[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: (FC-Devel) Have You Looked at The Dia Project?

> I will jump in here and make myself flame-bait.

I'll byte ;). Although I can only talk about my experience with DIA...

Q: What does KUml do? is it the same kind of diagramming like DIA does or does it take care of meta-information?

> First, I guess I don't see what so wrong with Argo or Dia. 
> Unless, I really
> wrong there is no executable of any type called FreeCase.  
> I'm all for a
> native Case tool, but I just think that the design is way out 
> of portion. I
> would much rather us work with Argo and Dia and create a tool 
> that works on
> Windows and Linux than worry about CORBA right now. I just 
> seems to me that
> we are building something far too complex.

The problem is not the tools in itself, they work excelent (as far as DIA is conserned at least). The problem is the underlaying architecture. DIA doesn't have any features for updating diagrams by another source (like a repository) and making it would basically mean a redesign of DIA (and that's what Alex doesn't want!).

> I can understand if we want to design by Interface as 
> suggested in Large
> Scale C++ Design by Lakos. If we were using Argo as the 
> prototype, than the
> CORBA could make sense. At the moment, we are not even a good 
> Diagramming
> tool. I can't make just a simple UML diagram in Freecase, 
> which I can in
> Argo and Dia.

If we are able to create some sort of interface that would make it possible to adjust objects on the client machine, there is a possibility that (and I'm sure they want that, hearing the reacions on the DIA list) they'll adopt it and make DIA interact with the repository. 

> So, I think we are way ahead of ourselves here. I do have 
> experince with
> this type of issue on an OpenSource project. I was key player 
> in the Free
> Builder Java IDE last year. I can tell you, I did the same 
> thing. Hell, I
> didn't know what some of the stuff really was when I talked 
> about it. It
> sounded cool. It just wasn't pratical. We made the project 
> too difficult for
> new people to join. Each module was so difficult there was 
> very little you
> do in the few minutes here and there that most of us have to 
> devote to a
> project. This is exactly what I see here.
> My suggestion is that we work with Dia or Argo. I would suggest Argo,
> because it closer to what we want to do. Use that project as 
> our prototype.
> Drop the whole CORBA thing, until we have 1.0 released. I'd 
> like see you
> consider using a toolkit like wxWindows, so that it worked on Windows,
> Gnome, Mac and BeOS... They are even forming a development 
> IDE group that
> will be create an application framework that could be used.

I can understand your worries. I know it's an amazing bunch of work. The problem is that you have to build a solid foundation first (=repository) before you can make any clients work (they can be developed in parallel for a graet deal, but that's another story). Creating one big app is probably not the solution in this case ('cause that's what's happening when you drop CORBA until 1.0). The advantage of using CORBA in this case is that we already have a strict interface to talk to.

> I wish I could help, but right now I feel this project is way 
> over my head.
> I barely know what UML standards for let alone how to create 
> a CASE tool.
> Plus, I'm off to work on my Visual Python stuff.

Damn. I wish I had my own computer here... This sounds like a challange to me ;-).

> Regards,
> Jeff