RFR: 8276058: Some swing test fails on specific CI macos system

[Alexander Zvegintsev]

On Wed, 27 Oct 2021 16:33:27 GMT, Phil Race <prr at openjdk.org> wrote:

>> As per JDK-8252813, some tests fails recurringly in CI macos system. This is an attempt to fix the swing tests.
>> It was seen from the logs that we have color mismatch in these tests.
>> 
>> For example, in PressedIcon test, we had following log
>> 
>> ----------System.out:(1/75)----------
>> color java.awt.Color[r=55,g=55,b=55] COLOR_1Xjava.awt.Color[r=255,g=0,b=0]
>> ----------System.err:(13/842)----------
>> java.lang.RuntimeException: Colors is different for scale=1!
>> 	at PressedIconTest.main(PressedIconTest.java:97)
>> 	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
>> 	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:76)
>> 	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:51)
>> 	at java.base/java.lang.reflect.Method.invoke(Method.java:569)
>> 	at com.sun.javatest.regtest.agent.MainWrapper$MainThread.run(MainWrapper.java:127)
>> 	at java.base/java.lang.Thread.run(Thread.java:833)
>> 
>> and JInternalFrame test, we had 
>> ----------System.err:(15/939)----------
>> FRAME_COLOR Red: 255; Green: 200; Blue: 0
>> Pixel color Red: 55; Green: 55; Blue: 55
>> java.lang.RuntimeException: Internal frame is not correctly dragged!
>> 	at bug8069348.main(bug8069348.java:96)
>> 	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
>> 	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:76)
>> 	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:51)
>> 	at java.base/java.lang.reflect.Method.invoke(Method.java:569)
>> 	at com.sun.javatest.regtest.agent.MainWrapper$MainThread.run(MainWrapper.java:127)
>> 	at java.base/java.lang.Thread.run(Thread.java:833)
>> 
>> where color is obtained as 55,55,55 instead of required color. The png image generated at failure also did not reveal anything abnormal apart from presence of mouse cursor around the same area where the getPixelColor location is pointing to. And since mouse cursor is also grayish black, it seems robot was wrongly picking up cursor pixels instead of background color.
>> Modified tests to use location.x-10, location.y-10 to obtain the pixel color. It does not hamper the test as the whole area around the location is coloured with same color in the test so getting pixel color from a bit wide of the centre will not affect the test.
>> Modified swing tests are passing in the affected system and also in other macos systems and platforms. Link in JBS.
>> 
>> The awt tests that are failing seems to have different root cause and will be handled separately.
>
> test/jdk/javax/swing/JButton/8151303/PressedIconTest.java line 88:
> 
>> 86:         robot.waitForIdle();
>> 87:         Color color = robot.getPixelColor(centerX-10, centerY-10);
>> 88: 
> 
> So the problematic pattern is a mouse move to the location from which you subsequently capture a pixel?
> I can see how on a compositing window manager that might be an issue as the cursor is part of the window
> whereas on old X11 it isn't part of the root window .. but why isn't this ALWAYS an issue on macOS ?
> 
> I wonder how many other tests have the same issue ?
> 
> And is an offset of 10 enough ? Its a bit arbitrary and cursors could be a larger shape or different orientation ..

Robot doesn't capture a cursor, we have a `Robot#createScreenCapture` spec which clearly says that:

> Creates an image containing pixels read from the screen.  
> **This image does not include the mouse cursor.**

Both `Robot#createScreenCapture` and `Robot#getPixelColor` are using the same `CRobot#nativeGetScreenPixels` method for getting pixel data.


<details>
  <summary>So the mouse cursor should not affect `Robot#getPixelColor`, this is easy to check on practice</summary>  


import javax.imageio.ImageIO;
import java.awt.AWTException;
import java.awt.Color;
import java.awt.Frame;
import java.awt.Rectangle;
import java.awt.Robot;
import java.awt.image.BufferedImage;
import java.io.File;
import java.io.IOException;

public class CursorTest {
    public static void main(String[] args) throws AWTException, IOException {
        Robot robot = new Robot();

        Frame frame = new Frame();
        frame.setBackground(Color.red);
        frame.setUndecorated(true);

        Rectangle rectangle = new Rectangle(100,100,50,50);
        frame.setBounds(rectangle);
        frame.setVisible(true);

        robot.waitForIdle();
        robot.delay(1000);
        robot.mouseMove(rectangle.x + 5, rectangle.y + 5);
        robot.waitForIdle();

        System.err.println("Moved");
        robot.waitForIdle();
        robot.delay(3000);

        System.err.println("Starting");

        BufferedImage screenCapture = robot.createScreenCapture(rectangle);
        ImageIO.write(screenCapture, "png", new File("/tmp/out.png"));

        for (int i = rectangle.x; i < rectangle.x + rectangle.width; i++) {
            for (int j = rectangle.y; j < rectangle.y + rectangle.height; j++) {
                Color pixelColor = robot.getPixelColor(i, j);
                if (!pixelColor.equals(Color.red)) {
                    System.err.println(pixelColor);
                    frame.dispose();
                    throw new RuntimeException("Unexpected pixel color " + pixelColor);
                }
            }
        }

        System.err.println("Finished");
        frame.dispose();
    }
}

  
</details>

-------------

PR: https://git.openjdk.java.net/jdk/pull/6140