Re: [dali-dev] recent EMFBinding/Selection modifications

[Markus Kuppe <mkuppe@xxxxxxxxxxx>]

Karen Moore wrote:
> Markus,
>
> I meant to send an email out yesterday afternoon, so here's my 
> explanation now.  The bug I was trying to fix was that changing the 
> PersistentAttribute attributeMapping would cause problems.  I tried to 
> solve this problem as you suggested a few weeks ago by having the 
> Persistence Properties 'Map As' combo viewer be a SelectionTracker, but 
> was unable to do this.  My solution here was to only store the selected 
> object EClass and any possible parent EClasses because this simplified 
> things for the Dali case.  I don't know that we need to try and have a 
> selection and binding framework that will work for other clients yet.  
> It is nearly impossible to do that in a first pass, so I felt like we 
> should at least make things work for Dali, keep it simple,  and then we 
> can expand on it later if we feel that this is useful code for other 
> clients to use.
>
> I have attached the original email that explained the bug I was seeing 
> and your solution for how to fix it.  When I tried to make the 'Map As' 
> combo viewer an ISelectionTracker I hit issues because the combo viewer 
> does not actually hold emf objects, the emf object is created when the 
> model is told the the combo selection occurred.  So, at the time of 
> selection, depending on who is informed first, the ISelectionTracker or 
> the emf model, the new AttributeMapping will not be in place in the 
> model yet.  I can enter a bug in bugzilla if you would like to try and 
> fix this using the selection framework as it was before.  It just seems 
> much simipler to me to only store the EClass of the object that is 
> actually selected, in the Dali project this will only ever be a 
> PersistenceFile, PersistentType, PersistentAttribute(possibly a few 
> others as our outline tree expands).  It seemed to make more sense to 
> only store that one EClass and then you wont have as many problems with 
> child objects being replaced.
>
> I agree with number 3 and am not sure why I didn't take that approach.
>
> Karen

Hi,

i will look into this and try to solve it with the old selection version. :)
Can you please open a bug in Bugzilla for this issue?

--
Regards,
Markus Kuppe

Versant GmbH, European Headquarters
Wiesenkamp 22b, 22359 Hamburg, Germany
+49(40)60990-215, http://www.versant.com

(CEST, UTC+2)