class TGLViewer: public TVirtualViewer3D, public TGLViewerBase, public TQObject

 Construct the viewer object, with following arguments:
    'pad' - external pad viewer is bound to
    'x', 'y' - initial top left position
    'width', 'height' - initial width/height

gl-embedded viewer's ctor
 Construct the viewer object, with following arguments:
    'pad' - external pad viewer is bound to
    'x', 'y' - initial top left position
    'width', 'height' - initial width/height

 Common initialization.

 Destroy viewer object.

 Entry point for updating viewer contents via VirtualViewer3D
 interface.
 We search and forward the request to appropriate TGLScenePad.
 If it is not found we create a new TGLScenePad so this can
 potentially also be used for registration of new pads.

 Force update of pad-scenes. Eventually this could be generalized
 to all scene-types via a virtual function in TGLSceneBase.

 Resets position/rotation of current camera to default values.

 Setup cameras for current bounding box.

 Perform post scene-build setup.

 Initialise GL state.

 Post request for redraw of viewer at level of detail 'LOD'
 Request is directed via cross thread gVirtualGL object.

 Setup clip-object. Protected virtual method.

 Initialize objects that influence rendering.
 Called before every render.

 Normal rendering, used by mono and stereo rendering.

 Restore state set in PreRender().
 Called after every render.

 Draw out the viewer.

 Draw out in monoscopic mode.

 Draw out in stereoscopic mode.

 Save current image using the default file name which can be set
 via SetPictureFileName() and defaults to "viewer.jpg".
 Really useful for the files ending with 'gif+'.

 Save current image in various formats (gif, gif+, jpg, png, eps, pdf).
 'gif+' will append image to an existng file (animated gif).
 'eps' and 'pdf' do not fully support transparency and texturing.
 The viewer window most be fully contained within the desktop but
 can be covered by other windows.
 Returns false if something obvious goes wrong, true otherwise.

 The mage is saved using a frame-buffer object if the GL implementation
 claims to support it -- this claim is not always true, especially when
 running over ssh with drastically different GL implementations on the
 client and server sides. Set this in .rootrc to enforce creation of
 pictures using the back-buffer:
   OpenGL.SavePicturesViaFBO: off

 Save current image in various formats (gif, gif+, jpg, png).
 'gif+' will append image to an existng file (animated gif).
 Back-Buffer is used for capturing of the image.
 The viewer window most be fully contained within the desktop but
 can be covered by other windows.
 Returns false if something obvious goes wrong, true otherwise.

 Save current image in various formats (gif, gif+, jpg, png).
 'gif+' will append image to an existng file (animated gif).
 Frame-Buffer-Object is used for capturing of the image - OpenGL
 1.5 is required.
 The viewer window does not have to be visible at all.
 Returns false if something obvious goes wrong, true otherwise.

 pixel_object_scale is used to scale (as much as possible) the
 objects whose representation size is pixel based (point-sizes,
 line-widths, bitmap/pixmap font-sizes).
 If set to 0 (default) no scaling is applied.

 Returns current image.
 Back-Buffer is used for capturing of the image.
 The viewer window most be fully contained within the desktop but
 can be covered by other windows.

 Returns current image.
 Frame-Buffer-Object is used for capturing of the image - OpenGL
 1.5 is required.
 The viewer window does not have to be visible at all.

 pixel_object_scale is used to scale (as much as possible) the
 objects whose representation size is pixel based (point-sizes,
 line-widths, bitmap/pixmap font-sizes).
 If set to 0 (default) no scaling is applied.

 Save picture with given width (height scaled proportinally).
 If pixel_object_scale is true (default), the corresponding
 scaling gets calculated from the current window size.

 Save picture with given height (width scaled proportinally).
 If pixel_object_scale is true (default), the corresponding
 scaling gets calculated from the current window size.

 Save picture with given scale to current window size.
 If pixel_object_scale is true (default), the same scaling is
 used.

 Draw reference marker and coordinate axes.

 If in debug mode draw camera aids and overall bounding box.

 Perform GL work which must be done before each draw.

 Perform GL work which must be done after each draw.

 Draw a rectangle (background color and given alpha) across the
 whole viewport.

 Make GL context current

 Swap GL buffers

 Post request for selection render pass viewer, picking objects
 around the window point (x,y).

 Perform GL selection, picking objects overlapping WINDOW
 area described by 'rect'. Return kTRUE if selection should be
 changed, kFALSE otherwise.
 Select lock should already been taken in other thread in
 TGLViewer::ReqSelect().

 Request secondary select.

 Secondary selection.

 Process result from last selection (in fSelRec) and
 extract a new current selection from it.
 Here we only use physical shape.

 Post request for secondary selection rendering of selected object
 around the window point (x,y).

