Hello, Anthony.

> I suppose we might even want to put the GenerateTestImage.java file to the 
> test/java/awt/SplashScreen/ directory so that we could use it for other 
> splash screen tests in the future.
Sure. I'll do that prior to the push.

With best regards. Petr.

On 19.12.2013, at 13:53, Anthony Petrov <[email protected]> wrote:

> Looks great. Thank you!
> 
> I suppose we might even want to put the GenerateTestImage.java file to the 
> test/java/awt/SplashScreen/ directory so that we could use it for other 
> splash screen tests in the future.
> 
> --
> best regards,
> Anthony
> 
> On 12/19/2013 12:06 PM, Petr Pchelko wrote:
>> Hello, Anthony.
>> 
>> Thank you for the review.
>> 
>> Actually we have quite a lot of binary images in tests, but in this 
>> particular case the image could be generated on the fly.
>> Please review the updated version of the fix:
>> http://cr.openjdk.java.net/~pchelko/9/8024185/webrev.01/
>> 
>> Only test test is changed. We now generate an image for the splashscreen on 
>> the fly.
>> 
>> With best regards. Petr.
>> 
>> On 16.12.2013, at 23:04, Anthony Petrov <[email protected]> wrote:
>> 
>>> Hi Petr, Phil,
>>> 
>>> The fix looks fine to me. However, I'm not sure we want to add binary files 
>>> to the repo, no matter how good they are from "legal" perspective.
>>> 
>>> 
>>> Phil: what do you think about .png files in tests?
>>> 
>>> 
>>> --
>>> best regards,
>>> Anthony
>>> 
>>> On 12/16/2013 12:36 PM, Petr Pchelko wrote:
>>>> Hello, AWT Team.
>>>> 
>>>> Please review the fix for the issue:
>>>> https://bugs.openjdk.java.net/browse/JDK-8024185
>>>> The fix is available at:
>>>> http://cr.openjdk.java.net/~pchelko/9/8024185/webrev/
>>>> The fix also resolves the issue:
>>>> https://bugs.openjdk.java.net/browse/JDK-8009203
>>>> 
>>>> The problem:
>>>> When showing the splash screen the NSApplicationAWT runAWTLoopWithApp: was 
>>>> invoked from within the dispatch_async.
>>>> This is a blocking method, so it blocked the main dispatch queue which is 
>>>> used in Cocoa internally. So we've got different bugs.
>>>> The fix replaces the Grand Central Dispatch API with the JNFRunLoop 
>>>> performOnMainThreadWaiting which is used in other places in splashscreen.
>>>> 
>>>> The test verifies that the native FS support works after showing the 
>>>> splashscreen. Mac OS X specific APIs are accessed with reflection, so the 
>>>> test is compilable on other platforms.
>>>> The test.png is an image added to the test folder, it's not in the webrev 
>>>> as it does not support binary file diffs. I took the image from an 
>>>> existing 2d open test, so it should be fine from the legal point of view.
>>>> 
>>>> With best regards. Petr.
>>>> 
>> 

Reply via email to