[vtk-developers] Widgets etc

[John Biddiscombe]

Happy new year everyone...

Dean wrote
> I agree that whatever the base class for widgets is,
> regardless of handle type, there should be functionality to 
> position the widget. This does not apply to widgets like 
> vtkScalarBarWidget though ... hmm, so I see that vtk2DWidget 
> as a potential base class.  What about John's ideas along the 
> lines of Geometric Widgets, ControllerWidgets so something 
> like a transfer function editor widget might be a 
> ControllerWidget and bypass the notion of 3D and 2D at this level?

Yes. This was my thinking. Geometric widgets line/plane etc can exist
independent of any data, but placewidget will usually be used to
initialize them inside the world bounds of some data/object which you
wish to probe/clip/slice etc. Controller widgets would not normally
exist without a related object. For example a ScalarBarWidget wouldn't
exist without a scalarbar. A TransferfunctionWidget would be a complete
"environment" in the sense that it would consist of several embedded
objects and would require inputs too (an imagedata to use as a reference
for histograms etc). 
Another category might be ImagePlaneWidget, because it serves as a
controller and also has some geometry. There seem to be endless
combinations of these types of widgets etc.

Perhaps the existing framework is ok, but we simply need to consolidate
as much as possible of the duplicated functionality into more usable
classes like polylinewidgets shared between
tracers/tr_fn_editors/splines etc.

On the subject of handles...
HandleWidget seems like an obvious one, but box widget for example allow
picking of a face, others allow lines.. Do we also have a facewidget and
an edgewidget to complete the set? (any more we need? Mesh widget for
distorting a whole mesh?)

I guess what I'm aiming at is a question...How far do we go into
partitioning widgets into classes that encapsulate small features, which
can then be amalgamated into the more complex widgets. Can we let it
evolve naturally or must we do some additional serious planning now...

I've got Both ArgoUml and the other one Will posted about up and running
so shall we get some diagrams flowing?

JB