<AWT Dev> [9] Review request for 8032078: CPlatformWindow.setWindowState throws RuntimeException, if windowState=ICONIFIED|MAXIMIZED_BOTH

[Petr Pchelko]

Ok. I see.

Thank you for the explanation. Now I agree that it's better to chose the easiest approach here as all Toolkits are different in handling this flag.

The fix looks good to me.

With best regards. Petr.

On 24.01.2014, at 17:43, Anton Litvinov <anton.litvinov at oracle.com> wrote:

> Hello Petr,
> 
> Thank you very much for review of this fix. I am not sure that the behavior of "ICONIFIED | MAXIMIZED_BOTH" should be completely the same as "ICONIFIED" on OS X, but I think that in this bitwise mask the most important bit is "ICONIFIED". And, if this compound state is valid, then as a result the window should be minimized.
> 
> This combination of flags is handled absolutely differently in Windows, Solaris/Linux implementation of JDK. Particularly on Windows the frame becomes minimized and will be always maximized, if the frame's icon on the taskbar is clicked. But "Frame.getExtendedState", "WFramePeer.getState()" will not always return "ICONIFIED | MAXIMIZED_BOTH", instead of this "ICONIFIED" will be returned, when before a call to "Frame.setExtendedState" method "Frame.state" was not "MAXIMIZED_BOTH".
> 
> The common feature of Windows and Solaris/Linux code is the fact that the implementation of "java.awt.peer.FramePeer.setState" does not throw an exception unlike OS X code.
> 
> Yes, this bug may be fixed in "Frame.isFrameStateSupported", but the change will require introduction of code checking, either it is executed on OS X platform or not, because this method should not be changed for other platforms, since they depend on it for a long time since 2004, when it was introduced by the fix for the bug JDK-4987087. From my opinion, making "Frame.isFrameStateSupported" return false for this compound state on OS X platform will make the code of this method not truly shared and will not let code calling "Frame.setExtendedState" with the state "ICONIFIED | MAXIMIZED_BOTH" to minimize the frame.
> 
> Thank you,
> Anton
> 
> On 1/21/2014 4:26 PM, Petr Pchelko wrote:
>> Hello, Anton.
>> 
>> Are we sure that the behavior of ICONIFIED | MAXIMIZED_BOTH should be the same as just ICONIFIED?
>> What does this combination of flags do on other platforms?
>> 
>> I would expect that if the frame is not maximized, this combination would iconify it, but when you deiconify the frame
>> should go to maximized state. However, I’m quite sure you can’t do it like this on Mac OS X, because we use the native
>> zoom mechanism for maximization and do not have enough control over it. So, in my opinion this should be fixed by
>> making Frame.isFrameStateSupported return false for this combination. What do you think?
>> 
>> With best regards. Petr.
>> 
>> 21 ÿíâ. 2014 ã., â 3:27 ïîñëå ïîëóäíÿ, Anton Litvinov<anton.litvinov at oracle.com>  íàïèñàë(à):
>> 
>>> Hello,
>>> 
>>> Could you please review the following fix for the bug.
>>> 
>>> Bug:https://bugs.openjdk.java.net/browse/JDK-8032078
>>> Webrev:http://cr.openjdk.java.net/~alitvinov/8032078/jdk9/webrev.00
>>> 
>>> The bug consists in undocumented throwing of "RuntimeException" from the method "Frame.setExtendedState" for the compound state "ICONIFIED | MAXIMIZED_BOTH" that is supported according to a result of "Frame.isFrameStateSupported" method call on OS X.
>>> 
>>> The solution adds handling of the mask "ICONIFIED | MAXIMIZED_BOTH" to "switch" block of the method "sun.lwawt.macosx.CPlatformWindow.setWindowState" which duplicates existing handling of the state "ICONIFIED" and prevents from throwing of the exception.
>>> 
>>> Thank you,
>>> Anton
>