<AWT Dev> [9] Review request for 8173853: IllegalArgumentException in java.awt.image.ReplicateScaleFilter

[Phil Race]

I am fine with the proposed fix as it fixes the exception and should be 
safe to backport.

-phil.

On 3/3/2017 4:09 AM, dmitry markov wrote:
> Hi Sergey,
>
> In my opinion if zero size is returned as a result of 
> getBestCursorsize() invocation, the cursor should have initial size 
> defined by the parameter of its constructor. I am sorry, but it will 
> look very odd especially for end user, if we silently change an 
> original cursor's size to zero even in the environment where custom 
> cursor is not supported. Also I guess the usage of the cursor with 
> zero size may cause some problems like exceptions, etc. in other parts 
> code.
>
> Thanks,
> Dmitry
> On 02/03/2017 22:08, Sergey Bylokhov wrote:
>> Hi, Dmitry.
>> I have only the one question: the size 0,0 can be returned if the 
>> custom cursors are not supported. I wonder should our CustomCursor 
>> have an initial size, of it should have zero size?
>>
>>> Hello,
>>>
>>> Could you review a fix for jdk9, please?
>>>
>>>     bug: https://bugs.openjdk.java.net/browse/JDK-8173853
>>>     webrev: http://cr.openjdk.java.net/~dmarkov/8173853/webrev.00/
>>>
>>> Problem description:
>>> Sometimes XQueryBestCursor may return zero-dimension to the code 
>>> inside XCustomCursor.getBestCursorSize(). As a result the subsequent 
>>> invocation of CustomCursor constructor will fail with 
>>> IllegalArgumentException during scaling of the image.
>>>
>>> Fix:
>>> It is necessary to avoid scaling if getBestCursorSize() returns 
>>> zero-dimension.
>>>
>>> Thanks,
>>> Dmitry
>>>
>