[news.eclipse.modeling.gmf] Re: A clear definition of "children feature"

[zippo29@xxxxxxxxxx (Mous)]

Hi Alex!

I had to search on google to understand what "AFAIK" means! My english is 
not that good :p!

AW(AnyWay ;)), I tried to create an editor for this case :

EClass D <>------>EClass E<>------>EClass B <>------> EClass C ------> 
EClass A
EClass D <>------>EClass A 

I created a main editor (Root D) Es contain Bs => no problem. 
I created a sub-editor for the Bs: in the gmfmap B is the root element. I 
created a top node for C and a top node for A with (containment 
feature)and (children feature) left blank. So it's a phantom node. I 
created a linkMapping representing the reference between C and A.

I generated the gmfgen ignoring the validating error (phatom nodes exist 
not targeted ...) of the gmfmap. I generated the code(No complain from the 
gmfgen).

My editor works fine. A nodes appears at the resource level as expected 
from the phantom nodes.

> From now, I have 3 questions :
- Is this a bug in gmf : gmfmap not valid and gmfgen valid and no problem 
in generated code ?
- I want the As to be created in the D level instead of the resource 
level. Is this possible ? If yes, how to do it ? EMF custom code or GMF 
custom code ?
- I noticed that the As are shared between all the instances of B 
diagrams i-e if I created a new instance of B all the As that ewxist in 
the resource are loaded in my new diagram. I managed to handle this by 
changing the XDiagramUpdater.getB_VisualIDSemanticChildren() method. Is 
there a another/better way to do this ?


Thanks for your patience until the end!
Best regards