<Swing Dev> <AWT Dev> [8] Review Request: JDK-8027157 [SwingNode] needs explicit expose for JWindow

[Anton V. Tarasov]

Hi Anthony,

On 10/23/13 8:37 PM, Anthony Petrov wrote:
> Hi Anton,
>
> Not sure if this matters much, but normally we don't use *.swing.* 
> classes directly from *.awt.* classes to avoid inter-dependencies. It 
> would look better if you could use an approach based on the 
> Class.getName()/Class.getSuperClass() methods here.
Right, I missed that. Thanks.

>
> Also, I don't quite understand why "in case of interop WM_PAINT is 
> never triggered". W/o understanding this, I'm not sure if the current 
> fix is correct. Please elaborate on the root cause of the bug.

Well, I agree we should understand why WM_PAINT is not generated. 
However, I'm afraid the investigation may take too much time and with no 
guarantee (that it can't be solved). This is a simple and, imo, harmless 
workaround which gets the issue solved in jfx8/jdk8. What about leaving 
the WM_PAINT issue open, but targeting it to the next release (when we 
have more time to investigate)?

Thanks,
Anton.


>
> -- 
> best regards,
> Anthony
>
> On 10/23/2013 10:15 PM, Anton V. Tarasov wrote:
>> Hello,
>>
>> Please, review the fix:
>>
>> jira: https://bugs.openjdk.java.net/browse/JDK-8027157
>> webrev: http://cr.openjdk.java.net/~ant/RT-32570/webrev.0
>>
>> This is to support SwingNode. On Windows, in the interop mode, when a
>> popup (JWindow) is shown it doesn't get WM_PAINT native message.
>> The message should trigger adding a dirty component to RepaintManager
>> that will eventually paint it.
>> Currently, a popup is shown blank. Please, see
>> https://javafx-jira.kenai.com/browse/RT-32570 for more details.
>>
>> The fix explicitly calls for WWindowPeer.handleExpose(..) on showing a
>> window owned by JLightweightFrame.
>>
>> Thanks,
>> Anton.