<AWT Dev> [8] Review Request for JDK-8015477 Support single threaded AWT/FX mode.

[Anthony Petrov]

Hi Petr,

Thanks for all the changes. Looks much better, indeed. However, there 
still seems to be an issue with using inAWT in LWCToolkit.m and with the 
?: operator. Please see the quote below for details.

--
best regards,
Anthony

On 05/30/13 13:19, Petr Pchelko wrote:
> Hello, Thank you for the reviews.
>
> I have updated the fix, it is now available at:
> http://cr.openjdk.java.net/~pchelko/8015477/webrev.01/
> And the fx part:
> http://cr.openjdk.java.net/~pchelko/8015477/webrev.rt.01/
>
> I have updated comments and used AWTAccessor.
>
> Anthony wrote:
>> Regarding the FX part, I'm a bit uncertain if it's a good idea to perform so many actions with undefined semantics. I mean, the SwingFXUtils.runOnFxThread() either executes the runnable here and now (blocking the calling code), or just schedules the execution for later time. Sometimes this indeed makes sense. But in some cases this may be the wrong thing to do because e.g. we may want to block always. I believe we should mention this explicitly in the spec for this method, so that people know what they do when they use this method. Anyway, as long as this code is experimental and is off by default I'm OK with it.
> This method was added as an optimization so that we do not post a runnable if we are already on the needed thread. Currently it does not make any trouble as I have tested, but it is possible to remove it if we get any problems with it any time later. I have reduced it's visibility as it should only be used internally.
>
> With best regards. Petr.
>
> On May 29, 2013, at 6:54 PM, Anthony Petrov wrote:
>
>> Hi Petr,
>>
>> src/macosx/classes/sun/lwawt/macosx/LWCToolkit.java
>>>   88     /* If we operate in FX/AWT single threaded mode the nested loop should run in default mode */
>>>   89     private static final boolean inAWT;
>>
>> src/macosx/native/sun/awt/LWCToolkit.m
>>> 314         isRunning = [[NSRunLoop currentRunLoop] runMode:inAWT ? [JNFRunLoop javaRunLoopMode] : NSDefaultRunLoopMode
>>
>> If I read the comment above right, then we should swap the modes in ?:, no? Also, since ':' has a special meaning in ObjC, I suggest to enclose the ?: expression in parentheses for better readability.
>>
>> src/share/classes/sun/awt/FwDispatcher.java
>>>    2  * Copyright (c) 2011, 2013, Oracle and/or its affiliates. All rights reserved.
>>
>> Is it true that this code exists since 2011?
>>
>>> 41  * @since 8.0
>>
>> We use "1.8" to indicate that an API is available since JDK8.
>>
>> Regarding the FX part, I'm a bit uncertain if it's a good idea to perform so many actions with undefined semantics. I mean, the SwingFXUtils.runOnFxThread() either executes the runnable here and now (blocking the calling code), or just schedules the execution for later time. Sometimes this indeed makes sense. But in some cases this may be the wrong thing to do because e.g. we may want to block always. I believe we should mention this explicitly in the spec for this method, so that people know what they do when they use this method. Anyway, as long as this code is experimental and is off by default I'm OK with it.
>>
>> --
>> best regards,
>> Anthony
>>
>> On 05/28/2013 01:28 PM, Petr Pchelko wrote:
>>> Hello, AWT Team.
>>>
>>> Please review the fix for the issue:
>>> http://bugs.sun.com/view_bug.do?bug_id=8015477
>>> The fix is available at:
>>> http://cr.openjdk.java.net/~pchelko/8015477/webrev.00/
>>>
>>> The issue is just created, so it might not show up on bugs.sun.com
>>>
>>> This fix adds a private API to support a single threaded mode for AWT/FX interop.
>>> The main idea is to delegate some methods of an EventQueue to FX, so that AWT events were processed on FX thread.
>>>
>>> Here's the webrev for FX part of changes, for a usage sample:
>>> http://cr.openjdk.java.net/~pchelko/8015477/webrev.rt/
>>>
>>> With best regards. Petr.
>>>
>