<AWT Dev> Bug 853079: focus problems with openjdk 1.7.0 under gnome3 when selcted keyboard is not the first in keyboard list

[Artem Ananiev]

On 12/14/2012 5:30 PM, Mario Torre wrote:
> Il giorno ven, 14/12/2012 alle 16.35 +0400, Artem Ananiev ha scritto:
>> On 12/14/2012 4:10 PM, Roman Kennke wrote:
>>> Am Freitag, den 14.12.2012, 15:35 +0400 schrieb Artem Ananiev:
>>>> On 12/10/2012 11:57 PM, Mario Torre wrote:
>>>>> Hello Anthony,
>>>>>
>>>>> Sorry for the delay, but I've been pretty busy lately.
>>>>>
>>>>> Here is the new webrev with the corrections you requested:
>>>>>
>>>>> http://cr.openjdk.java.net/~neugens/853079/webrev.02/
>>>>
>>>> It looks good, just a single question:
>>>>
>>>> XBaseWindow.isFullRelease(): could you provide scenario, when this
>>>> method is not the same as "return buttonState == 0", please?
>>>
>>> I am not sure. IIRC, the button mask for button-release events in X11 is
>>> the mask *before* the release (i.e. for a full-release the
>>> currently-released button is still in there). That's why buttonState==0
>>> would not be correct in this case.
>>
>> Here is the code:
>>
>> 1036         if (button < 0 || button > buttonsNumber) {
>> 1037             return buttonState == 0;
>> 1038         } else {
>> 1039             return buttonState == XlibUtil.getButtonMask(button);
>> 1040         }
>>
>> If 0 <= button <= buttonsNumber, getButtonMask() will return 0, right?
>>
>> Thanks,
>
> No, it returns:
>
> return 1 << (7 + button);

Oh, Friday is not the best day to review fixes, I read this condition 
incorrectly twice (when looked to the code and when wrote an email)...

I don't have further comments/questions. The fix looks fine.

Thanks,

Artem

> Which is dependent of the value of button. For example, here is what I
> get with release:
>
> 1 - 256
> 2 - 512
> 1 - 256
> 4 - 2048
> 5 - 4096
>
> The first is button, the second is XlibUtil.getButtonMask(button);
>
> The code result in false (buttonState != XlibUtil.getButtonMask(button))
> when, for example, you press one button, keep it pressed, then press a
> second button, then release the second button, but keep the first button
> pressed.
>
> In that case, the release event for the second button doesn't match the
> button state:
>
> Press and release 1st:
> 1 - 256 - true
>
> Press and release 2nd (id=3):
> 3 - 1024 - true
>
> Press 1, press 2, release 2, release 1:
> 3 - 1024 - false
> 1 - 256 - true
>
> Cheers,
> Mario
>