<AWT Dev> [8] Review request for 8028995: Write regression test for JDK-8016356

[Sergey Bylokhov]

It is not just block of the thread, both threads are synchronized via 
wait/notifyAll() on AWTInvocationLock {} inside EventQueue.invokeAndWait/.

http://docs.oracle.com/javase/specs/jls/se7/html/jls-17.html
....
The |wait| methods of class |Object| (§17.2.1 
<http://docs.oracle.com/javase/specs/jls/se7/html/jls-17.html#jls-17.2.1>) 
have lock and unlock actions associated with them; their 
/happens-before/ relationships are defined by these associated actions.
....

  *

    An unlock on a monitor /happens-before/ every subsequent lock on
    that monitor.

  *

    A write to a |volatile| field (§8.3.1.4
    <http://docs.oracle.com/javase/specs/jls/se7/html/jls-8.html#jls-8.3.1.4>)
    /happens-before/ every subsequent read of that field.

....

Synchronization is required when we block main thread via Thread.sleep() 
and use ActionListeners() on EDT;


On 22.11.2013 20:20, Anthony Petrov wrote:
> All I know is as long as you access a variable from multiple threads, 
> you must use synchronization (locks or volatile). I've never heard 
> that simply blocking a thread is equivalent to using synchronization 
> primitives. Are there any documents that specify that?
>
> -- 
> best regards,
> Anthony
>
> On 11/22/2013 07:55 PM, Sergey Bylokhov wrote:
>> On 22.11.2013 18:51, Anthony Petrov wrote:
>>> Hi Oleg,
>>>
>>> The frLoc and frSize (and maybe other variables) should be declared
>>> volatile. Generally, if you access a variable from different thread,
>>> you have to synchronize the access using either the volatile modifier
>>> (the easiest way, esp. for tests), or a lock.
>> I always thought that in such cases two threads already synchronized,
>> because invokeAndWait is used, which block one thread and wait another.
>>>
>>> -- 
>>> best regards,
>>> Anthony
>>>
>>> On 11/22/2013 06:35 PM, Sergey Bylokhov wrote:
>>>> Thanks!
>>>> The fix looks good.
>>>>
>>>> On 22.11.2013 18:19, Oleg Pekhovskiy wrote:
>>>>> Hi Sergey,
>>>>>
>>>>> thanks you for the review,
>>>>> please take a look at the next version of fix:
>>>>> http://cr.openjdk.java.net/~bagiras/8028995.2/
>>>>>
>>>>> Changes:
>>>>> 1. OSInfo used.
>>>>> 2. Location and Dimensions of Frame are retrieved on EDT.
>>>>>
>>>>> Thanks,
>>>>> Oleg
>>>>>
>>>>> On 22.11.2013 17:16, Sergey Bylokhov wrote:
>>>>>> Hi, Oleg.
>>>>>> Usually we check the type of OS via sun.awt.OSInfo
>>>>>> Note that all these things should be on EDT as well:
>>>>>>
>>>>>>    // Retrieving the color of window expanded area
>>>>>>    86                 p = frame.getLocationOnScreen();
>>>>>>    87                 d = frame.getSize();
>>>>>>    88                 Insets insets = frame.getInsets();
>>>>>>
>>>>>>    91
>>>>>>    92                 frame.dispose();
>>>>>>
>>>>>>
>>>>>> On 22.11.2013 16:37, Oleg Pekhovskiy wrote:
>>>>>>> Hi all,
>>>>>>>
>>>>>>> please review the fix
>>>>>>> http://cr.openjdk.java.net/~bagiras/8028995.1/
>>>>>>> for
>>>>>>> https://bugs.openjdk.java.net/browse/JDK-8028995
>>>>>>>
>>>>>>> It's just a regression test for JDK-8016356.
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Oleg
>>>>>>
>>>>>>
>>>>
>>>>
>>
>>


-- 
Best regards, Sergey.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/awt-dev/attachments/20131122/41a98c2e/attachment-0001.html