Hello Alexander,
Alexandr Scherbatiy wrote > Could you review the updated fix: > http://cr.openjdk.java.net/~alexsch/8043869/webrev.03/ I think this change caused a bug in the SplashScreen#getBounds method (SplashScreen.java file) when a high-dpi splash screen is available and scale > 1. if (0 < scale && scale != 1) { bounds.setSize((int) (bounds.getWidth() / scale), (int) (bounds.getWidth() / scale)); } I think that in the above code, the second getWidth() should be replaced by a call to getHeight(), resulting in: if (0 < scale && scale != 1) { bounds.setSize((int) (bounds.getWidth() / scale), (int) (bounds.getHeight() / scale)); } If not, the getBounds method will always return square bounds, even when the splash screen is a rectangle where width != height. This problem is not detected by the MultiResolutionSplashTest.java. I think the test would detect this problem if the MultiResolutionSplashTest#testSplash is updated to include the following check: if ( splashScreen.getBounds().getWidth() != IMAGE_WIDTH ){ throw new RuntimeException( "Splash screen width is incorrect" ); } if ( splashScreen.getBounds().getHeight() != IMAGE_HEIGHT ){ throw new RuntimeException( "Splash screen height is incorrect" ); } Robin -- View this message in context: http://openjdk.5641.n7.nabble.com/AWT-Dev-9-Review-request-for-8043869-macosx-java-splash-does-not-honor-2x-hi-dpi-notation-for-retinat-tp192308p272361.html Sent from the OpenJDK AWT Development mailing list archive at Nabble.com.
