<cite>@normen said:</cite>But I am not really interested in theorizing software failures.
LOL yeah, except that “more likely the new lwjgl than anything else” is actually theorizing a software failure.
Oh, and the mandatory downvoting is there, too. I hope you feel better for that.
Plus no answer to the technical issues. Which is what I’d be really interested in.
@toolforger said: LOL yeah, except that "more likely the new lwjgl than anything else" is actually theorizing a software failure. Oh, and the mandatory downvoting is there, too. I hope you feel better for that. Plus no answer to the technical issues. Which is what I'd be really interested in.
No, its empiric data leading me to this conclusion, not looking at awt behavior that seems strange to me. When awt uses gl surfaces then you might see multiple threads in awt. Combined with lwjgls native calls to gl, a combination of all three (lwjgl displays a canvas in awt or lwjgl renders in a backbuffer and displays it in awt) is likely to be the cause for freezes. Why else would the GUI freeze, its because its AWT rendering that. If it was a subcomponent the GUI would not freeze.
But I said all that already, I hope you get something out of your research. Thanks for confirming that these problems are effectively long solved.
Well, it was an AWT-EventQueue-2 and an AWT-EventQueue-4 thread. (Whatever happened to the normal AWT-EventQueue-0 I don’t know.)
One of them was holding the TreeLock, the other was waiting for it. Neither had anything LWJGL-related in their stack traces (but both were dealing with some HTML panel, dunno if it was the same HTML panel).
Root cause could still be LWJGL, of course. It’s just that I can’t see anything pointing in that direction.
Hi
<cite>@toolforger said:</cite> Well, it was an AWT-EventQueue-2 and an AWT-EventQueue-4 thread. (Whatever happened to the normal AWT-EventQueue-0 I don't know.) One of them was holding the TreeLock, the other was waiting for it. Neither had anything LWJGL-related in their stack traces (but both were dealing with some HTML panel, dunno if it was the same HTML panel).Root cause could still be LWJGL, of course. It’s just that I can’t see anything pointing in that direction.
Your symptom is exactly what this regression causes, several event queues, several event dispatching threads. Please switch to OpenJDK 1.7 update 25 (or later). I use it by default under Mageia Linux 2.
Thanks for the feedback, I noticed your info already.
Trouble is that some things wouldn’t work with 1.7. Combination of some Eclipse plugins (I think) not yet working in 1.7, trouble with some tools that choke as soon as multiple Javas are installed, Windows-vs.-Linux issues, and all that. Can’t describe the exact combination of events that led us to finally go for 1.6, but we decided to in the end.
But nevermind, I’m currently dropping the SDK anyway. I came for the shadernode editor, I went because I found it isn’t covering my use case (can’t inject a 3D texture for testing, and that’s what I wanted to do). That JDK update issue was just the last straw that sent me back.
(Plus I’m faster with Eclipse simply because I’m used to it.)