[stack] first order concatenative -> push-based dataflow

I hand-translated some very simple first order concatenative programs  
to a dataflow representation. I implemented the dataflow versions in  
Pure Data (http://puredata.info/). Pd is a complete piece of garbage,  
but it did work for this purpose at least.

Anyway, here it is:
http://johnnowak.com/heap/pd-concat.png

Note that 'swap', 'dup', and 'drop' are "primitives" here. The rest  
follow the normal rules for a concatenative to push-based dataflow  
translation: lines do not cross, there are no loops, and each port has  
one (and only one) connection. The 'dip' combinator is the equivalent  
of sliding a function one "column" to the left.

I'm interested in auto-generating visual representations of 5th code  
along these lines. 5th has a "flow" system (in addition to the type  
system and effect system) that keeps track of objects as they move  
around on the stack. This will allow me to color-code lines to make  
swapping, dup-ing, and so on, much more apparent. It may be worth  
something... at least for educational purposes.

If anyone wants the source code, let me know. Making sense of Pd  
without prior experience isn't much fun though. It's not fun even with  
prior experience.

- John