Hi Adrian,
I have introduced various extra measures to ensure that the
DataVariance is set correctly, and in the case of osgshaders this
changes what was a STATIC into a DYNAMIC setting for the StateSet
holding the updating uniforms. I have also changed osgpick to set the
DataVariance of the text it updates.
Could you do a cvs update and test again?
Thanks for your patience.
Robert.
On 2/14/07, Adrian Egli <[EMAIL PROTECTED]> wrote:
i did following test, and crash !
****************** TEST with SingleThreaded
C:\Dokumente und Einstellungen\adegli>set OSG_THREADING=SingleThreaded
C:\Dokumente und Einstellungen\adegli>osgshaders
GraphicsWindowWin32::setPixelFormat() - Found a matching
pixel format but without the WGL_SWAP_METHOD_ARB specification for screen #0
Warning: deleting still referenced object 014AB0C0 of type 'class
osg::Referenced *'
the final reference count was 1, memory corruption possible.
******************DEADLOCK
2007/2/14, Robert Osfield < [EMAIL PROTECTED]>:
>
> Hi Adrian,
>
> On 2/14/07, Adrian Egli < [EMAIL PROTECTED]> wrote:
> > CPU:
> > Intel Dual Core : T7600 2.33 GHz
> > 2 GB RAM
> >
> > GPU:
> > ATI Mobility Radeon X1600 : Driver ATI Version 8.24
> >
> > Windows:
> > Windows XP : Version 2002, Service Pack 2, all securtiy updates
installed /
> > updated today !
> >
> > Screen:
> > 1
>
> Thanks, this is exactly the info I needed.
>
> The ATI driver probably doesn't swap the WGL_SWAP_METHOD_ARB control,
> this may or may not be an issue, perhaps Andreas can comment. If its
> not an issue then we can probably change the warning into a info.
>
> My guess is that the crash is probably unrelated to the above warning,
> the fact you you dual core and have one screen means that
> osgViewer::Viewer should be choosing to run using the
> DrawThreadPerContext threading model. This threading model double
> buffers the rendering backend, allow update and cull to run in
> parallel to draw.
>
> The parallism of DrawThreadPerContext also requires the scene graph to
> be set up so any changing Drawables and StateSet's are protected by
> setting DataVariance to STATIC. I will look at the example and do
> lots of testing on to see if I can recreate the crash, once I have I
> can test fixes for it.
>
> If you could override the threading model selection so that it is
> SingleThreaded (as per my previous email) then we'll be able to know
> whether my hunch above is correct.
>
> Robert.
> _______________________________________________
> osg-users mailing list
> [email protected]
> http://openscenegraph.net/mailman/listinfo/osg-users
> http://www.openscenegraph.org/
>
_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/
_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/
