[News.eclipse.technology.gmf] Re: Semantical container != Graphical cont

[Thomas Weiler <jw377@xxxxxxxxxxxxx>]

Hello!

Dmitry Stadnik schrieb:
> Hello!
>
>     Note is a pure design element and it's quite different. Basically 
> you have to patch two things:
>
> 1. Container node canonical edit policy: modify 
> getSemanticChildrenList() to use another container (it's parent I guess; 
> that corresponds to diagram element; I assume that child node is stored 
> in diagram element).
>
> 2. Container node item semantic edit policy: in generated 
> CreateElementCommand modify getEClassToEdit() and getElementToEdit() to 
> provide another container info (diagram element instead of container 
> element).
>
> In gmfmap model you should define structure that follows graphical 
> definition; in container node mapping there should be a ChildReference 
> to the child node and child node mapping within it. ContainmentFeature 
> of ChildReference could be written directly in text editor (ui filters 
> it out since it's not available amongst container node features). 
> Validation also will complain so you have to ignore it (((

I have tried your approach and now I am able to add an element to 
another diagram element than the diagram element corresponding to it's 
semantical container. But now I have a different problem:

On top level I have the diagram canvas (corresponding to semantical 
element A). A references Elements of type B and C. Furthermore I have a 
graphical node X (for the semantical element B) and a graphical node Y 
(for the semantical element C).

By following your advice I am able to create Y-nodes within X-nodes. But 
when I create this node within the X-node, another Y-node is also 
created in the canvas. So I have two Y-nodes for one C element (one in 
the X-node and one in the canvas).

Any idea how to prevent the second Y-node to appear in the canvas?

(I hope the A,B,C's were not too confusing ;-))

Thomas

F'up to eclipse.modeling.gmf