
>> http://www.aicas.com   * Tel: +49-721-663 968-48
>> USt-Id: DE216375633, Handelsregister HRB 109481, AG Karlsruhe
>> Geschäftsführer: Dr. James J. Hunt
>>


-- 
Dipl.-Inform. (FH) Roman Kennke, Software Engineer, http://kennke.org
aicas Allerton Interworks Computer Automated Systems GmbH
Haid-und-Neu-Straße 18 * D-76131 Karlsruhe * Germany
http://www.aicas.com   * Tel: +49-721-663 968-48
USt-Id: DE216375633, Handelsregister HRB 109481, AG Karlsruhe
Geschäftsführer: Dr. James J. Hunt
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TestModalDialog.java
Type: text/x-java
Size: 784 bytes
Desc: not available
Url : http://mail.openjdk.java.net/pipermail/awt-dev/attachments/20090213/2e9eb1c5/TestModalDialog.java