Hi Vincent, It does sound like the destruction of threads and wait for them to complete it not functioning correctly somehow. I'm afraid this isn't something one can debug remotely, could you please modify one of the OSG examples to behave in similar way to your app so that it recreates the bug in way that others can test it out first hand.
Robert. On Thu, Jun 25, 2009 at 2:40 PM, Vincent Bourdier<[email protected]> wrote: > Hi Robert, > > The 3d engine is called by an other application, frame by frame, so the > application call a personal method to close a View (we have one or more > Viewer). There is a map to get the viewer pointer associated to the Opengl > Context it has, which allow us modifying a viewer or an other one. So when > we decide to close a viewer, I set the viewer as done, verify the thread are > stopped, and I remove it from our viewer map. This implicitly call the > viewer destructor, then causes the crash. > > The map is a ref_ptr map and the viewer before being deleted only has a ref > count of 1, so I can erase the map element, which call the viewer destructor > causing the crash. > > Sorry for the lack of explanation, I hope this is better. > > Thanks. > > Regards, > Vincent. > > 2009/6/25 Robert Osfield <[email protected]> >> >> Hi Vincent, >> >> What exactly do you mean by "when I use the other way and call the >> destructor (I clear the viewer map I get) there is a crash in", as >> these seems odd, you never call the destructor, all you can do is call >> delete, but in the case of the Viewer class you should never be >> calling delete yourself, rather leave it to C++ to clean up in the >> case of creating the Viewer in local scope on the stack, or using a >> ref_ptr<> when you manage the Viewer on the heap. >> >> Robert. >> >> On Thu, Jun 25, 2009 at 12:30 PM, Vincent >> Bourdier<[email protected]> wrote: >> > Hi all, >> > >> > I have a crash on a composite viewer destructor, and I do not see >> > anything >> > strange or wrong in my code. >> > >> > My viewer is managed as an external lib, and the frame() call are done >> > by a >> > method, like the close() call to finish a viewer. >> > >> > So, if I use the viewer->run() there is no problem, but when I use the >> > other >> > way and call the destructor (I clear the viewer map I get) there is a >> > crash >> > in >> > >> > GLObjects.cpp : >> > >> > void osg::flushAllDeletedGLObjects(unsigned int contextID) >> > { >> > double currentTime = DBL_MAX; >> > double availableTime = DBL_MAX; >> > >> > >> > osg::BufferObject::flushDeletedBufferObjects(contextID,currentTime,availableTime); >> > osg::Drawable::flushAllDeletedDisplayLists(contextID); >> > >> > >> > osg::FragmentProgram::flushDeletedFragmentProgramObjects(contextID,currentTime,availableTime); >> > >> > >> > osg::FrameBufferObject::flushDeletedFrameBufferObjects(contextID,currentTime,availableTime); >> > >> > >> > //osg::Program::flushDeletedGlPrograms(contextID,currentTime,availableTime); >> > //CRASH Here >> > >> > >> > osg::RenderBuffer::flushDeletedRenderBuffers(contextID,currentTime,availableTime); >> > >> > >> > //osg::Shader::flushDeletedGlShaders(contextID,currentTime,availableTime); >> > //CRASH Here >> > osg::Texture::flushAllDeletedTextureObjects(contextID); >> > >> > >> > osg::VertexProgram::flushDeletedVertexProgramObjects(contextID,currentTime,availableTime); >> > >> > >> > osg::OcclusionQueryNode::flushDeletedQueryObjects(contextID,currentTime,availableTime); >> > >> > } >> > >> > in DEBUG_INFO I get : >> > ... >> > Releasing GL objects for Camera=05EAE0C8 _state=05EB7050 >> > Closing still viable window 0 _state->getContextID()=0 >> > Doing Flush >> > GraphicsContext::setWindowingSystemInterface() 00000000 05DE9198 >> > >> > Any idea ? suggestion ? >> > >> > Thanks a lot. >> > >> > Regards, >> > Vincent. >> > >> > _______________________________________________ >> > osg-users mailing list >> > [email protected] >> > >> > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org >> > >> > >> _______________________________________________ >> osg-users mailing list >> [email protected] >> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > > _______________________________________________ > osg-users mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
