Bug 1225

Summary: Nothing is drawn after AWT-Eventqueue thread is interrupted.
Product: [JogAmp] Jogl Reporter: ozgurkurt
Component: awtAssignee: Sven Gothel <sgothel>
Status: RESOLVED FIXED    
Severity: blocker    
Priority: ---    
Version: 2.3.2   
Hardware: pc_all   
OS: windows   
Type: DEFECT SCM Refs:
3235a4ae91e0202f9a36b8df4893f83be129dea2
Workaround: ---
Attachments: A java class with main method to demo explained behaviour.

Description ozgurkurt 2015-09-22 15:25:27 CEST
Created attachment 739 [details]
A java class with main method to demo explained behaviour.

We have a GLCanvas, working very well until some other AWT component calls Thread.currentThread.interrupt() (for whatever reason) during paint method execution. Afterwards GLCanvas doesn't get updated on the screen even though it's display(glad) method is executed completely.
Comment 1 ozgurkurt 2015-09-23 14:17:34 CEST
can be an intel chipset problem :(
Comment 2 Sven Gothel 2015-09-28 03:53:55 CEST
commit 3235a4ae91e0202f9a36b8df4893f83be129dea2:

Unit test to validate that Thread.interrupt() 
does not disturb continued rendering of AWT and GLCanvas.

This test is based on your attached test case.

Tests are displaying an ever color rotating rectangle in an AWT component alone and with an additional GearsES2 within a GLCanvas.

The AWT component is issuing an interrupt during paint on the AWT-EDT.

The reporter claims that an interrupt on the AWT-EDT shall not disturb neither AWT nor JOGL's GLCanvas and rendering shall continue. This seems to be true.

The test passes on GNU/Linux and Windows using JRE 1.8.0_60.

+++

Hence, I cannot reproduce this issue here.
Comment 3 ozgurkurt 2015-09-28 10:44:32 CEST
(In reply to Sven Gothel from comment #2)

Hi Sven, did you have a chance to try it on intel graphics platform?
Comment 4 Sven Gothel 2015-09-28 17:22:36 CEST
com.jogamp.opengl.test.junit.jogl.awt.TestBug1225EventQueueInterruptedAWT
hangs w/ java 1.7.0_45
Comment 5 Sven Gothel 2015-09-28 22:44:45 CEST
commit 45448a1bcdca7a5d7605de072fb39c8c31afc00f

Make unit test more robust,
notify if JRE 1.7.0 issue w/ interrupted AWT-EDT
    
The reporter claims that an interrupt on the AWT-EDT shall not disturb 
neither AWT nor JOGL's GLCanvas and rendering shall continue.
    
- This seems to be true for JRE 1.8.0_60

- This seems to be false for JRE 1.7.0_45. 
  This JRE's AWT-EDT even dies occasionally when interrupted.

On both JRE's occasionally an interrupt could be detected
on the AWT-EDT w/ or w/o GLCanvas being used.

On JRE 1.7.0_45 GNU/Linux, the AWT-EDT even dies 
while not using GLCanvas (1st test).

The behavior differs depending on the machine.

Hence this test case purpose is more to
document the different AWT-EDT behavior on the running
machine / JRE version.