 Perform GL selection, picking overlay objects only.
 Return TRUE if the selected overlay-element has changed.

 Animate fading from curernt value to fade over given time (sec)
 and number of steps.

 Use the dark color-set.

 Use the light color-set.

 Swtich between dark and light colorsets.

 Set usage of the default color set.

 Check if the viewer is using the default color set.
 If yes, some operations might be disabled.

 Set background method.
 Deprecated method - set background color in the color-set.

 Returns reference to the default color-set.
 Static function.

 Sets static flag that determines if new viewers should use the
 default color-set.
 This is false at startup.

 Returns the value of the static flag that determines if new
 viewers should use the default color-set.
 This is false at startup.

 Returns true if curremt color set is dark.

 Set viewer viewport (window area) with bottom/left at (x,y), with
 dimensions 'width'/'height'

 Set viewr viewport from TGLRect.

 Return camera reference by type.

 Set current active camera - 'cameraType' one of:
   kCameraPerspX,    kCameraPerspY,    kCameraPerspZ,
   kCameraOrthoXOY,  kCameraOrthoXOZ,  kCameraOrthoZOY,
   kCameraOrthoXnOY, kCameraOrthoXnOZ, kCameraOrthoZnOY

 Set an orthographic camera to supplied configuration - note this
 does not need to be the current camera - though you will not see
 the effect if it is not.

 'camera' defines the ortho camera - one of kCameraOrthoXOY / XOZ / ZOY
 'left' / 'right' / 'top' / 'bottom' define the WORLD coordinates which
 corresepond with the left/right/top/bottom positions on the GL viewer viewport
 E.g. for kCameraOrthoXOY camera left/right are X world coords,
 top/bottom are Y world coords
 As this is an orthographic camera the other axis (in eye direction) is
 no relevant. The near/far clip planes are set automatically based in scene
 contents

 Set a perspective camera to supplied configuration - note this
 does not need to be the current camera - though you will not see
 the effect if it is not.

 'camera' defines the persp camera - one of kCameraPerspXOZ, kCameraPerspYOZ, kCameraPerspXOY
 'fov' - field of view (lens angle) in degrees (clamped to 0.1 - 170.0)
 'dolly' - distance from 'center'
 'center' - world position from which dolly/hRotate/vRotate are measured
             camera rotates round this, always facing in (in center of viewport)
 'hRotate' - horizontal rotation from initial configuration in degrees
 'hRotate' - vertical rotation from initial configuration in degrees

 Change base-vectors defining the camera-base transformation of current
 camera. hAxis and vAxis are the default directions for forward
 (inverted) and upwards.

 Get the auto-rotator for this viewer.

 Set the auto-rotator for this viewer. The old rotator is deleted.

 Enable stereo rendering.
 If quad_buf is true rendering is done into separate left and right GL
 buffers. This requires hardware support. Otherwise left and right images
 get rendered into left and right half of the window.
 Note that mouse highlighting and selection will not work exactly right
 as image for each eye gets slightly shifted and there are two different
 directions through the mouse pointer, one for each eye.

 Fetch the state of guides (axes & reference markers) into arguments

 Set the state of guides (axes & reference markers) from arguments.

 Draw camera look at and rotation point.

 Return selected physical shape.

 Emit MouseOver signal.

 Emit MouseOver signal.

 Emit MouseOver signal.

 Emit MouseOver signal.

 Emit UnMouseOver signal.

 Emit Clicked signal.

 Emit Clicked signal with button id and modifier state.

 Emit ReClicked signal with button id and modifier state.

 Emit UnClicked signal with button id and modifier state.

 Emit MouseIdle signal.

 Calcaulate and return pixel distance to nearest viewer object from
 window location px, py
 This is provided for use when embedding GL viewer into pad

 Process event of type 'event' - one of EEventType types,
 occuring at window location px, py
 This is provided for use when embedding GL viewer into pad

 Pass viewer for print capture by TGLOutput.

 Update GUI components for embedded viewer selection change.

 An overlay operation can result in change to an object.
 Refresh geditor.

 Update GED editor if it is set.

 Set the event-handler. The event-handler is owned by the viewer.
 If GLWidget is set, the handler is propagated to it.

 If called with handler=0, the current handler will be deleted
 (also from TGLWidget).

 Remove overlay element.

 Reset current overlay-element to zero, eventually notifying the
 old one that the mouse has left.
 Usually called when mouse leaves the window.

 TVirtualViewer3D interface ... mostly a facade
 Forward to TGLScenePad

 Only implemented because they're abstract ... should throw an
 exception or assert they are not called.

 External GUI component interface

 Stereo

 Push / drag action

 Draw and selection
 Scene rendering timeouts