Ok, so I'm making some progress with the dragging and dropping of classes onto sequence diagrams, but I have a couple of problems that maybe someone can help me with:
1) I seem to get 2 events raised, and thus it adds 2 classifier roles to the diagram. I'm not sure why the event isn't being consumed... 2) I can't see how to associate the Class with the classifier role. There looks to be an underlying model somewhere that needs to be updated. I have attached a patch containing the changes thus far. If someone who knows a bit more than I can take a look, that would be great. lee On 1/23/07, Bob Tarling <[EMAIL PROTECTED]> wrote:
> 1. No ability to drag a class onto the diagram. This would > automatically add a new classifier role, and associate it with the > dropped on class. Could also consider dropping a class onto a > classifier role to associate it. You need to look at DnDJGraph to see the starting point of how the existing drag from explorer to diagrams work. Currently there is no example where drag from explorer creates a new model element or modifies the model element of the fig it is dropped on. I'd rather sequence diagram specific code doesn't go into DnDJGraph so I need to think about how to best delegate any specialist work from there to either the diagram or the graph model. > 2. No ability to drag an actor onto the diagram. This would > automatically add a new classifier role, and associate it with the > dropped on actor. Should this have a different icon? Should only be > allowed as the initiating role in the diagram. From what I recall of the code you may well find changing the box to a stick man not so easy. At least for the first round attempt you could add an <<actor>> keyword to the existing stereotype compartment of FigClassifierRole if its related classifier is an actor. > There may be other issues, but these are the ones that strike me right > now. Should these be raised in the issue tracker? Anything should be raised in IZ that is not already there. > For now, I will try and just 'fix' the existing code rather than > starting afresh. Find or create the issues and attach any patches to them. I'm living the hotel lifestyle at the moment travelling with work. I'll look at them when I can if nobody else gets around to them. If changes start looking really complicated then I'd feel better delaying them to the reimplementation. My experience with the amount of time spent with the existing implementation to make small changes and how easy it is to break something else in the process leaves me feeling rather negative about putting a great deal of time into this. However if improvements can be made to what we have while a cleaner implementation is worked on then great. Anyone learning GEF would do well to look at some of the simpler example apps and at simpler diagrams in ArgoUML such as the class diagrams. All other diagram in ArgoUML seem to be getting simpler over time. Sequence diagrams just get more complicated. Sequence diagrams are the most complicated I've seen and IMO not implemented in the best way. I think a series of developers have tried and stopped and then the next continued from where someone else gave up. Not a good sign that the right direction was taken in the first place and work continues with limited understanding of what the previous persons intentions were. This class diagram will also be a good reference to the GEF base classes that are extended in ArgoUML - http://gef.tigris.org/files/documents/9/36076/file_36076.dat?filename=Tigris%20GEF%20Diagrams.pdf Regards Bob. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
patch.txt.tgz
Description: GNU Zip compressed data
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
