Bug description says: "The test java/awt/Robot/HiDPIScreenCapture/ScreenCaptureTest.java fails on MacOSX, Windows. This failure was also observed on Kubuntu (KDE) “
Should there be sub-bugs created for Mac OS and also other platforms? Shura > On Mar 6, 2019, at 10:08 PM, Sergey Bylokhov <sergey.bylok...@oracle.com> > wrote: > > Hello. > Please review the fix for JDK 13. > > Bug: https://bugs.openjdk.java.net/browse/JDK-8215105 > Fix: http://cr.openjdk.java.net/~serb/8215105/webrev.00 > > Notes about the current implementation. > 1. The native system reads the pixels using the current "Display Profile". > 2. The robot API saves pixels into BufferedImage, which uses the sRGB color > profile. > 3. To convert pixels from native representation to sRGB we used > "kCGColorSpaceGenericRGB" mode. > > Note that during conversion we may lost a color precision, this is why we > always configured our test systems to use "Generic RGB Profiles" > But the logic above does not work on macOS 10.13.6. > > > I have tested macOS 10.12/13/14 and the only constant which works on all > systems is a kCGColorSpaceSRGB + Display Profile "sRGB IEC 61966-2.1". > So we can: > - Change nothing and skip "support" of 10.13.6 where > "kCGColorSpaceGenericRGB" + "Generic RGB Profiles" does not work. > - Apply the current fix, but then we need to change the requirement to the > test systems(macos 10.12->10.14) to use "sRGB IEC 61966-2.1" > > But since "kCGColorSpaceGenericRGB is available only for legacy reasons." it > make sense to use kCGColorSpaceSRGB anyway. > https://developer.apple.com/documentation/coregraphics/kcgcolorspacegenericrgb > > -- > Best regards, Sergey.
