Review request for 7124430: [macosx] LWCToolkit.grab() and LWCToolkit.ungrab() events are not implemented yet

[Anthony Petrov]

The fix looks fine. Thanks!

--
best regards,
Anthony

On 1/17/2012 7:41 PM, Artem Ananiev wrote:
> 
> On 1/17/2012 6:09 PM, Anton V. Tarasov wrote:
>> Here's the new version:
>>
>> http://cr.openjdk.java.net/~ant/7124430/webrev.3/
> 
> I don't have any further comments. Anthony?
> 
> Thanks,
> 
> Artem
> 
>> Thanks for the catch! I've added the case to the test as well.
>>
>> Thanks,
>> Anton.
>>
>>
>> On 17.01.2012 14:34, Artem Ananiev wrote:
>>> Hi, Anton,
>>>
>>> as we discussed offline, the fix looks fine with except the code in
>>> LWWindowPeer.notifyNCMouseDown(): it doesn't send ungrab events when
>>> user clicks on a title of a frame, that is not related (neither owner
>>> nor child) to the current grabbing window.
>>>
>>> So waiting for the next version of the fix.
>>>
>>> Thanks,
>>>
>>> Artem
>>>
>>> On 1/17/2012 11:30 AM, Anton V. Tarasov wrote:
>>>> Anthony was right (thanks for the persistence! :)
>>>>
>>>> Please review the new version, with -[NSWindow sendEvent:] overridden:
>>>>
>>>> http://cr.openjdk.java.net/~ant/7124430/webrev.2/
>>>>
>>>> Thanks,
>>>> Anton.
>>>>
>>>> On 1/13/12 9:17 PM, Anthony Petrov wrote:
>>>>> Overriding NSWindow -sendEvent: does allow one to catch mouse click
>>>>> events on the Mac. We use this technique in JavaFX for grab/ungrab
>>>>> functionality, and everything works smoothly.
>>>>>
>>>>> Anton was just trying to override NSApplication -sendEvent: which
>>>>> indeed won't help in that case (note the class name difference).
>>>>>
>>>>> -- 
>>>>> best regards,
>>>>> Anthony
>>>>>
>>>>> On 1/13/2012 9:34 PM, Mike Swingler wrote:
>>>>>> Don't worry about trying to catch the window title bar frame, you
>>>>>> really can't, and you shouldn't try. The titlebar/toolbar area is
>>>>>> specially marked so that the WindowServer will still allow the window
>>>>>> to be movable, even if the app is spinning and non-responsive. Due to
>>>>>> this fact, the app is infrequently notified when it's window is being
>>>>>> moved, because that app isn't even involved.
>>>>>>
>>>>>> Regards,
>>>>>> Mike Swingler
>>>>>> Apple Inc.
>>>>>>
>>>>>> On Jan 13, 2012, at 7:49 AM, Anton V. Tarasov wrote:
>>>>>>
>>>>>>> Hi Anthony,
>>>>>>>
>>>>>>> Thanks for the tip. Though I did look through some forum discussions
>>>>>>> where people asked question
>>>>>>> "is it possible to catch a mouse down event in a frame's titlebar?"
>>>>>>> The answer was "not a direct way".
>>>>>>> Commonly, a suggestion is to workaround it by means of
>>>>>>> -windowWillMove usage, but this looks quite
>>>>>>> odd to me...
>>>>>>>
>>>>>>> For instance:
>>>>>>>
>>>>>>> http://www.cocoabuilder.com/archive/cocoa/6725-catching-mousedown-in-an-nswindow-titlebar.html 
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> "PS: I have tried to overiding -sendEvent in the instance of a
>>>>>>> subclass
>>>>>>> of NSApplication object to see what events are going thru, mouseDown
>>>>>>> and mouseUp events in the titlebar don't go thru. It's seem that the
>>>>>>> windows are moved directly by the WindowServer, as you explained 
>>>>>>> me."
>>>>>>>
>>>>>>> Did you have an experience of trying it yourself?
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Anton.
>>>>>>>
>>>>>>> On 1/13/12 4:58 PM, Anthony Petrov wrote:
>>>>>>>> Hi Anton,
>>>>>>>>
>>>>>>>> You want to override NSWindow -sendEvent: in order to catch clicks
>>>>>>>> on the titlebar of windows on the Mac.
>>>>>>>>
>>>>>>>> -- 
>>>>>>>> best regards,
>>>>>>>> Anthony
>>>>>>>>
>>>>>>>> On 1/13/2012 6:36 PM, Anton V. Tarasov wrote:
>>>>>>>>> Hello,
>>>>>>>>>
>>>>>>>>> Please review a fix for 7124430.
>>>>>>>>>
>>>>>>>>> webrev: http://cr.openjdk.java.net/~ant/7124430/webrev.1/
>>>>>>>>>
>>>>>>>>> UngrabEvent dispatching is implemented according to the cases
>>>>>>>>> mentioned
>>>>>>>>> in the UngrabEvent class description, except for the case of
>>>>>>>>> clicking
>>>>>>>>> in an owner frame's title. The latter, as I found, can't be
>>>>>>>>> implemented
>>>>>>>>> on Mac OS X platform. Along with the implementation, a regression
>>>>>>>>> test
>>>>>>>>> is proposed.
>>>>>>>>>
>>>>>>>>> Thanks,
>>>>>>>>> Anton.
>>>>>>>>>
>>>>>>>>>
>>>>>>
>>>>
>>