The original idea was to allow eUML to have access to the ArgoUML command stack so that it could add its own commands to that stack.
However, as I think we're all agreed now that RCP is the correct direction for ArgoUML, I think we should make more direct use of RCP for this. That is the first step of having ArgoUML aware of RCP classes. There's a lot of work I've done in the past to try and build some undo framework which will support standalone ArgoUML and argoeclipse and in MDR and eUML. But if standalone ArgoUML is to be RCP and we wish to move away from MDR (which does not support undo itself) then I see no reason for that extra architecture. Bob. 2008/5/8 Tom Morris <[EMAIL PROTECTED]>: > I presume that Model.getCommandStack is another thing that we want to > go away before the release. What isn't clear to me is all the work > that needs to be done. The interface ModelImplementation doesn't > appear to have any other method that requires Model subsystem > implementations to implement an undo mechanism. Is it intended that > this be optional? How do they pass momentos back to get pushed on the > stack? > > Bob - can you clarify API issues? > Bogdan (Pistol) - can you comment on how the proposed API would work > (or not) with the eUML implementation? > > BTW - I'm doing a review of everything that's marked as deprecated, > starting with the Model API and Model implementations. > > Tom > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
