test_backspaceAndDelete StyledText and Text fail on I-build machine

[jonahgraham]

In #2549 I updated tests and they pass on GitHub actions, all platforms. But failed on Windows I-build https://download.eclipse.org/eclipse/downloads/drops4/I20250930-1800/testResults.php - so more work (and more for me to learn about different platforms)

Failure looks like:

Multiple Failures (2 failures) java.lang.AssertionError: <no message> java.lang.RuntimeException: AwtScreenshot VM finished with exit code 1: Exception in thread "main" java.awt.AWTError: no screen devices at java.desktop/sun.awt.Win32GraphicsEnvironment.getDefaultScreenDevice(Win32GraphicsEnvironment.java:101) at java.desktop/java.awt.Robot.<init>(Robot.java:135) at org.eclipse.test.AwtScreenshot.main(AwtScreenshot.java:43)

org.opentest4j.MultipleFailuresError: Multiple Failures (2 failures)
java.lang.AssertionError: <no message>
java.lang.RuntimeException: AwtScreenshot VM finished with exit code 1:
Exception in thread "main" java.awt.AWTError: no screen devices
at java.desktop/sun.awt.Win32GraphicsEnvironment.getDefaultScreenDevice(Win32GraphicsEnvironment.java:101)
at java.desktop/java.awt.Robot.<init>(Robot.java:135)
at org.eclipse.test.AwtScreenshot.main(AwtScreenshot.java:43)

Suppressed: java.lang.AssertionError
at org.junit.Assert.fail(Assert.java:87)
at org.junit.Assert.assertTrue(Assert.java:42)
at org.junit.Assert.assertTrue(Assert.java:53)
at org.eclipse.swt.tests.junit.Test_org_eclipse_swt_custom_StyledText.test_backspaceAndDelete(Test_org_eclipse_swt_custom_StyledText.java:5916)
at java.base/java.lang.reflect.Method.invoke(Method.java:580)
Suppressed: java.lang.RuntimeException: AwtScreenshot VM finished with exit code 1:
Exception in thread "main" java.awt.AWTError: no screen devices
at java.desktop/sun.awt.Win32GraphicsEnvironment.getDefaultScreenDevice(Win32GraphicsEnvironment.java:101)
at java.desktop/java.awt.Robot.<init>(Robot.java:135)
at org.eclipse.test.AwtScreenshot.main(AwtScreenshot.java:43)

at org.eclipse.test.AwtScreenshot.dumpAwtScreenshot(AwtScreenshot.java:113)
at org.eclipse.test.Screenshots.takeScreenshot(Screenshots.java:102)
at org.eclipse.test.Screenshots$ScreenshotOnFailure.failed(Screenshots.java:40)

Originally posted by @jonahgraham in #2549 (comment)

[jonahgraham]

Best guess is this fails because of UIPI issues - in the past the Windows machine needed a VNC client connected to it to allow test to work, see Bug 407862, but at some point all the other tests that do Display.post were disabled too so there is no visibility on it and we just haven't been testing Display.post consistently.

Therefore I am going to try to deal with any bitrotting in this area, and if I can't I will disable this test (at least on Jenkins/I-builds)

[jonahgraham]

Naturally this is connected to the other KeyEvent issues on GitHub actions 😢 #2516

[jonahgraham]

With #2576 merged test_postLorg_eclipse_swt_widgets_Event (in Test_org_eclipse_swt_widgets_Display) may fail on the I-build. If so, and I can't resolve the UIPI issues, then I will disable the tests on Jenkins.

cc: @iloveeclipse FYI because I know you often do triaging of new test failures and I wanted to make sure you knew what I am doing here.

[jonahgraham]

TL;DR

A combination of further investigation on how UIPI and azure interact and celebrating my win so far here I have decided to proceed and just disable tests on non-GitHub actions.

---

Basically I think we would need to get the EF IT to have a permanent RDP connection to the Windows Azure dedicated machine so UIPI would allow SendEvent to work. There used to be a VNC connection to the Windows machine for this purpose in the distant past (see Bug 407862) but this seems to have long ago been abandoned, so when the transition was made to azure (when??) the functionality didn't carry over. By this point the tests requiring it had all been disabled, so no one took much notice. (Much of this paragraph is educated guessing and conjecture)

For the "celebrating my win" part, I started down this path of fixing the false positive test result in test_backspaceAndDelete as an exercise in learning SWT. These tests, and now the related test_postLorg_eclipse_swt_widgets_Event pass again, albeit only on GHA. This is the win and progress I am going to celebrate and close off this issue here with the merging of #2578

#2578 was not merged in time for the build (I20251001-1800) currently running, but if all goes to plan these tests will be skipped from the next I-build.