Thats an interesting way to put programs together but I'm not sure it quite fits the direction that processing is heading in. It almost makes things too complicated and (as a programmer) I don't see the advantages.

Thanks for sharing your point of view about DFP. It helps because I’ve never used DataFlow programming intensively but I got this comment few times.

As I gain experience with JAVA, I don't feel that I'm missing something with Data-Flow models except maybe for one thing, when I have a set of well identified methods that I'm using in different ways. I could imagine that it could be helpful. At a certain stage of a design process, there’s a point where the exploration of variations on some parameters relationships on the general aspect could be helpful. It seems to be easier to visualize the structure and control changes with DataFlow. There's a kind of proximity with the result. Is it a compromise between coding and interface usability? But, as Markavian mentioned, for an experienced programmer, I could imagine that it is easier to read|use the code.

If my understanding is right, in the early CAD systems, there was a lack of relationship description between the forms and assembly; it was a low-level description (point-line-faces). This led to the development of CAD software as ProEngineer and CATIA based on Parametric Modeling (or Features Based Modeling) where the description of the relationship gives a sort of smartness to the geometry. There's a set of well identified topological grammar's rules and primitive that you combine to form complex objects. The usability the parametric relationship in "Moment" is really nice. I like software about geometry as Cabri or Cinderella, but I feel that the DataFlow is adding something nice in “Moment”.

One thing convincing about Simon's Applet is the possibility to write new methods within the Data Flow model. It looks more flexible in this way.

I must admit that I'm a Simon's fan.