hi all,
just polling for reactions.
i noticed 2 things: * pd and forth mix very well * forth and microcontrollers mix very well
i had this idea of writing a forth for a picmicro, which consists of 2 linked forths, one running on the host, one on the target. which effectively enables you to have a 'shell' on a microcontroller over a serial line, or midi, or another bus.
it's not working yet, mostly because i just was too scared to apply power to that picmicro circuit, but the start of the idea is on my zwizwa site. the design is mostly ready, the implementation sucks and needs to be redone.
this forth (badnop) is actually the one that made me write mole, and mole, which is no more than a geek toy atm, eventually lead to a better pf. it took me a while to get right. once you start writing forths..
now, i was thinking. why not just make mole to be the host forth, choose an architecture, and start develloping some standard microcontroller protocol for pd, so you could eventually just abstract (a program running on) a microcontroller as a pd object.
i'm currently using the picmicro 18f452 as a target chip, which is not extremely fast and cheap and all, but has an instruction set that maps very well to forth, so i can compile directly to binary code. no need for closed source tools etc.. (generated code is peephole-optimized and actually works in the simulator :) i'd like to eventually design it that the target does not matter, but i'd like to get the rest working first.
now, i know this has a lot of 'forth' in it, but would anyone be remotely interested in helping to design/implement it further? it's not too hard to actually do, most of it already works except the hardware part, and might be pretty useful.
the thing for me to find out is how 'standard' i can make the way to plug this into pd. i can go on a solo tour and say 'just use mole', but i'd like to somehow take this to a higher level. maybe writing some basic library to run on a micro.
any idea would be welcome.
cheers tom