The following version of the test passes for OS X + retina (as well as for Windows and Linux):
http://cr.openjdk.java.net/~avstepan/8150258/webrev.01/

not sure if "-Dsun.java2d.uiScale" option was intended for use with OS X (probably an enhancement bug should be created).

So for OS X now we have only single test run in fact.

Thanks,
Alexander

On 2/29/2016 5:12 PM, Alexander Stepanov wrote:
Hello Sergey,

When running the following trivial test case (having Color set to "Generic RGB Profile")

import java.awt.*;

public class ColorTest extends Frame {

    private void UI() {
        setSize(100, 100);
        setBackground(Color.BLUE);
        setVisible(true);
    }

    public ColorTest() throws Exception {
        EventQueue.invokeAndWait(this::UI);
    }

    public static void main(String[] args) throws Exception {
        ColorTest test = new ColorTest();
        Thread.sleep(1000);

        Point pt = test.getLocationOnScreen();
        System.out.println("color = " +
            (new Robot()).getPixelColor(pt.x + 50, pt.y + 50));
        test.dispose();
    }
}

the expected output appears:

color = java.awt.Color[r=0,g=0,b=255]

At the same time the Color for the Icon is "[r=0,g=0,b=248]"

Could you please let me know if the test still fails for you when simply set "expected = C2X" on the ln 116 and comment out the 1st "@run"?

Not sure if it should be considered as a bug, or some additional settings are required, or the color check should be simply softened (some tolerance used).

Thanks,
Alexander


On 2/26/2016 6:13 PM, Alexander Stepanov wrote:
It seemingly does not help (but maybe I'm doing something wrong).

On 2/26/2016 5:45 PM, Sergey Bylokhov wrote:
On 26.02.16 17:27, Alexander Stepanov wrote:
It seems that for osx + retina

1. -Dsun.java2d.uiScale option is ignored for multires. image (in
contrast for HiDPI + Win. 8, for example). Should it be considered as a
bug? At a 1st glance, yes.
2. even in case of correct color robot's getPixelColor() returns for
retina a bit shifted color ((0, 0, 251) for blue instead of (0, 0, 255))
- not sure if that could be considered as an issue, or the color check
should be simply softened.

Can you try to set Generic color RGB profile for the Display?


Thanks,
Alexander

P.S. I performed test runs on HiDPI Windows 8 and regular Ubuntu 14.04
Linux initially (which is incomplete)



On 2/26/2016 4:41 PM, Alexander Stepanov wrote:
Hello Sergey,

No, as well as for 8150724. I have to re-check, thank you.

Regards,
Alexander

On 2/26/2016 4:36 PM, Sergey Bylokhov wrote:
Hi, Alexander.
The test failed on my OSX 10.11 + retina. Is it expected?

On 25.02.16 17:51, Alexander Stepanov wrote:
Sorry, just a reminder.

Thanks,
Alexander

On 2/19/2016 7:17 PM, Alexander Stepanov wrote:
Hello,

Could you please review the following fix
http://cr.openjdk.java.net/~avstepan/8150258/webrev.00/
for
https://bugs.openjdk.java.net/browse/JDK-8150258

Just a single test added.

Thanks,
Alexander










Reply via email to