<AWT Dev> [9] Review request for 8075244 [macosx] The fix for JDK-8043869 should be reworked

[Alexander Scherbatiy]

On 3/19/2015 2:03 PM, Anton V. Tarasov wrote:
> Hi Alexander, Sergey,
>
> I can't say exactly if this is ok to init AWT from that point, but 
> even if it is, this seems to introduce new risks. At the same time, 
> AFAICS, the root of the original focus problem is unknown. Is that the 
> case? Should we try to understand it, instead? It is possible that 
> fixing the root of the issue would be less risky.

     It is new requirement that NSApplication should be initialized 
before NSScreen using on Mac OS X 10.10 Yosemite:
   -----------------------------
   NSScreen should be called the first time from the main thread 
(usually as part of -[NSApplication init]) to maintain consistency. (
   0 AppKit 0x00007fff84c1609a +[NSScreen _prepare] + 56
   1 libobjc.A.dylib 0x00007fff8f2e0bd0 _class_initialize + 655
   -----------------------------

    In our case it should be [NSApplicationAWT init].

    I can't dig deeper in it because problem is in between intersection 
of NSScreen initialization and [JRSAppKitAWT registrationOptions: 
options] calls on Mac OS X Yosemite. Both of them do not have public 
sources.

   Which risks do you mean? [NSApplicationAWT sharedApplication] is 
called from the SplashInitPlatform() method during splash screen 
creation so we just initialize the NSApplicationAWT a bit earlier.

   Thanks,
   Alexandr.

>
> Thanks,
> Anton.
>
> On 17.03.2015 14:27, Alexander Scherbatiy wrote:
>>
>> Hello,
>>
>> Could you review the fix:
>>   bug: https://bugs.openjdk.java.net/browse/JDK-8075244
>>   webrev: http://cr.openjdk.java.net/~alexsch/8075244/webrev.00
>>
>>   [NSApplicationAWT sharedApplication] call is added for the 
>> application initialization.
>>
>> Thanks,
>> Alexandr.
>>
>