Re: [gl4java-usergroup] GLCanvas Dispose on NT

[Kenneth B. R. <kbr...@al...>]

Thanks, this looks like a good fix. I tested it and it clears up
the warning that Arkanae was printing upon exit as well. It's
been checked in to the source tree on SourceForge.


> After doing some debugging I made the following change to my source.
> 
> In file : OpenGL_Win32_jawt.c
> In the method : gljDestroyNative
> 
> We were trying to delete a context with : disp__wglDeleteContext(gc). An
> error was occurring on my system (WinNT), I think, because the context
> being deleted (gc) belonged to a thread other than the current thread and
> thus threw an EXCEPTION_ACCESS_VIOLATION.
> Previous to this (a few lines up from the above) we were calling  :
> disp__wglMakeCurrent( NULL, NULL ) , which detaches the rendering context
> from the current thread and makes it non-current, in preparation for
> deletion. This call, however, was failing on my system. I am assuming it
> was failing, again, because the current thread did not hold the rendering
> context.
> 
> I am not exactly sure why the thread did not own the context. I must admit,
> my experience with GL Windows programming is very limited. But here is what
> I did to fix my error :
> 
> I changed disp__wglMakeCurrent( NULL, NULL ) to disp__wglMakeCurrent(
> thisWin, gc ) so that the current thread now owns the context. Then, when
> it gets around to deleting it a few lines later, no error occurs. According
> to MSDN the disp__wglDeleteContext(gc) call will actually change the
> rendering context to being non-current before deleting so really the
> original disp__wglMakeCurrent( NULL, NULL ) was redundant. The only
> important thing is to make sure that the current thread owns the context
> before deleting it.
> 
> Is it a hack? I don't know. Thoughts? It has stopped my error though.