i actually did some algorythm stuff a LONG time ago on tool pathing, the logic behind it for a reasonable level of detail is very simple if you don't optimize it. in a commercial setting though you have to be very concerned about cut duration, if you can mill it in 3 hours it doesn't make sense to have the machine take 6 because it was laid out as a human machinist would.

i don't have any functional programming skills that would approach it, but could probably explain the basic concept to someone.

(side note, old pen plotters have had their innards hacked to do exactly what we're talking about. you do each z axix step as a new plot, and pull out the steppers and replace them with the x and y axis drives)

For my money, I'm wanting to build parts to hold the off-the-shelf stuff in the orientation I need for a given project.

Example: my laser-aimer. I milled up the holder for the potentiometer out of aluminum and the sliding mount for it from polycarb. I probably could have sketched it up in a cad program and had a RepRap turn it out much easier- especially given how loose my tolerances were.

I guess for the projects I'm imagining doing, I'd rather have a manual mill than a CNC of the same price, as I doubt we could get the quality from the CNC that we'd get from the manual. A quick glance at Craigslist recently showed Bridgeport-type vertical milling machines at between $2k and $4k. I think it would be hard to get a CNC of that level for that price.

Don Lancaster has some interesting articles on building CNC machines.
He refers to them as "Flutterwumpers", any machine that can either spit, or chomp.
He talks about using a postscript engine (ghostscript, adobe, etc...) on a pc to generate
movement codes that can then be sent out a serial port to a microcontroller driving
the actual hardware.

Google for it

ok so i'm pretty speedy with inventor, and what you described is pretty simple, but i'm not seeing that being faster than hitting a drill press for a few holes.

aside from fun factor, the point of cnc is (for a fab shop, not for production runs) to be able to do complex machining without the need for constant supervision and to acchieve accuracy that hand machining would require repeated setup and indexing at each step.

for many things the wrong tool for the job is *fun* but not the fastest or most appropriate.