On 26/03/2018 17:50, Philip Race wrote:
I presume the headful keyword means thie can't be run in our nightly
tests ..and
isn't supposed to be run in headless mode, so I am a bit puzzled by the
change.
I also don't understand why it would be "random".
An updated test just emulates the problem, it run the first instance
using "-Djava.awt.headless" and then unconditionally tries to change the
system display mode, this causes an exception in the headless instance.
This bug cause the random failures of our headless tests
Here I have referenced any other tests in nightly, which fails if we get
such notifications from the OS.
* @key headful
* @summary Tests that no exception is thrown when display mode is changed
* externally
* @compile UninitializedDisplayModeChangeTest.java DisplayModeChanger.java
* @run main/othervm UninitializedDisplayModeChangeTest
+ * @run main/othervm -Djava.awt.headless=true
UninitializedDisplayModeChangeTest
*/
-phil.
On 3/20/18, 11:48 AM, Sergey Bylokhov wrote:
Hello.
Please review fix for jdk11.
Bug: https://bugs.openjdk.java.net/browse/JDK-8198335
Webrev can be found at:
http://cr.openjdk.java.net/~serb/8198335/webrev.01
This bug cause the random failures of our headless tests. The reason
is that we try to notify the
GraphicsEnvironment(HeadlessGraphicsEnvironment), which does not
implement DisplayChangedListener, about display change.
It is possible to skip this notification for headlesGE only, but I
think that it is better to make instanceof before cast to
DisplayChangedListener, usually we use this pattern in other places.
PS. Note that WToolkit.paletteChanged and WToolkit.displayChanged use
different threads for notifications, I have created a bug to
investigate this: https://bugs.openjdk.java.net/browse/JDK-8199881
--
Best regards, Sergey.
