<AWT Dev> [8] Review request for 8013563: Memory leak in JFrame on Linux

[Artem Ananiev]

This version looks fine.

Thanks,

Artem

On 10/2/2013 5:59 PM, Alexander Zvegintsev wrote:
> Hi Artem,
>
> 1. Oh, sure.
> 2. In initDeserializedWindow()  owner is not set yet, but we can call it
> from connectOwnedWindow().
>
> The new webrev is available here:
> http://cr.openjdk.java.net/~serb/alexz/8013563/webrev.01
>
> Thanks,
>
> Alexander.
>
> On 10/02/2013 02:12 PM, Artem Ananiev wrote:
>>
>> Tho short questions:
>>
>> 1. Is a new weak reference (victimWindowRef) really required? Can't we
>> re-use weakThis, which is victim.weakThis?
>>
>> 2. When a window is deserialized, does it have an owner? Shouldn't we
>> call WDR.updateOwner() in initDeserializedWindow()?
>>
>> Thanks,
>>
>> Artem
>>
>> On 10/1/2013 6:57 PM, Alexander Zvegintsev wrote:
>>> Hello,
>>>
>>> please review a fix for following issue:
>>> https://bugs.openjdk.java.net/browse/JDK-8013563
>>> webrev:
>>> http://cr.openjdk.java.net/~serb/alexz/8013563/webrev.00/
>>>
>>> This issue is a regression of 6758673 changes in java.awt.Window, where
>>> call to Disposer.addRecord
>>> was moved from init(gc) to ownedInit(). But we have constructors without
>>> a ownedInit() call, hence
>>> we never call removeFromWindowList() for Frame.
>>>
>